Luminous Landscape Forum

Equipment & Techniques => Medium Format / Film / Digital Backs – and Large Sensor Photography => Topic started by: photolinia on April 25, 2009, 02:13:55 am

Title: DXOmark ranks DB image quality well below DSLR!
Post by: photolinia on April 25, 2009, 02:13:55 am
Reading through a photo magazine I saw an ad for DXO mark - http://www.dxomark.com/ (http://www.dxomark.com/) -
a company that claims to test raw image quality of most cameras and backs based on
color deapth, dynamic range, and low light ISO.

Looking through their database, they rate D3X the highest, and it was quite shocking to see H3DII-39 and P45+ come in well behind
many prosumer grade DSLR's...  
I did not expect medium format to do well in low light ISO, but they rank behind many DSLR's even in color deapth and dynamic range...
Is there any sense in what these guys are doing?  I'm curious if anyone is familiar with their test methods.

thanks!
-ilya
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Nick_T on April 25, 2009, 02:39:19 am
Quote from: photolinia
Reading through a photo magazine I saw an ad for DXO mark - http://www.dxomark.com/ (http://www.dxomark.com/) -
a company that claims to test raw image quality of most cameras and backs based on
color deapth, dynamic range, and low light ISO.

Looking through their database, they rate D3X the highest, and it was quite shocking to see H3DII-39 and P45+ come in well behind
many prosumer grade DSLR's...  
I did not expect medium format to do well in low light ISO, but they rank behind many DSLR's even in color deapth and dynamic range...
Is there any sense in what these guys are doing?  I'm curious if anyone is familiar with their test methods.

thanks!
-ilya
This was discussed maybe a month ago.. I think DXO have their own agenda, I have (and love) a D3 as well as an H3D31 (love it too), each have their place just as the equivalents did in the film days. I'm no scientist and frankly cannot be bothered to examine the science behind DxO's claims.
 I do know for a fact that certain gear works better than other gear for certain jobs..
Nick-T
Title: DXOmark ranks DB image quality well below DSLR!
Post by: photolinia on April 25, 2009, 02:44:47 am
Quote from: Nick_T
This was discussed maybe a month ago.. I think DXO have their own agenda, I have and love a D3 as well as an H3D31 each have their place just as the equivalents did in the film days. I'm no scientist and frankly cannot be bothered to examine the science behind DxO's claims. I do know for a fact that certain gear works better than other gear for certain jobs..
Nick-T


I agree with you, and I missed the original discussion...  from personal experience, I know that D3 is a LOT better at low light and fast action photography than an H3D-39, but when it comes to image quality, D3 can not really compare.  That's why it seems so shocking that DXO found D3 to produce better
image quality...

oh well - I was only curious.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: elf on April 25, 2009, 04:08:22 am
Quote from: photolinia
I agree with you, and I missed the original discussion...  from personal experience, I know that D3 is a LOT better at low light and fast action photography than an H3D-39, but when it comes to image quality, D3 can not really compare.  That's why it seems so shocking that DXO found D3 to produce better
image quality...

oh well - I was only curious.

If I were buying a camera today, I certainly wouldn't base my decision on Dxo (or even consider using Dxo in the decision process).
Title: DXOmark ranks DB image quality well below DSLR!
Post by: rethmeier on April 25, 2009, 04:15:31 am
Well,
I'm getting better results in the shadows(less noise) with my D3x,compared with my Hy6/75LV.
I certainly don't believe that a MFDB has more DR than my D3x.
Also,I sold my MFDB kit,so I don't really care anyway.
I don't make big prints and I'm not a landscape shooter.
All I want is an easy camera to work with.
Best,
Willem.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: ErikKaffehr on April 25, 2009, 11:03:19 am
Hi,

DxO essentially looks at noise related parameters and also low light ISO. They put no weight on resolution just take it into account regarding noise measurements. The measurements are probably correct. Their only agenda is probably to sell their DxO product, which is probably quite good as Alan Briot obviously likes it.

One area that may be of significance here is that MFDB use sensors from Kodak and Dalsa which need more calibration data than the Canon and Sony sensors employed by most of the DSLRs. According to PhaseOne each "raw" image from a Phase One DB contains something like one Megabyte of proprietary info about the individual sensor chip. This info can be utilized by the PhaseOne software but probably not by DxO.

The fact that MFDBs do not perform well at high ISO indicates that they have some form of weakness regarding noise. I never did fully understand this, if DR is high it should be possible to utilize it as a high ISO.

It's quite obvious that people who own but MFDBs and DSLRs seem to agree that the MFDBs deliver better image quality, but it's probably not in the area that DxO measures. For instance there is some discussion of "microcontrast" a term that is not very well defined but seems to be related to MTF for fine details and this may also relate to quaility of optics and presence/absence of optical low pass filter (AA-filter). Without the AA-filter thee image will have higher MTF but also contain false detail which may improve perception of sharpness but still is an artifact.

Best regards
Erik

Quote from: photolinia
Reading through a photo magazine I saw an ad for DXO mark - http://www.dxomark.com/ (http://www.dxomark.com/) -
a company that claims to test raw image quality of most cameras and backs based on
color deapth, dynamic range, and low light ISO.

Looking through their database, they rate D3X the highest, and it was quite shocking to see H3DII-39 and P45+ come in well behind
many prosumer grade DSLR's...  
I did not expect medium format to do well in low light ISO, but they rank behind many DSLR's even in color deapth and dynamic range...
Is there any sense in what these guys are doing?  I'm curious if anyone is familiar with their test methods.

thanks!
-ilya
Title: DXOmark ranks DB image quality well below DSLR!
Post by: jimgolden on April 25, 2009, 01:46:20 pm
we've been down this road...keep dreaming...
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Graham Mitchell on April 25, 2009, 02:08:48 pm
Quote from: ErikKaffehr
They put no weight on resolution just take it into account regarding noise measurements.

It seems to be a very common mistake to disregard resolution when comparing noise samples. People tend to view both images at 100% to look at the noise, which is understandable, but they forget that each pixel of noise on a 39MP camera is only going to be one quarter the area on the final print compared to a pixel of noise from a 10 MP camera. Big difference, and this means that on a per-pixel basis, the 39MP camera can afford to be a lot more noisy. It also means that noise reduction can be applied more aggressively and you will still be left with more detail.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: photolinia on April 25, 2009, 02:30:27 pm
Quote from: foto-z
It seems to be a very common mistake to disregard resolution when comparing noise samples. People tend to view both images at 100% to look at the noise, which is understandable, but they forget that each pixel of noise on a 39MP camera is only going to be one quarter the area on the final print compared to a pixel of noise from a 10 MP camera. Big difference, and this means that on a per-pixel basis, the 39MP camera can afford to be a lot more noisy. It also means that noise reduction can be applied more aggressively and you will still be left with more detail.
I thought that too, but then d3x came in ahead of d3. How can that be?
Title: DXOmark ranks DB image quality well below DSLR!
Post by: yaya on April 25, 2009, 02:56:35 pm
Quote from: photolinia
I thought that too, but then d3x came in ahead of d3. How can that be?

Sony sensor and a newer processor
Title: DXOmark ranks DB image quality well below DSLR!
Post by: ErikKaffehr on April 25, 2009, 03:11:21 pm
Hi,

They actually take resolution into account by normalizing to 8 MPixel size. Problem is that most of us regard a high resolution picture to be better than a low resolution one if the noise level is the same. I guess that very few would spend more than 2000 USD on any camera (except a golden Leica) if they would not print larger than 8x10". Cooking everything into a single figure of merit is more that dubious IMHO.

The DxO-site has good information on this: http://www.dxomark.com/index.php/eng/Insig...s-offsets-noise (http://www.dxomark.com/index.php/eng/Insights/More-pixels-offsets-noise)!

Best regards
Erik

Quote from: foto-z
It seems to be a very common mistake to disregard resolution when comparing noise samples. People tend to view both images at 100% to look at the noise, which is understandable, but they forget that each pixel of noise on a 39MP camera is only going to be one quarter the area on the final print compared to a pixel of noise from a 10 MP camera. Big difference, and this means that on a per-pixel basis, the 39MP camera can afford to be a lot more noisy. It also means that noise reduction can be applied more aggressively and you will still be left with more detail.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: ErikKaffehr on April 25, 2009, 03:20:36 pm
Hi,

There are a couple of questions about the Nikon 3Dx. It has a 14 bit capture mode which is much slower than 12-bit mode. It is not very clear how this is achieved, as it seems that it uses on chip converters whether 12 bits or 14 bits are used as ADC components seem to absent from the circuit boards in the camera.

Best regards
Erik


Quote from: yaya
Sony sensor and a newer processor
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Guillermo Luijk on April 25, 2009, 03:21:52 pm
Quote from: foto-z
It seems to be a very common mistake to disregard resolution when comparing noise samples. People tend to view both images at 100% to look at the noise, which is understandable, but they forget that each pixel of noise on a 39MP camera is only going to be one quarter the area on the final print compared to a pixel of noise from a 10 MP camera. Big difference, and this means that on a per-pixel basis, the 39MP camera can afford to be a lot more noisy. It also means that noise reduction can be applied more aggressively and you will still be left with more detail.
DxOMark put the different resolution into the equation with the 'Print' option in the DR and SNR18% tests:
- SNR18% represents well the amount of noise in correctly exposed areas of the RAW file. Both Nikon D3X and H3DII 39 perform the same here.
- DR represents well the amount of noise in the deep shadows of the RAW file. The Nikon D3X is a clear winner over H3DII 39 here.


I was a bit surprised when I saw the ISO plot for the back:

(http://img18.imageshack.us/img18/5356/isod.gif)

My interpretation is that the H3DII 39 has just one real ISO: ISO50. All over values are fake ISOs obtained from ISO50.

BR
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Graham Mitchell on April 25, 2009, 03:25:32 pm
Quote from: GLuijk
My interpretation is that the H3DII 39 has just one real ISO: ISO50. All over values are fake ISOs obtained from ISO50.

I think that's true for all MFDBs
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Guillermo Luijk on April 25, 2009, 04:16:14 pm
Quote from: foto-z
I think that's true for all MFDBs
It seems the Phase One has ISO50 and 100, and the Leaf ISO50, 100, 200, 400 and 800. Only the Hasselblad has a unique real ISO, of 50.
Probably this does not mean anything, just that users should avoid fake ISOs if there is a risk of losing highlights information.

(http://img26.imageshack.us/img26/5977/isos.gif)
Title: DXOmark ranks DB image quality well below DSLR!
Post by: BernardLanguillier on April 25, 2009, 04:45:05 pm
Quote from: foto-z
It seems to be a very common mistake to disregard resolution when comparing noise samples. People tend to view both images at 100% to look at the noise, which is understandable, but they forget that each pixel of noise on a 39MP camera is only going to be one quarter the area on the final print compared to a pixel of noise from a 10 MP camera. Big difference, and this means that on a per-pixel basis, the 39MP camera can afford to be a lot more noisy. It also means that noise reduction can be applied more aggressively and you will still be left with more detail.

True, on the other hand if you do the same computation between a 39MP back and a D3x, you will realize that a the same DPI, the linear size of the print you can achieve is only 1.25 times larger, meaning that you can go from an A4 print (29.7 cm long) to one that is 37cm long.

It can of course be argued that the 39MP AAfilter less back has higher pixel quality, which is true to some extend.

Now, these 39MP back are old technology, and it would be interesting to see what companies like Kodak and Dalsa could do with 39MP if they tried today, but it seems that they have been stucked in the MP race like everybody else and will not develop any new sensor with that kind of resolution.

Cheers,
Bernard
Title: DXOmark ranks DB image quality well below DSLR!
Post by: photolinia on April 25, 2009, 05:38:11 pm
Quote from: foto-z
I think that's true for all MFDBs

Hmmm...  Could somebody explain this real ISO vs fake ISO issue?
I thought that all cameras had one base ISO  (50 for most MFDB, 100 for D3X, 200 for D3), but then
used amps to simulate the effect of higher ISOs - is that not the case?

what is the difference between the way D3X and H3D-39 derive higher ISO's?

-ilya
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Guillermo Luijk on April 25, 2009, 06:28:39 pm
Quote from: photolinia
what is the difference between the way D3X and H3D-39 derive higher ISO's?
Real ISOs refer to the gain of the amplifier that feeds the captured information from the sensor into the AD converter.
Fake ISOs are shots obtained at a different (usually lower) real ISO value and then overexposed by the camera software.

A camera usually has some real ISOs, and also several fake ISOs. For example on a Canon 5D, ISO3200 is fake ISO. The shot is actually taken at ISO1600 and then overexposed by 1 stop (all levels are multiplied by 2) before saving the RAW data.

A fake ISO like the ISO3200 on the Canon 5D has no noise advantage when shooting RAW over shooting at the same aperture/shutter using the lower real ISO (ISO1600 in the example), because it's actually a ISO1600 shot. But the fake ISO can make us loose up to 1 stop of highlights information because of the overexposure.
So fake ISOs should always be avoided when shooting in RAW mode or we can loose DR in the highlights when comparing to shooting at the same aperture/shutter with the highest real ISO.

In JPEG mode all ISOs are fine, as long they help us to achieve the correct exposure in the final image, since this will be hardly corrected.

Looking at the plot, it seems the Hasselblad has only one real ISO (ISO50). When setting ISO100, 200 or 400 on that camera, the shot will be internally done at ISO50 and then respectively overexposed by 1, 2 or 3 stops in software. So shotting at any ISO over 50 in the Hassy is useless and can make the user loose highlights information with respect to using the same aperture/shutter and ISO50.

BR
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on April 25, 2009, 09:00:56 pm
Quote from: foto-z
I think that's true for all MFDBs
This is not so straightforward. I can make definitive statements only relating to a few models and a few ISO steps:

Phase One P25 Plus: no difference from 100 to 800, i.e. fake latest from ISO 200

Phase One P30 Plus: real up to 200; 400, 800 and 1600 are fake

Phase One P45 Plus: real at least up to 800

Sinar e54: already ISO 100 is fake

Hasselblade announced a short while ago ISO 1600 for whatever back. They (the users on this forum at least) have not noticed, that this happens without any hardware upgrading. I saw only a small side-note stating, that the ISO increase is due to the better noise reduction of the raw processor. I would have loved to have such  digitally challenged customers as Hasselblade's are.

Title: DXOmark ranks DB image quality well below DSLR!
Post by: photolinia on April 25, 2009, 09:09:16 pm
Quote from: GLuijk
Real ISOs refer to the gain of the amplifier that feeds the captured information from the sensor into the AD converter.
Fake ISOs are shots obtained at a different (usually lower) real ISO value and then overexposed by the camera software.

A camera usually has some real ISOs, and also several fake ISOs. For example on a Canon 5D, ISO3200 is fake ISO. The shot is actually taken at ISO1600 and then overexposed by 1 stop (all levels are multiplied by 2) before saving the RAW data.

A fake ISO like the ISO3200 on the Canon 5D has no noise advantage when shooting RAW over shooting at the same aperture/shutter using the lower real ISO (ISO1600 in the example), because it's actually a ISO1600 shot. But the fake ISO can make us loose up to 1 stop of highlights information because of the overexposure.
So fake ISOs should always be avoided when shooting in RAW mode or we can loose DR in the highlights when comparing to shooting at the same aperture/shutter with the highest real ISO.

In JPEG mode all ISOs are fine, as long they help us to achieve the correct exposure in the final image, since this will be hardly corrected.

Looking at the plot, it seems the Hasselblad has only one real ISO (ISO50). When setting ISO100, 200 or 400 on that camera, the shot will be internally done at ISO50 and then respectively overexposed by 1, 2 or 3 stops in software. So shotting at any ISO over 50 in the Hassy is useless and can make the user loose highlights information with respect to using the same aperture/shutter and ISO50.

BR

Oh - very interesting - thanks!

So, does that mean that in Nikons all the"Hi" and "Low" settings are the fake software ISO's and the ones designated with actual numbers are real internal amp setting ISO's?

-ilya
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Ray on April 25, 2009, 09:21:44 pm
Quote from: GLuijk
DxOMark put the different resolution into the equation with the 'Print' option in the DR and SNR18% tests:

Yes it does. When I first saw this option "print", I actually thought it was asking me if I wanted to print out the results. (Duh!)

At this relatively small print size one can see a fairly dramatic shift in most of the results. The P45 actually has better SNR than the D3X, at base ISO of 50. But only slightly better   .
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Guillermo Luijk on April 25, 2009, 10:23:56 pm
Quote from: photolinia
Oh - very interesting - thanks!

So, does that mean that in Nikons all the"Hi" and "Low" settings are the fake software ISO's and the ones designated with actual numbers are real internal amp setting ISO's?

I wouldn't say that, it has to be analysed because manufacturers lie more than they speak.
I don't know about Nikon's ISOs, but I can tell you in the Canon 5D MKII, ISO6400 is fake but if you read the camera user manual ISO6400 is groupped together with 100, 200, 400, 800, 1600 letting you believe it's a real ISO. They never say it is an electronic ISO, but they let the user believe it.

Funnily yesterday I found someone in the DPreview forums (http://forums.dpreview.com/forums/read.asp?forum=1032&message=31685274) claiming that ISO6400 on the 5D2 is fantastic, noise-free and far better than ISO3200.

BR


Title: DXOmark ranks DB image quality well below DSLR!
Post by: dfarkas on April 25, 2009, 10:30:10 pm
Quote from: BernardLanguillier
Now, these 39MP back are old technology, and it would be interesting to see what companies like Kodak and Dalsa could do with 39MP if they tried today, but it seems that they have been stucked in the MP race like everybody else and will not develop any new sensor with that kind of resolution.

Cheers,
Bernard

The Leica S2 uses a newly designed Kodak 37.5 MP CCD sensor based on the latest 6um pixel architecture.

David
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on April 25, 2009, 10:46:46 pm
Quote from: photolinia
So, does that mean that in Nikons all the"Hi" and "Low" settings are the fake software ISO's and the ones designated with actual numbers are real internal amp setting ISO's?
The designation is meaningful only in one direction: the ISO steps with special designation like "Hi" or "boosted" are usually fake; however, those not designated as such may be real or fake.

Examples

Nikon D300 and D90: ISO 6400 is designated as "boosted", but already 3200 is fake

Nikon D3: this is "correct", i.e. ISO 6400 is analogue supported.

The same is happening in the Canon line.

However, this is is much less interesting than the factual usefulness. Most of the top analogue ISO steps are purely eye-wash; there is nothing to gain over 1600, no matter if Canon or Nikon. The only exception is perhaps the Canon 1DMkIII, 3200 seems to contribute a tiny bit compared to 1600.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: EricWHiss on April 26, 2009, 01:16:20 am
Besides the noise and resolution, I'll bet there are color accuracy differences between MFDB and DSLR cameras.   Does DXO measure this too?    


Quote from: Panopeeper
The designation is meaningful only in one direction: the ISO steps with special designation like "Hi" or "boosted" are usually fake; however, those not designated as such may be real or fake.

Examples

Nikon D300 and D90: ISO 6400 is designated as "boosted", but already 3200 is fake

Nikon D3: this is "correct", i.e. ISO 6400 is analogue supported.

The same is happening in the Canon line.

However, this is is much less interesting than the factual usefulness. Most of the top analogue ISO steps are purely eye-wash; there is nothing to gain over 1600, no matter if Canon or Nikon. The only exception is perhaps the Canon 1DMkIII, 3200 seems to contribute a tiny bit compared to 1600.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: lisa_r on April 26, 2009, 09:51:05 am
I think it's interesting that anywhere above 50 ISO the DSLRs beat the backs in terms of D.R. Often by one or two full stops, even though the backs have larger pixels. I am really wondering where the truth lies.

It's easy to find threads here and elsewhere where one guy claims that in terms of D.R. his back "blows away" his Canons, and then right after that someone else says that their back is no better than their Canon. Maybe it's because they are shooting at different ISOs...

Click the Dynamic Range tab here  (http://www.dxomark.com/index.php/eng/Image-Quality-Database/Compare-cameras/(appareil1)/305|0/(appareil2)/301|0/(appareil3)/304|0/(onglet)/0/(brand)/Canon/(brand2)/Hasselblad/(brand3)/Phase%20One)to see what I am talking about:

http://www.dxomark.com/index.php/eng/Image...d3)/Phase%20One (http://www.dxomark.com/index.php/eng/Image-Quality-Database/Compare-cameras/(appareil1)/305|0/(appareil2)/301|0/(appareil3)/304|0/(onglet)/0/(brand)/Canon/(brand2)/Hasselblad/(brand3)/Phase%20One)

Maybe it's the fashion guys who are often shooting at higher ISOs because they are mixing daylight with strobes and such - and are therefore not seeing the DR advantage of the backs (because there is apparently no advantage at 400ISO) and the landscape/architecture guys who are seeing the backs as having better D.R. only because they are shooting at ISO 50?

I do not currently have a back in my possession or I would attempt to confirm this on my own...
Title: DXOmark ranks DB image quality well below DSLR!
Post by: ErikKaffehr on April 26, 2009, 10:15:45 am
Hi,

I got the impression that "dynamic range" is not that easy to test.

My guess is simply that DSLRs lead regarding noise related factors, because of better sensor technology (including preamplification) while MFDBs lead with regard to resolution/MTF related factors.

Best regards
Erik

Quote from: lisa_r
I think it's interesting that anywhere above 50 ISO the DSLRs beat the backs in terms of D.R. Often by one or two full stops, even though the backs have larger pixels. I am really wondering where the truth lies.

It's easy to find threads here and elsewhere where one guy claims that in terms of D.R. his back "blows away" his Canons, and then right after that someone else says that their back is no better than their Canon. Maybe it's because they are shooting at different ISOs...

Click the Dynamic Range tab here  (http://www.dxomark.com/index.php/eng/Image-Quality-Database/Compare-cameras/(appareil1)/305|0/(appareil2)/301|0/(appareil3)/304|0/(onglet)/0/(brand)/Canon/(brand2)/Hasselblad/(brand3)/Phase%20One)to see what I am talking about:

http://www.dxomark.com/index.php/eng/Image...d3)/Phase%20One (http://www.dxomark.com/index.php/eng/Image-Quality-Database/Compare-cameras/(appareil1)/305|0/(appareil2)/301|0/(appareil3)/304|0/(onglet)/0/(brand)/Canon/(brand2)/Hasselblad/(brand3)/Phase%20One)

Maybe it's the fashion guys who are often shooting at higher ISOs because they are mixing daylight with strobes and such - and are therefore not seeing the DR advantage of the backs (because there is apparently no advantage at 400ISO) and the landscape/architecture guys who are seeing the backs as having better D.R. only because they are shooting at ISO 50?

I do not currently have a back in my possession or I would attempt to confirm this on my own...
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Guillermo Luijk on April 26, 2009, 10:49:12 am
Quote from: lisa_r
I think it's interesting that anywhere above 50 ISO the DSLRs beat the backs in terms of D.R. Often by one or two full stops, even though the backs have larger pixels. I am really wondering where the truth lies.
If you read carefully this thread you will find the answer: backs have only one or two real ISOs, the rest are fake ISOs obtained by overexposing in software a lower real ISO. With every extra fake ISO stop, you are losing 1 stop in the ability to capture DR.

Are backs wrongly designed? no way, they were designed for applications where usually the base native ISO is used everytime.

BR

Title: DXOmark ranks DB image quality well below DSLR!
Post by: photolinia on April 26, 2009, 12:13:01 pm
Quote from: GLuijk
If you read carefully this thread you will find the answer: backs have only one or two real ISOs, the rest are fake ISOs obtained by overexposing in software a lower real ISO. With every extra fake ISO stop, you are losing 1 stop in the ability to capture DR.

Are backs wrongly designed? no way, they were designed for applications where usually the base native ISO is used everytime.

BR


Regarding this REAL ISO question...  Do MFDB people admit to this?  I'm attending a Hasselblad event on Tuesday and will ask them about it.  
If the higher ISO's are fake software options, then no one should really ever use them, but I've heard from many people (including on this
forum) who use higher ISO's regularly...

-ilya
Title: DXOmark ranks DB image quality well below DSLR!
Post by: R_Medvid on April 26, 2009, 12:50:43 pm
Quote from: GLuijk
With every extra fake ISO stop, you are losing 1 stop in the ability to capture DR.

If you increase ISO by roughly shifting the pixels in the histogram one stop to the right, then yes, 1 stop of DR is lost.

But is there a way to fake ISO by smartly pushing up only shadows and slightly lighting up the midtones, without blowing out the highlights?

16-bit depth would allow this no problem, I guess.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Guillermo Luijk on April 26, 2009, 02:50:33 pm
Quote from: photolinia
Regarding this REAL ISO question...  Do MFDB people admit to this?  I'm attending a Hasselblad event on Tuesday and will ask them about it.
Sadly they will probably not understand a word about what you're talking about.

Quote from: R_Medvid
is there a way to fake ISO by smartly pushing up only shadows and slightly lighting up the midtones, without blowing out the highlights?
Sure! you are talking about... postprocessing   that's easily done in Photoshop, or even in the RAW developer with the Brightness control, which unlike the Exposure control allows to lift the shadows without blowing the highlights.

BR
Title: DXOmark ranks DB image quality well below DSLR!
Post by: douglasf13 on April 26, 2009, 03:01:11 pm
The A900 is somewhat similar in that it actually performs better by never going over ~ISO 400, and boosting in the converter....assuming you're not using ACR, but rather a converter with high quality output like RPP, Raw Therapee, etc.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: paulmoorestudio on April 26, 2009, 03:12:38 pm
[quote name='GLuijk' date='Apr 26 2009, 07:50 PM' post='279096']
Sadly they will probably not understand a word about what you're talking about.


give me a break!
when I bought my hasselblad back a few years back the back rep clearly stated to me that the base iso is what the back performs best at and that any higher than that, it is just software manipulation..
so I think they would understand..maybe they aren't as clever as you but I think they understand how their backs work.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: R_Medvid on April 26, 2009, 03:18:20 pm
Quote from: GLuijk
Sure! you are talking about... postprocessing   that's easily done in Photoshop, or even in the RAW developer with the Brightness control, which unlike the Exposure control allows to lift the shadows without blowing the highlights.

BR

But, this is understood. My question was obviously about the on-chip shadow-pushing-up before making up of the RAW file. I'm talking about a smarter ISO faking. Is it possible?
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Guillermo Luijk on April 26, 2009, 03:30:59 pm
Quote from: paulmoorestudio
so I think they would understand..maybe they aren't as clever as you but I think they understand how their backs work.
It's not a question of being clever, but just interested in those matters. I am glad to see the people you bought the back from were not only informed but also honest. I hope photolinia will tell us how was his event.


Quote from: R_Medvid
My question was obviously about the on-chip shadow-pushing-up before making up of the RAW file.
That's RAW cooking, and nobody should be interested in that. First because is useless (it can be done better in postprocessing), and second because it would ruin the virgin and linear condition of the RAW file.

BR
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Mort54 on April 26, 2009, 03:45:08 pm
OK, I'll throw some gasoline on the fire :-)

I shoot primarily landscapes and use a P45+ on a Mamiya 645 II, as well as a D3 (one of these days I'll upgrade to a D3X). The only thing the P45+ gives me is more pixels, and no AA filter getting in the way of those pixels. There, I said it out loud. Believe me, I take no pleasure in saying that, given how much I spent on the P45+ kit. But that's what I'm seeing.

More pixels is no small thing, and the lack of an AA filter further helps out,  and in fact these are the two primary reasons I got the P45+ kit. The detail available in a MFDB file is breathtaking. But as for all the other so-called advantages attributed to MFDBs, I'm not seeing them. If anything, I agree with DxO's assessment here. Maybe these other advantages were real a few years ago, but with the latest crop of DSLRs, the gap has narrowed considerably. I have no trouble pulling clean sharp detail out of the shadows with my D3. If anything, I rank the D3 slightly superior to the P45+ in this regard (based on using the D3 at ISO 200 and the P45+ at ISO 50, their respective base ISOs). And the D3 has more mid-tone to highlight DR than the P45+, again at base ISOs. The P45+ saturates at around 2 and 2/3 to 3 stops over mid-tone. The D3 highlights don't hit the limit for another two thirds to a full stop beyond that. And the D3X is reportedly even better in this regard.

On top of all this is the convenience factor in Nikon's (and Canon's) favor. I have excellent quality super wides to super telephotos, and everything in between. I have an excellent selection of tilt/shift lenses. I have blazingly fast AF. My batteries last days, not hours. And I have superb in-camera live view that allows me to achieve critical focus to a much more repeatable and precise degree than I've ever been able to achieve with my MF kit.

One of these days I'll get a D3X, and I doubt I'll pull the P45+ kit out much after that, based on what I've seen to date.

Regards,
Mort.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: ErikKaffehr on April 26, 2009, 03:48:14 pm
Hi,

Real ISO is achieved with preamplification. Fake ISO is achieved by essentially multiplying the signal after analogue to digital conversion. Preamplification may be boosted so far that signal to noise ratio is not improved, in this case the increase in ISO will be essentially be fake.

Check this article: http://www.clarkvision.com/imagedetail/dig...ary/#unity_gain (http://www.clarkvision.com/imagedetail/digital.sensor.performance.summary/#unity_gain)

Best regards
Erik




Quote from: photolinia
Hmmm...  Could somebody explain this real ISO vs fake ISO issue?
I thought that all cameras had one base ISO  (50 for most MFDB, 100 for D3X, 200 for D3), but then
used amps to simulate the effect of higher ISOs - is that not the case?

what is the difference between the way D3X and H3D-39 derive higher ISO's?

-ilya
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on April 26, 2009, 03:57:02 pm
Quote from: lisa_r
I think it's interesting that anywhere above 50 ISO the DSLRs beat the backs in terms of D.R. Often by one or two full stops, even though the backs have larger pixels. I am really wondering where the truth lies
The cases have to be separated.

For example the Phase One P45 Plus has analogue ISO steps up to 800 (I don't know about 1600; perhaps there is no such step at all). This means, that the dynamic range gets reduced by every ISO step increase; the magnitude of the loss needs to be measured. I don't have suitable raw files, so I can't make any statement, but I find it hardly believable, that the P45 Plus does not profit anything from the analogue gain, according to the DxO chart.

On the other hand, the fake ISO steps of the MFDBs do not reduce the DR by any amount, in contrast to DSLRs.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: feppe on April 26, 2009, 04:03:58 pm
Quote from: Mort54
OK, I'll throw some gasoline on the fire :-)

That you definitely did

I've been always skeptical about the hand-waving of MFDB aficionados, and the dismissive claims of "micro-tonality" (whatever that is) and "3D-look" (I don't see it). I shoot MF and 35mm film (well, shot, although I'm seriously thinking re-starting limited MF shooting), the only difference in Velvia and Provia out of those two cameras is that there's lots more detail.

Sure, if I had 50k to throw away I'd buy a 65+ or S2 in a heartbeat. But I'll be happy with 12 megapixels - I was happy with 8 previously...
Title: DXOmark ranks DB image quality well below DSLR!
Post by: ErikKaffehr on April 26, 2009, 04:18:22 pm
Hi!

So you see benefits from MFDB regarding detail but not on dynamic range? That makes sense to me...

In a sense this is academic. If the MFDB images are better than that's it. It doesn't really matter if they are better because of high MTF, better microcontrast or higher DR. On the other hand, being an engineer I prefer explanations which are consistent with theory and measurements.

Best regards
Erik

Quote from: Mort54
OK, I'll throw some gasoline on the fire :-)

I shoot primarily landscapes and use a P45+ on a Mamiya 645 II, as well as a D3 (one of these days I'll upgrade to a D3X). The only thing the P45+ gives me is more pixels, and no AA filter getting in the way of those pixels. There, I said it out loud. Believe me, I take no pleasure in saying that, given how much I spent on the P45+ kit. But that's what I'm seeing.

More pixels is no small thing, and the lack of an AA filter further helps out,  and in fact these are the two primary reasons I got the P45+ kit. The detail available in a MFDB file is breathtaking. But as for all the other so-called advantages attributed to MFDBs, I'm not seeing them. If anything, I agree with DxO's assessment here. Maybe these other advantages were real a few years ago, but with the latest crop of DSLRs, the gap has narrowed considerably. I have no trouble pulling clean sharp detail out of the shadows with my D3. If anything, I rank the D3 slightly superior to the P45+ in this regard (based on using the D3 at ISO 200 and the P45+ at ISO 50, their respective base ISOs). And the D3 has more mid-tone to highlight DR than the P45+, again at base ISOs. The P45+ saturates at around 2 and 2/3 to 3 stops over mid-tone. The D3 highlights don't hit the limit for another two thirds to a full stop beyond that. And the D3X is reportedly even better in this regard.

On top of all this is the convenience factor in Nikon's (and Canon's) favor. I have excellent quality super wides to super telephotos, and everything in between. I have an excellent selection of tilt/shift lenses. I have blazingly fast AF. My batteries last days, not hours. And I have superb in-camera live view that allows me to achieve critical focus to a much more repeatable and precise degree than I've ever been able to achieve with my MF kit.

One of these days I'll get a D3X, and I doubt I'll pull the P45+ kit out much after that, based on what I've seen to date.

Regards,
Mort.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: David Grover / Phase One on April 26, 2009, 04:41:39 pm
Quote from: GLuijk
Sadly they will probably not understand a word about what you're talking about.


Sure! you are talking about... postprocessing   that's easily done in Photoshop, or even in the RAW developer with the Brightness control, which unlike the Exposure control allows to lift the shadows without blowing the highlights.

BR

Charming.

Anyway, there are differences in the DSP functions regarding higher ISO settings on the camera, but then again, I probably don't know what I am talking about.

 

Best,


David


Title: DXOmark ranks DB image quality well below DSLR!
Post by: David Grover / Phase One on April 26, 2009, 04:43:58 pm
On a separate note..

Regardless of how A to B is reached, if the result is a useable image of expected quality, who really cares?
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Guillermo Luijk on April 26, 2009, 04:47:41 pm
Quote from: David Grover / Hasselblad
Charming.

Anyway, there are differences in the DSP functions regarding higher ISO settings on the camera, but then again, I probably don't know what I am talking about.
Amazingly valuable and detailed information about your product you are providing here. You can take a rest now.


Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on April 26, 2009, 05:10:55 pm
Quote from: David Grover / Hasselblad
Regardless of how A to B is reached, if the result is a useable image of expected quality, who really cares?
My first reaction was this is the argument of a used car salesman. However, my experience shows, that not only many but perhaps the majority of the owners don't have much idea about the characteristics of their cameras in general. The owners of MFDBs are not excepted, or if they are, then rather on the ignorant side. No problem, it's their money. Thus the position "don't care for details as long as you are happy" is justified.

I remember to have read in some car related magazine, that the Rolls Royce specifications do not mention the power output, nor the top speed of their cars; asked for that data, the answer was enough - and from the perspective of their customers, they were right.

Beside the total confusion about the role of ISO and about proper exposure, the second most common misconception is that about the raw data bit depth, particularly regarding the imaginary 16bit depth of Phase One backs.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: R_Medvid on April 26, 2009, 05:13:32 pm
Quote from: Panopeeper
.... the imaginary 16bit depth of Phase One backs....

 Could you elaborate on this please?
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Nick-T on April 26, 2009, 05:25:17 pm
Quote from: David Grover / Hasselblad
On a separate note..

Regardless of how A to B is reached, if the result is a useable image of expected quality, who really cares?


Well there's clearly a group of prolific posters here who one might call imaging scientists or something. These people are fascinated by the inner workings of digital imaging and good luck to them.
None of these people as far as I can see make a living taking photos. The people at DXo don't make a living from taking photos.

I make a living from taking photos and my digital backs out perform my DSLRs and my clients can tell the difference, end of story.

Frankly I care a great deal more about my clients than I do graphs..

NickT
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Bill Caulfeild-Browne on April 26, 2009, 05:40:13 pm
Quote from: BernardLanguillier
True, on the other hand if you do the same computation between a 39MP back and a D3x, you will realize that a the same DPI, the linear size of the print you can achieve is only 1.25 times larger, meaning that you can go from an A4 print (29.7 cm long) to one that is 37cm long.

It can of course be argued that the 39MP AAfilter less back has higher pixel quality, which is true to some extend.

Now, these 39MP back are old technology, and it would be interesting to see what companies like Kodak and Dalsa could do with 39MP if they tried today, but it seems that they have been stucked in the MP race like everybody else and will not develop any new sensor with that kind of resolution.

Cheers,
Bernard

Actually, the linear size is only 20% bigger, unless the Nikon version of the Sony sensor is different. My a900 is 6048 pixels wide; the P45+ is 7216 wide. (Height, on the other hand, is 34% greater.)

Having said that, I have on several occasions shot the same image with both sensors, adjusting zoom lenses to ensure the images are the same size linearly, and my eyes on-screen and more importantly in print, tell me the MFDB image is clearly superior in detail and tonality.

Bill
Title: DXOmark ranks DB image quality well below DSLR!
Post by: EricWHiss on April 26, 2009, 05:47:54 pm
There's like a mile of gap between my 5D and my older P20 (non plus) and that's even when I use the brilliant Leica R lenses on my canon.  There's several stops of usable DR more on my p20 than the 5D.  If the 5D 2 got 1 stop more DR than the 5D classic then it still wouldn't match up to my 5 year old P20.     I am a big doubter that the 5D2 or D3X are going to replace even 2 or 3 generation older MFDB's for studio use.    I think Michael wrote in the first DXO/MFDB thread why the DSLR numbers come out so favorable for DR - something about in camera noise reduction while MFDB's rely on on RAW processing software.   The DXO numbers for at least DR are not useful for comparing DSLR to MFDB.  

Title: DXOmark ranks DB image quality well below DSLR!
Post by: Carsten W on April 26, 2009, 06:20:01 pm
Quote from: EricWHiss
There's like a mile of gap between my 5D and my older P20 (non plus) and that's even when I use the brilliant Leica R lenses on my canon.  There's several stops of usable DR more on my p20 than the 5D.  If the 5D 2 got 1 stop more DR than the 5D classic then it still wouldn't match up to my 5 year old P20.     I am a big doubter that the 5D2 or D3X are going to replace even 2 or 3 generation older MFDB's for studio use.    I think Michael wrote in the first DXO/MFDB thread why the DSLR numbers come out so favorable for DR - something about in camera noise reduction while MFDB's rely on on RAW processing software.   The DXO numbers for at least DR are not useful for comparing DSLR to MFDB.

It is interesting to note that the people who claim that MFDBs are "better" generally talk about what they see, and what their customers see, i.e. a clear visual superiority for the MFDBs. The people who claim that the difference is very small are generally talking about statistics, plots, measured noise, and so on. There are a small number of people who go against the stream and claim that they can see no difference, or measure no difference.

This leads me to wonder about two things:

1) Could they both be true? In other words, could it be that the MFDBs give visually noticeably more appealing results, while there is little measurable difference? That would mean that what we measure doesn't translate directly to what we feel about the results. This would lead me to think that if we want to measure how good something is, we need to come up with new tests and new ways of measuring. Especially the DR results make me think this. One hears repeatedly how much cleaner the shadows are with MFDBs, but this somehow doesn't translate into better DR results, counter-intuitively, I suppose because what we like and what we can measure don't match.

2) Could it be that the 35mm-FF camera manufacturers know about the lack of significant measurable differences, and are actively pushing this angle by "cooking" the results to obtain the best possible *measured* result? They must surely know that very few people will ever see top-notch MFDB results, and that cooking the results to minimize the measurable differences helps them sell more cameras. This is somewhat analogous to graphics card manufacturers writing drivers which are specifically optimised for certain gaming performance benchmarks, while performing no better than other cards in actual play...
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Carsten W on April 26, 2009, 06:21:33 pm
Quote from: billcb
Actually, the linear size is only 20% bigger, unless the Nikon version of the Sony sensor is different. My a900 is 6048 pixels wide; the P45+ is 7216 wide. (Height, on the other hand, is 34% greater.)

Having said that, I have on several occasions shot the same image with both sensors, adjusting zoom lenses to ensure the images are the same size linearly, and my eyes on-screen and more importantly in print, tell me the MFDB image is clearly superior in detail and tonality.

I am curious about this comment (genuinely curious, not looking to poke holes in it). I presume you are comparing the P45+ and the A900, not some other cameras. Could you explain in a bit more detail what the differences are that you see, and that you feel while working? Do you use the same software for both?

I would love a P45+, but my budget today barely stretches to an A900, which is why I am interested.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Graeme Nattress on April 26, 2009, 06:34:17 pm
Talking fake and not fake ISO is pejorative. We should be talking analogue gain and digital gain.

The benefit of analogue gain is you get analogue noise instead of digital noise. For each stop you gain up, you loose a stop in the highlights though.

Digital gain can be configured as a straight gain, working just like analogue gain, but digitally. Or you can configure the gain on a curve to preserve highlights instead of loosing them.

Different sensors / post processing will behave differently. Some will look better with analogue gain and some with digital gain.

Some cameras will use one or the other, and others a combination of the two.

Graeme
Title: DXOmark ranks DB image quality well below DSLR!
Post by: photolinia on April 26, 2009, 06:35:34 pm
Mort,

Thanks for throwing in some gasoline... this is basically what I've been trying to get to for the past several weeks.

I'm looking to upgrade my 12MP Nikon system to something with better resolution and detail.  
The D3X would be a logical choice, but I am somewhat put off by the obnoxious price tag.  I've had a chance to test P45+ and the H3DII-39 and absolutely
loved the detail I got with those backs.  I liked the handling of the H3D a bit more (plus the P45+ I used had a major AF problem)
and the fact that I can find a used H3DII-39 kit with an 80mm lens for not too much more than a D3X made me look into H3D very seriously.

I shoot about 50% in a studio and 50% on location.  I think in a studio MFDB wins by a mile, but on location it can be tricky.  
Basically, I'm at a cross road between a D3X and a used MFDB, and it sounds like given the choice you would go for a D3X...

Obviously, I also have many more lens options with a Nikon.  With an H3D I would probably be stuck with one or two for a while...

Nick,
Other than file size and resolution, what other aspects of MF photos do your clients prefer?  I think the clients do not always know best
and often know very little, but of course they pay, so they get to choose...

-ilya
Title: DXOmark ranks DB image quality well below DSLR!
Post by: feppe on April 26, 2009, 06:45:22 pm
Quote from: carstenw
It is interesting to note that the people who claim that MFDBs are "better" generally talk about what they see, and what their customers see, i.e. a clear visual superiority for the MFDBs. The people who claim that the difference is very small are generally talking about statistics, plots, measured noise, and so on. There are a small number of people who go against the stream and claim that they can see no difference, or measure no difference.

I believe it has a lot to do with preconceived notions, and seeing what one wants to see. I posit that in a double-blind test the differences would not be nearly as obvious as many claim.

There was a lot of discussion quite a while on this forum that MFDB shots are obvious from DSLR ones even at screen resolution which I found preposterous. I ran a test  (http://luminous-landscape.com/forum/index.php?showtopic=19420&hl=), in which 122 of LL members participated in, and they were not able to tell the difference (not better than chance). I'd be thrilled to have someone run the same test with prints or larger files.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on April 26, 2009, 06:55:59 pm
Quote from: Graeme Nattress
Talking fake and not fake ISO is pejorative. We should be talking analogue gain and digital gain
This is BS. "Digital gain" in the current context is an oxymoron.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Nick-T on April 26, 2009, 07:08:35 pm
Quote from: photolinia
Nick,
Other than file size and resolution, what other aspects of MF photos do your clients prefer?  I think the clients do not always know best
and often know very little, but of course they pay, so they get to choose...

-ilya

You are right, some clients know very little, for them the difference might be perceptual, the look of the camera itself. Many others though, have been burned by DSLR shooters providing poor quality files. Granted some of the blame may well be laid at the foot of the un-skilled photographer but I get calls complimenting the quality of my files. I had a senior creative ring me the other day to compliment me on an image of some dog biscuits, so I don't think my artistic abilities had much to do with it  
In my experience the MFDB files look good "out of the box". By that I mean that as they pop up on screen (tethered shooting) they look how I want them to look, and I can (bar say a bit of dust busting in Photoshop) deliver files direct from the capture software. Whenever I shoot DSLR I find myself working a lot harder to get a decent result which still falls short of the quality I get from the backs.

HTH
Nick-T
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on April 26, 2009, 07:09:57 pm
Quote from: Nick-T
These people are fascinated by the inner workings of digital imaging and good luck to them. None of these people as far as I can see make a living taking photos...

I make a living from taking photos and my digital backs out perform my DSLRs and my clients can tell the story.
The greatness of the MFDBs is in my eyes, that they outperform the DSLRs even if their users don't know how to use it optimally.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: EricWHiss on April 26, 2009, 07:18:43 pm
Carsten,
A while back I measured the DR of my Leica DMR against the then 1D3 and found the DMR had 1.5 stops more "useable" DR than the Canon using a software package called imatest and a stoufer transmission step wedge.  I still have the DMR and shoot it side by side with the P20.  It's obvious from the shots that the P20 has about the same advantage over the DMR that it had over the canon.  It's also true that not all MFDB have the same DR.  Some have 1 stop more than others. Just check the chip manufacturers data.   The p20 has a SNR of 73db while the DMR chip had only 68db.

Anyhow the fact that the technical definition of DR is not a useful measurement for photographers has also been beaten around quite a bit.  In my Imatest data the Canon actually had more DR than the leica according to the technical definition of DR (don't have the data handy but about 12stops to 11.5) , but using the threshold advised by the Imatest program for photographers it was a completely different story with the DMR having about 1.5 stops advantage (in this case about 10 stops to 8.5).  

Usable DR is a lot less than what you come up if you measure just the DR as technically defined and there are big differences in how these numbers shift by camera.  


Anyhow let's say that current line-up of cameras   D3, 5d2, A900 plus the MFDB are all pretty awesome.   I could see a reason to have several cameras.  
Eric






 



Quote from: carstenw
It is interesting to note that the people who claim that MFDBs are "better" generally talk about what they see, and what their customers see, i.e. a clear visual superiority for the MFDBs. The people who claim that the difference is very small are generally talking about statistics, plots, measured noise, and so on. There are a small number of people who go against the stream and claim that they can see no difference, or measure no difference.

This leads me to wonder about two things:

1) Could they both be true? In other words, could it be that the MFDBs give visually noticeably more appealing results, while there is little measurable difference? That would mean that what we measure doesn't translate directly to what we feel about the results. This would lead me to think that if we want to measure how good something is, we need to come up with new tests and new ways of measuring. Especially the DR results make me think this. One hears repeatedly how much cleaner the shadows are with MFDBs, but this somehow doesn't translate into better DR results, counter-intuitively, I suppose because what we like and what we can measure don't match.

2) Could it be that the 35mm-FF camera manufacturers know about the lack of significant measurable differences, and are actively pushing this angle by "cooking" the results to obtain the best possible *measured* result? They must surely know that very few people will ever see top-notch MFDB results, and that cooking the results to minimize the measurable differences helps them sell more cameras. This is somewhat analogous to graphics card manufacturers writing drivers which are specifically optimised for certain gaming performance benchmarks, while performing no better than other cards in actual play...
Title: DXOmark ranks DB image quality well below DSLR!
Post by: PeterA on April 26, 2009, 07:19:00 pm
I use different technology for different purposes. With MFD I get more megapixels and no AA filter. This trandslates to larger print sizes with a lot of detail. I also get the optionality to use a MFD back on a technical camera like my Alpa. This gives me access to wide angle lenses from Schneider and Rodenstock - which have no peer in 35mm land. I also get to sync @ up to 1/800th using my Fujiblad lenses - this is very useful - no?

For many shots though - there is no need for the hassle and technical precision required. I am not surprised that working pros have discovered that for many situations there are no great advantages between CaNikon and a MFD back. In many situations a 35mm system delivers strong advantages and practical functionalities that MFD backs and systems do not.

These tests -say nothing about how person uses a digital back in ways that a 35mm camera system cant be used. They just say a lot about the irrelevant ways where the Venn diagrams overlap in terms of functionality - and convenently ignore the major practical benefits of a MFD back.

To put it in another way - the size of a magazine spread hasnt changed and 35mm chips can more than adequately handle what is required for this sized shot. So what? The practical size of a panorama is now measured in meters when using a MFD back. 35mm stuff doesn't play this game well.

Title: DXOmark ranks DB image quality well below DSLR!
Post by: Bill Caulfeild-Browne on April 26, 2009, 07:19:02 pm
Quote from: carstenw
I am curious about this comment (genuinely curious, not looking to poke holes in it). I presume you are comparing the P45+ and the A900, not some other cameras. Could you explain in a bit more detail what the differences are that you see, and that you feel while working? Do you use the same software for both?

I would love a P45+, but my budget today barely stretches to an A900, which is why I am interested.

I wish I could show you prints on-line!

I used the Sony with the Zeiss 24-70 at approx. 50 mm, and the Mamiya AFD III with the 75-150 zoom at approx 75 mm - getting a subject to fill the frame horizontally, but not vertically. I printed the resulting images to 24 inches horizontally. (Using an iPF6100 on Epson Premium Luster or whatever they call it these days.)

The P45+ images resemble 4 by 5 in. images in detail, micro-contrast and gradation. There is a smoothness that's hard to describe but can be seen quite easily. The a900 files are very very good, but they "look digital" and just don't hold the detail as well.

If the RAW files from each camera were not so huge, I'd post them so you could decide for yourself. I do STRESS however, that at smaller sizes like A3, you'd be very hard pressed to tell the difference. I only see teal value in MFDB if you're going to print big. For all other work, especially wildlife, DSLRs are way superior.

Bill
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on April 26, 2009, 07:20:13 pm
Quote from: R_Medvid
Could you elaborate on this please?
The Phase Ones create nominally 16bit raw data. In fact, the pixel values are 16bit wide, i.e. they go up to 65535. However, about two bits of that are not only useless, they are actually detrimental; if the raw processing requires so much pushing, that the low-order two bits become apparent in the result, then the photographer becomes unhappy with the camera (back).
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Mort54 on April 26, 2009, 07:26:23 pm
Quote from: photolinia
I shoot about 50% in a studio and 50% on location.  I think in a studio MFDB wins by a mile, but on location it can be tricky.  
Basically, I'm at a cross road between a D3X and a used MFDB, and it sounds like given the choice you would go for a D3X...
If you really really really need to print big, and your revenues will benefit from printing big, then MFDBs make a lot of sense. However, it's all too easy to be seduced by those gorgeously detailed files, and to convince yourself you need MFDB when you really don't. Only you can answer that question. If I were you, I'd rent a D3X (or a 1DsIII) and shoot the same subjects that you shot with the MF system, and then print them side by side at the same print sizes you're currently selling to customers. Unless you are printing really big, I doubt you're going to see a huge difference, if any. And after all, it's prints you sell, not the files.

I haven't bought a D3X yet because, like you, I'm put off by Nikon's pricing strategy. A D3X costs way less than I paid for my P45+ system, of course, but I don't want to reward Nikon for what I consider to be a horrible marketing decision. Plus I already have the P45+ system for my high MP needs, so I'm reluctant to purchase something else that satisfies the same purpose. Nevertheless, knowing what I know now about how D3 files compare against  P45+ files, and the inconvenience of using two systems, and the inferiority of MF autofocus, metering, lenses, live view, etc, I think a D3X would be a better match for my needs than the P45+ system. If a D700X with the D3X sensor in it ever appears, it'll be a no brainer, and that's basically what I'm waiting for (although it's getting harder and harder to resist the D3X :-)
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on April 26, 2009, 07:32:04 pm
Quote from: EricWHiss
Usable DR is a lot less than what you come up if you measure just the DR as technically defined and there are big differences in how these numbers shift by camera
It depends on the definition of dynamic range adopted for the measurement. The engineering (scientific) definitions is IMO total waste of time and bandwidth relating to photography, so is DxO's 18% etc. measurement.

When I am measuring the noise/dynamic range, I don't presume any specific limit of acceptance of noise. For example the 5D2 noise graphs (http://www.panopeeper.com/Demo/Canon5DMkII_Noise.GIF) can be used by selecting the level of accepted noise on the y axis, and the x value shows the dynamic range, depending on the ISO (I have a collection of captures showing how the nosie of let's say 40% looks).

Of course, noise reduction is a subject on its own.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Graeme Nattress on April 26, 2009, 07:39:21 pm
Quote from: Panopeeper
This is BS. "Digital gain" in the current context is an oxymoron.

So you don't like the term - then suggest a more appropriate one and be positive about it rather than heckle and call BS. This can be a very useful and educational discussion....

Graeme
Title: DXOmark ranks DB image quality well below DSLR!
Post by: EricWHiss on April 26, 2009, 07:41:26 pm
Quote from: Panopeeper
The Phase Ones create nominally 16bit raw data. In fact, the pixel values are 16bit wide, i.e. they go up to 65535. However, about two bits of that are not only useless, they are actually detrimental; if the raw processing requires so much pushing, that the low-order two bits become apparent in the result, then the photographer becomes unhappy with the camera (back).

Gabor,
Can you tell us what kind of camera(s) you use, what you shoot, and if you make prints? Don't think it doesn't matter.  We all have our tricks on handling the files and experience doing real work is very valuable - much more so than pushing the files around in a program that measures noise.  Noise is subjective - there's nice noise that mimics what we are used to with film and there's ugly noise.     Don't forget that a lot of people add grain to files to get the look they want.   Noise is also added in up-sampling programs now.  Does that make you crazy or what?
Eric

Title: DXOmark ranks DB image quality well below DSLR!
Post by: Mort54 on April 26, 2009, 07:57:18 pm
Quote from: billcb
I do STRESS however, that at smaller sizes like A3, you'd be very hard pressed to tell the difference. I only see teal value in MFDB if you're going to print big. For all other work, especially wildlife, DSLRs are way superior.
That's the key. I totally agree with being hard pressed to see a difference at A3. That's true even with my D3 files compared to my P45+ files. At A2 I can more readily see a difference, but both prints still look fantastic. Someone thinking about getting a MFDB system should ask themselves if this difference is enough to justify the extra cost and inconvenience of going MFDB. For some people, the answer will be yes, for others it'll be no. If printing bigger than A2 helps the bottom line, the answer becomes more and more obvious in favor of MFDB the bigger you print.

If A2 were my final goal (and in most cases that's what I print), I think the money would be better spent on upgrading to the best primes available, making use of live view whenever possible to achieve more precise critical focus, and maybe even having the DSLR AA filter removed by a company like MaxMax. I think if someone did all those things, he'd even be hard pressed to see a difference at A2 between a D3X and P45+, and it would be way less expensive than buying a P45+ system.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: photolinia on April 26, 2009, 08:02:18 pm
Quote from: Mort54
If you really really really need to print big, and your revenues will benefit from printing big, then MFDBs make a lot of sense. However, it's all too easy to be seduced by those gorgeously detailed files, and to convince yourself you need MFDB when you really don't. Only you can answer that question. If I were you, I'd rent a D3X (or a 1DsIII) and shoot the same subjects that you shot with the MF system, and then print them side by side at the same print sizes you're currently selling to customers. Unless you are printing really big, I doubt you're going to see a huge difference, if any. And after all, it's prints you sell, not the files.

I haven't bought a D3X yet because, like you, I'm put off by Nikon's pricing strategy. A D3X costs way less than I paid for my P45+ system, of course, but I don't want to reward Nikon for what I consider to be a horrible marketing decision. Plus I already have the P45+ system for my high MP needs, so I'm reluctant to purchase something else that satisfies the same purpose. Nevertheless, knowing what I know now about how D3 files compare against  P45+ files, and the inconvenience of using two systems, and the inferiority of MF autofocus, metering, lenses, live view, etc, I think a D3X would be a better match for my needs than the P45+ system. If a D700X with the D3X sensor in it ever appears, it'll be a no brainer, and that's basically what I'm waiting for (although it's getting harder and harder to resist the D3X :-)


Agree - I do not need 39Mp.  I shot the same subject with a D700 + Nikkor 24-70 at about 45-50mm and H3D-II 39 with an 80mm.
Looking at the files on a computer screen without blowing up the images I now have trouble figuring out which one is which...

The huge difference I felt were during the retouching of the portraits - with the 39mp files it was a LOT simpler and more convenient to retouch the eyes and skin and the few times when I decided to do agressive crop after the shoot and was able to do it with a 39Mp files, but not with the 12M...

Of course, an H3D also looks a lot more impressive than a D700...

I had also decided to boycut the D3X and wait for a D700X, but I'm rethinking it now...

-ilya
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on April 26, 2009, 08:06:07 pm
Quote from: EricWHiss
We all have our tricks on handling the files and experience doing real work is very valuable - much more so than pushing the files around in a program that measures noise
This is not the same issue as the dispute over the use or uselessness of 14bit vs 12bit; the 16bit issue of the Phase Ones is not the question of noise. The two low order bits are not usable (not not useful) in the image reproduction.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on April 26, 2009, 08:26:28 pm
Quote from: Graeme Nattress
So you don't like the term - then suggest a more appropriate one and be positive about it rather than heckle and call BS
This is not the question of liking or not. Analogue ISO gain "reveals" image information captured by the sensor but not utilized at lower ISOs; that's the point of having them.

Multiplying the pixel values does not contribute to the useful image information. The current adoption of the fake ISOs with the DSLRs is a sad affair. There is absolutely no need to multiply the pixel values in order to make the resulting image brighter. MFDBs demonstrate this: the fake ISO is nothing but a tag.

It is important to note here, that the "real gain" of the analogue gain is limited: the higher analogue ISOs of almost all DSLRs don't deliver, i.e. their effect is worse than the digital adjustment.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Graeme Nattress on April 26, 2009, 08:35:36 pm
The "fake ISO" of DSLRs is done wrong though - they just apply a gain in post A-to-D digital signal before recording it, which brightens the image and throws away a stop. If  they'd left it alone and just metadata tagged +1EV or whatever, all would still be in the RAW and all would be good.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: cjmonty on April 26, 2009, 09:54:48 pm
Ive extensively used two Canon 21mp bodies, and now shoot with a Phamiya P45 refurb with assorted Mamiya glass.  With Canon i used Camera Raw or Lightroom, with Phase I use C1 4.7.

There is no comparison in image quality.  The Phase files are on a whole other level of clarity.  And not just in terms of 21MP vs 39MP relative size.

A Phase-derived image looks like my old 4x5 prints in describing detail.  The Canon prints should not be printed past 200% base resolution.

I could have been misusing the Canon gear, or have gotten two bad cameras, or should not have processed the files in Lightroom, but in my working experience some people are better served by the Canon and Nikon gear, and some are better served by MFDBs.  

However, neither is unequivocally better.

There are some very good reasons to stick with Canon

1- Not everyone prints large enough so the Canon dough-ball looking pixels become apparent.  When the prints are small, an invisible pixel is an invisible pixel.  If I didn't require a certain amount of detail in my work, I would have happily kept my Canon.  And several thousand dollars.

2- Much of the deficiency with Canon gear may be in the lenses, which when sharp, don't seem that sharp.  However, the whole problem is probably the AA filter.  As far as I can tell.  The description detail in a Canon image is rather fugly.  However, I have only had my work to judge this on... I would love to see a scientific comparison of Canon primes vs Mamiya vs Rodenstock/Schneider digitals in Helical mounts all compared on one sensor.

3- If I shot action, needed lowlight, or needed any real photographic versatility, I would happily have kept my the Canon.  And several thousand dollars.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on April 26, 2009, 10:20:07 pm
Quote from: Graeme Nattress
The "fake ISO" of DSLRs is done wrong though - they just apply a gain in post A-to-D digital signal before recording it, which brightens the image and throws away a stop. If  they'd left it alone and just metadata tagged +1EV or whatever, all would still be in the RAW and all would be good.
There is no justification for what the DSLR manufacturers are doing.

Some stick to the numerical range of the bit depth and fill that by multiplication; for example the Sony A700 multiplies the raw values at ISO 200 as well. In effect the camera fakes 12bit depth, when it uses in fact only less than 3200 levels.

The Nikon D300 stretches the range of red and blue by about 20% to "reach" the green range, instead of applying different white levels

The Nikon D200 stretches the red by 18.5% and the blue by 17%; the effect is, that the green clipping level (not a fixed value) is lower than the red and blue.

Canon is the worse: the clipping level of the newer models depends on the ISO, but still the fake ISOs are achieved by multiplication instead of by different white levels.

All this with the only effect of reducing the dynamic range. I would not believe it if I did not prove it myself.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: EricWHiss on April 26, 2009, 11:02:49 pm

Quote from: Panopeeper
This is not the same issue as the dispute over the use or uselessness of 14bit vs 12bit; the 16bit issue of the Phase Ones is not the question of noise. The two low order bits are not usable (not not useful) in the image reproduction.

Come on  - the two lowest bits represent 0, 1, and 2 values out of a total of 65536.  Who gives a cr*p about that? Most people adjust the black point higher anyhow even if they push up the shadows.  The problem is chromatic noise and blob type of noise and banding up in the 1/4 tones of the image, not those two last bits.  And back to the original topic - I am not certain you can take DXO's numbers across from DSLR to MFDB like apples to apples.  


As far as DR - thanks for clarifying how DXO is defining DR.  I did not know they were using anything other than the technical definition as the numbers they report seem too high overall to be anything else.

As to Fake ISO -  actually I understand why a camera company would do this - its just easier to operate the camera this way with exposure metering and exposure checking on the histogram.   If I were making a camera I'd probably do the same thing.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on April 26, 2009, 11:49:54 pm
Quote from: EricWHiss
Come on  - the two lowest bits represent 0, 1, and 2 values out of a total of 65536
and 3 :-( (i.e. four values out of 65536). On the other hand, this is not an issue of bits, that is only a convenient way of expressing such aspects.

Quote
Who gives a cr*p about that?
Let's keep things straight.

1. I did not criticize the image quality of the Phase One cameras. In fact I regard those cameras very highly.

2. I stated, that the 16bit depth is in practice only 14bit. You became outraged reading that, and now you are saying "who gives a cr*p about that".

So, pls make up your mind against what and for what you are arguing, i.e. if those bits are useful or not, or perhaps not even worth of arguing about.

Quote
Most people adjust the black point higher anyhow even if they push up the shadows
You can't adjust the black level in terms of raw pixel values. Nevertheless, you are right in that those pixels are not used in most cases. I wrote before: The two low order bits are not usable (not not useful) in the image reproduction.

Quote
The problem is chromatic noise and blob type of noise and banding up in the 1/4 tones of the image, not those two last bits
Some of the banding (or all of it) is caused by those bits (i.e. by the far too great depth of the data, which represents tiny sensor related issues, not image information).

Quote
I am not certain you can take DXO's numbers across from DSLR to MFDB like apples to apples
Actually, I don't take DxO's nor anyone else's numbers at all. I prefer my own measurements, which I can document and demonstrate. If I don't have raw files suitable for such measurements, then I don't make any related statement.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on April 27, 2009, 12:09:51 am
Eric, I was writing my post and have not noticed that you added to your previous post-

Quote from: EricWHiss
As far as DR - thanks for clarifying how DXO is defining DR
Hold on; I described, how I do the measurements (the graphs I linked to are mine). I don't know how exactly to interpret DxO's measurements.

Quote
As to Fake ISO -  actually I understand why a camera company would do this - its just easier to operate the camera this way with exposure metering and exposure checking on the histogram.   If I were making a camera I'd probably do the same thing
Don't be so sure you would do the same.

One of the fanciests is the Canon 5D2. I simply copy the definitions from the Rawnalyze program code. The second column in the red clipping level, the third is the green and the fourth is the blue; the fifth plays no role here. As you see, there are different clipping levels, depending on the ISO. Is that what you think makes it easier for the firmware?

ISOSat(25600,16383, 16383, 16383, 16384)
ISOSat(12800,16383, 16383, 16383, 16384)
ISOSat(6400,   16383, 16383, 16383, 16384)
ISOSat(5000,   15755, 15760, 15760, 15800)
ISOSat(4000,   15755, 15760, 15760, 15800)
ISOSat(3200,   15755, 15760, 15760, 15800)
ISOSat(2500,   15755, 15760, 15760, 15800)
ISOSat(2000,   15755, 15760, 15760, 15800)
ISOSat(1600,   15755, 15760, 15760, 15800)
ISOSat(1250,   12810, 12810, 12810, 12850)
ISOSat(1000,   15755, 15760, 15760, 15800)
ISOSat(800,   15755, 15760, 15760, 15800)
ISOSat(640,   12810, 12810, 12810, 12850)
ISOSat(500,   15755, 15760, 15760, 15800)
ISOSat(400,   15755, 15760, 15760, 15800)
ISOSat(320,   12810, 12810, 12810, 12850)
ISOSat(250,   15755, 15760, 15760, 15800)
ISOSat(200,   15755, 15760, 15760, 15800)
ISOSat(160,   12810, 12810, 12810, 12850)
ISOSat(125,   15755, 15760, 15760, 15800)
ISOSat(100,   15755, 15760, 15760, 15800)
ISOSat(50,   15755, 15760, 15760, 15800)

and then the Nikon D3X:

ISOSat(6400,4095,4095,4095,4096)
ISOSat(3200,4095,4095,4095,4096)
ISOSat(1600,4095,3830,4095,4096)
ISOSat(800, 4095,3830,4095,4096)
ISOSat(400, 4095,3830,4095,4096)
ISOSat(200, 4095,3830,4095,4096)
ISOSat(100,   4095,4095,4095,4096)
ISOSat(50,   4095,4095,4095,4096)
ISOSat(0,   4095,4095,4095,4096)


Title: DXOmark ranks DB image quality well below DSLR!
Post by: Guillermo Luijk on April 27, 2009, 05:10:24 am
Quote from: carstenw
Especially the DR results make me think this. One hears repeatedly how much cleaner the shadows are with MFDBs, but this somehow doesn't translate into better DR results, counter-intuitively, I suppose because what we like and what we can measure don't match.
Are you sure those who say how much cleaner the shadows are with MFDBs (funily without specifying which MFDB nor which DSLR) really know how to evaluate noise in the shadows of their MFDBs? even if they are MFDBs selling staff who think they know what they are talking about, or they are people making a living on using these backs and thinking making a living on using their backs is the only way to have a founded opinion.

This is a 100% crop of a H3D II:
(http://img129.imageshack.us/img129/3238/hasselzonl.jpg)

The black figure is located in the RAW range 9 stops from saturation. I am pretty sure the level of noise in the 9th stop on a Nikon D3X is the same or lower than this.

The sensor in a MFDB is larger than on a DSLR, that means with the same tecnology you can build a higher DR MFDB than the equivalent DSLR. But that does _not_ mean that all MFDBs will provide more DR than all DSLRs.

BR


Title: DXOmark ranks DB image quality well below DSLR!
Post by: Carsten W on April 27, 2009, 06:39:57 am
Quote from: GLuijk
The black figure is located in the RAW range 9 stops from saturation. I am pretty sure the level of noise in the 9th stop on a Nikon D3X is the same or lower than this.

It is an interesting image, but you left out a lot of information. At what ISO is this? Which H3DII is it? Do you have a similar crop from a D3X to demonstrate your assertion?
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Guillermo Luijk on April 27, 2009, 07:10:13 am
Quote from: carstenw
It is an interesting image, but you left out a lot of information. At what ISO is this? Which H3DII is it? Do you have a similar crop from a D3X to demonstrate your assertion?

Filename: hassel.dng
Timestamp: Thu Nov 13 16:54:51 2008
Camera: Hasselblad H3D II-31
DNG Version: 1.0.0.0
ISO speed: 100

Anyway ISO is irrelevant here, since all ISOs in this back are ISO50.
I have never seen any RAW file from Nikon D3X but I did for a Nikon D3 having similar noise in the 9th stop (signed as -8EV here):

(http://img504.imageshack.us/img504/7571/zonas2oq6.jpg)

I think the quality advantage for backs is not noise, DR, colour rendition,... but just sharpness and resolution, i.e. detail. In the same way as a Leica M8 with an average sensor can produce very good quality images thanks to the lack of AA filter and superb optics, backs have the same strongholds: no AA filter and good lenses. The sensor in backs is not better than in DSLRs.

BR
Title: DXOmark ranks DB image quality well below DSLR!
Post by: David Grover / Phase One on April 27, 2009, 07:30:00 am
Quote from: GLuijk
Filename: hassel.dng
Timestamp: Thu Nov 13 16:54:51 2008
Camera: Hasselblad H3D II-31
DNG Version: 1.0.0.0
ISO speed: 100

Anyway ISO is irrelevant here, since all ISOs in this back are ISO50.
I have never seen any RAW file from Nikon D3X but I did for a Nikon D3 having similar noise in the 9th stop (signed as -8EV here):

(http://img504.imageshack.us/img504/7571/zonas2oq6.jpg)

I think the quality advantage for backs is not noise, DR, colour rendition,... but just sharpness and resolution, i.e. detail. In the same way as a Leica M8 with an average sensor can produce very good quality images thanks to the lack of AA filter and superb optics, backs have the same strongholds: no AA filter and good lenses. The sensor in backs is not better than in DSLRs.

BR

Converting to DNG will see a significant rise in noise levels compared to retaining our 3F format and using Phocus.  Therefore showing such an image is an unfair evaluation.  I think this shows as that is a truly appalling example of a 100 ISO shot.  What did you do to make it so unusable?

Also saying 'all ISOs in this back are ISO50' is misleading to customers.  As I have said many times before on these frustrating threads is that when a photographer sets ISO400 or whatever on his camera, he/she wants it to behave in this way and give feedback via the light meter and histogram to that effect.

How the image is then translated via the DSP in the back (which I will say again varies depending on the ISO setting) and furthermore in the processing software to the end result is of little concern to many.

ISO is probably an ancient term which should not be used with digital, but it is what we know and is what is familiar to us photographers.

Best,



David
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Guillermo Luijk on April 27, 2009, 08:08:07 am
Quote from: David Grover / Hasselblad
Converting to DNG will see a significant rise in noise levels compared to retaining our 3F format and using Phocus.
Brilliant.

Title: DXOmark ranks DB image quality well below DSLR!
Post by: Carsten W on April 27, 2009, 08:11:41 am
Quote from: GLuijk
Anyway ISO is irrelevant here, since all ISOs in this back are ISO50.

Well, no. If you shoot at "fake" ISO 100 or 200 or 400, you are under-exposing, in effect, and thereby reducing the quality of the image. This is not irrelevant at all. Given that MF solutions are inherently much less flexible than DSLRs, there is only one reason to go the MF route, and that is to try to extract the absolute best image out of it that you can, and this means shooting at base ISO, IMO. If you do anything else, or get sloppy with processing, yes, you may as well use a DSLR. If you do go the extra mile, you should be rewarded with visibily better detail and shadows, from what I understand, but there are no shortcuts (until the S2 ).

I am interested in the A900 because it doesn't have Canon reds or Nikon greens, and the AA filter is apparently noticeably less strong than with either of those competitors, perhaps excepting the D3X, which is as expensive as a second-hand MF back in any case. Of course, there are also the Zeiss-Sony lenses to consider, which for me could be a big deal. The A900 is also meant to have very good skin tones. I am interested in what situations you are likely to miss the extra quality of the DB. So far, it seems that printing larger than A2 is one situation. Maybe there are others?
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Guillermo Luijk on April 27, 2009, 08:15:28 am
Quote from: carstenw
Well, no. If you shoot at "fake" ISO 100 or 200 or 400, you are under-exposing, in effect, and thereby reducing the quality of the image.
I know, but according to the ISO plots from DxO Mark and what Gabor commented (fake ISOs in this back are not software corrected, i.e. they are just metadata), the RAW data captured is the same at ISO50 as at ISO100, so the area analysed was actually 9 stops below sensor saturation, no matter what ISO was set in the camera.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: David Grover / Phase One on April 27, 2009, 08:16:59 am
Quote from: GLuijk
Brilliant.

I do my best.

But rather than take my word for it, simply try a test for yourself on a high (fake fake fake!) ISO shot.

Best,


David
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Dustbak on April 27, 2009, 08:53:14 am
If anything, you would want to test on a genuine Raw file wouldn't you, instead of something that is a conversion from the real thing? (where just about everyone that is using/involved with Hasselblad knows by now that DNG conversion results in loss of quality.)

Title: DXOmark ranks DB image quality well below DSLR!
Post by: Guillermo Luijk on April 27, 2009, 09:15:08 am
Quote from: Dustbak
If anything, you would want to test on a genuine Raw file wouldn't you, instead of something that is a conversion from the real thing? (where just about everyone that is using/involved with Hasselblad knows by now that DNG conversion results in loss of quality.)
Yes and no. If a genuine file from the back produces a better result when processed with the camera software, is because software corrections are being applied, not because there was a loss of quality when converting to DNG (or you think that for some strange reason noise will increase in the conversion?).
In addition to this the comparision to any other camera would not be fair since not the same RAW developer would have been used for both. That is why I always use DCRAW and ignore any metadata found in the RAW file.

Anyway I don't have any other file from a Hasselblad. Those who have are invited to show.

BR

Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on April 27, 2009, 02:28:33 pm
Quote from: carstenw
If you shoot at "fake" ISO 100 or 200 or 400, you are under-exposing, in effect, and thereby reducing the quality of the image. This is not irrelevant at all
If the analogue gain is identical for different ISOs, then the ISO selection is only for metering, thus the shot gets underexposed with higher ISO.

However, when you measure the noise relative to the pixel intensity, then the exposure of  the image as a whole plays no role. One selects untextured, smooth, uniform, evenly lit patches and measure the noise on them. The intensity of the selected patch is of relevance.

Following is a crop from a 5D2 shot woth ISO 1600 (created by out fellow poster Marc specifically for the measurement). I did not want to make huge captures, so this shows only six of the 24 patches of a color checker (printed by Marc). This shows the measured average intensities of the patches and the noise in percentage of the average intensity (the selection is on the first patch). Note, that the noise measurement has nothing to do with how bright the patch is displayed here.

The other captures show only the red, the green and the blue components, to give a feeling how much noise for example 43.4% is. Of course the appearance depends on how much the intensity is boosted. The paper The Source Of Noise (http://www.cryptobola.com/PhotoBola/SourceOfNoise.htm) contains samples brought closely to the same intensity to give a better feeling.

(http://www.panopeeper.com/Noise/MeasuringNoise_Sample_Composite.GIF)

(http://www.panopeeper.com/Noise/MeasuringNoise_Sample_Red.GIF)

(http://www.panopeeper.com/Noise/MeasuringNoise_Sample_Green.GIF)

(http://www.panopeeper.com/Noise/MeasuringNoise_Sample_Blue.GIF)
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on April 27, 2009, 02:49:10 pm
Quote from: GLuijk
Yes and no. If a genuine file from the back produces a better result when processed with the camera software, is because software corrections are being applied, not because there was a loss of quality when converting to DNG (or you think that for some strange reason noise will increase in the conversion?).
In addition to this the comparision to any other camera would not be fair since not the same RAW developer would have been used for both. That is why I always use DCRAW and ignore any metadata found in the RAW file
This is not a straightforward issue.

1. Most cameras do some adjustment of the raw data, usually with the aim of reducing the noise. There is nothing wrong with that, as long as those adjustments do not occur on the cost of some other aspect. For example the Sony noise reduction on the raw data greatly reduces the resolution; that is not acceptable without the user wanting that.

2. The technologies are different; there is no reason to declare that Canon's or Nikon's approach is the way to go. For example Canon keeps the negative pixel values as well, leaving it to the raw processing to remove as much as it wants to, with the algorythm it decides for. Nikon's do that in-camera. If Nikon did not carry out the adjustment in-camera but it passed the related info in a separate file, would you reject that because it can not be converted in DNG format?

3. The DNG format is not up to the job; so plain and simple. There is no basis to say that if something can not be converted in DNG, then it should be disregarded.

On the other hand, I do not accept blindly that the Hasselblad raw files could not be converted in DNG; perhaps they are not converted properly, because Hasselblad and/or Phocus don't give a damn for their customers' workflow. For example the Sinar files (multiple files for one raw image) are converted in DNG by two non-Adobe converters, differently but resulting largely in the same final output except for colors (which is the issue of the converter, not of DNG).

Keep in eyes, that Phocus does not make business by creating input files for Adobe.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: David Grover / Phase One on April 27, 2009, 03:46:40 pm
Quote from: Panopeeper
On the other hand, I do not accept blindly that the Hasselblad raw files could not be converted in DNG; perhaps they are not converted properly, because Hasselblad and/or Phocus don't give a damn for their customers' workflow. For example the Sinar files (multiple files for one raw image) are converted in DNG by two non-Adobe converters, differently but resulting largely in the same final output except for colors (which is the issue of the converter, not of DNG).

Keep in eyes, that Phocus does not make business by creating input files for Adobe.

Hasselblad files can be and are converted by Phocus to DNG.  I am simply saying that using Phocus you will get a better result - plain and simple.

However, if you want the benefit of our lens corrections, the way we handle noise and color and other aspects of Phocus then it is there for use, for free.

We write a DNG file conforming to the spec as anyone else would.  What happens after that point it beyond our control.

I am not sure I understand the venom towards us in this respect.  Already, 3F/3FR files are supported by Aperture and we hope the same for Adobe as well.  In the meantime there is the DNG option or simply to use Phocus and get a better result than above.

If there is an impression that we fudge the DNG conversion to make us look favorable, then I am afraid in our evil underground lair, we are not that clever.

Best,


David


Title: DXOmark ranks DB image quality well below DSLR!
Post by: BJL on April 27, 2009, 05:16:25 pm
About digital camera exposure index settings (a.k.a. ISO speeds, though they are not at all the same thing as the ISO speed rating of a film): there is no point applying different analog gain to different photographs once a single choice of gain achieves two things:
a. full well signals are converted to a voltage that is within what the subsequent processing can handle without clipping
b. the dark noise level in the signal is amplified to a level comfortably above all subsequent noise sources, including quantization error in A/D conversion, so that the amplified signal is strong enough to not be significantly affected by downstream noise sources.

Both conditions are are probably met by a MFDB with a good 16-bit or even a true 14-bit A/D convertor, because the gain amplification is done just before A/D conversion, so that the A/D convertor is the only downstream noise source to worry about, and the DR of such an A/D convertor should be far greater than the DR of the signal being converted. Amplifying a low light (high exposure index) signal more would simply risk blowing out highlights while not improving noise levels in any significant way.

In other words, for MFDB's with  adjusting of final output levels in the digital domain ("fake") might be as good as or better than adjusting them in the analog domain ("real") when dealing with high expsoure index shooting; that is, in limited light situations where the sensor has to be given less than full exposure, so that no photosites get close to full well signals.


Variable gain is far more useful in CMOS sensors, and when using lesser A/D convertors, like 12-bit and maybe even lower quality 14-bit. It is useful with CMOS because the variable gain can be applied far earlier in the signal path, such as at the bottom of each column of pixels, so that there are more noise sources downstream of the amplification, like noise arising in transportation across the sensor. Emil Martin has shown us good evidence that this noise down-stream of variable analog gain is the dominant noise source with some Canon DSLR's at low to moderate gain levels (low to moderate ISO speed settings.)
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Carsten W on April 27, 2009, 05:39:34 pm
Quote from: Panopeeper
For example the Sony noise reduction on the raw data greatly reduces the resolution; that is not acceptable without the user wanting that.

I presume you are talking about the A900? The NR can be turned off, can it not?
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on April 27, 2009, 07:24:32 pm
Quote from: David Grover / Hasselblad
If there is an impression that we fudge the DNG conversion to make us look favorable, then I am afraid in our evil underground lair, we are not that clever.
David, you posted earlier

Quote
Converting to DNG will see a significant rise in noise levels compared to retaining our 3F format and using Phocus. Therefore showing such an image is an unfair evaluation.

On the other hand, I found this a Phocus brochure (http://www.cameraelectronic.com.au/pdf/hasselblad/uk_phocus_v2_web.pdf):

The extra data contained in the Hasselblad 3FR format, and the corresponding set of image refining algorithms contained does not fit into today’s DNG format. Phocus does however, include a full export function for exporting 3FR files in the DNG format, delivering good color rendering, but with no optical lens corrections

Now we see three positions:

1. the DNG conversion is not suboptimal regarding noise

2. the DNG conversion results in higher noise

3. the DNG conversion has nothing to with noise

Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on April 27, 2009, 07:45:01 pm
Quote from: carstenw
I presume you are talking about the A900? The NR can be turned off, can it not?
Not for low ISOs, but most owners beleive it can. See a very technical study (http://www.cryptobola.com/PhotoBola/SonyA900/SonyA900_NR.htm) of this topic (although I thought you had seen this on GetDPI).
Title: DXOmark ranks DB image quality well below DSLR!
Post by: ErikKaffehr on April 28, 2009, 01:07:47 am
Hi,

What about sharpening? The Canon needs a lot of capture sharpening because of the AA-filter. A MFDB needs much less sharpening.

No doubt that MFDBs resolve better than DSLRs this discussion is about noise, where some of the differences between DSLRs and MFDBs is less well understood.

If you are comparing Canon with Camera Raw and Phase One with Capture one, what are you comparing Phase One to Canon or C1 to CR?

Best regards
Erik

Quote from: cjmonty
Ive extensively used two Canon 21mp bodies, and now shoot with a Phamiya P45 refurb with assorted Mamiya glass.  With Canon i used Camera Raw or Lightroom, with Phase I use C1 4.7.

There is no comparison in image quality.  The Phase files are on a whole other level of clarity.  And not just in terms of 21MP vs 39MP relative size.

A Phase-derived image looks like my old 4x5 prints in describing detail.  The Canon prints should not be printed past 200% base resolution.

I could have been misusing the Canon gear, or have gotten two bad cameras, or should not have processed the files in Lightroom, but in my working experience some people are better served by the Canon and Nikon gear, and some are better served by MFDBs.  

However, neither is unequivocally better.

There are some very good reasons to stick with Canon

1- Not everyone prints large enough so the Canon dough-ball looking pixels become apparent.  When the prints are small, an invisible pixel is an invisible pixel.  If I didn't require a certain amount of detail in my work, I would have happily kept my Canon.  And several thousand dollars.

2- Much of the deficiency with Canon gear may be in the lenses, which when sharp, don't seem that sharp.  However, the whole problem is probably the AA filter.  As far as I can tell.  The description detail in a Canon image is rather fugly.  However, I have only had my work to judge this on... I would love to see a scientific comparison of Canon primes vs Mamiya vs Rodenstock/Schneider digitals in Helical mounts all compared on one sensor.

3- If I shot action, needed lowlight, or needed any real photographic versatility, I would happily have kept my the Canon.  And several thousand dollars.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: andershald on April 28, 2009, 06:15:42 am
DXOlabs Suck anyway!

I bought the Filmgrain software from them, because I like the look and the way the effects are applied. The software kept crashing when I was usingit with my P30+ files (90mb). I contacted their support and they were surprised that I was using such extremely large files. Their conclusion was: I had to reduce my file sizes and (I quote Jeff): "Most images range from a few Mb to about 20 Mb."

So according to DXO normal image files are <20mb any thing above that is extreme!

When I asked for a refund, as the software is useless to me, and nowhere in the documentation is it specified that it only works with smaller files, they simply refused.

So I have very little respect for DXOlabs, they must have some kind of DSLR agenda going...
Title: DXOmark ranks DB image quality well below DSLR!
Post by: R_Medvid on April 28, 2009, 06:04:24 pm
Quote
When I asked for a refund, as the software is useless to me, and nowhere in the documentation is it specified that it only works with smaller files, they simply refused.

This is absolutely ridiculous.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: focusgroup on April 28, 2009, 09:20:17 pm
May I suggest simply using your eyes people? !??!!  Make up your own mind by looking at whatever medium you generally use or distribute.  Sheesh
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Nick-T on April 28, 2009, 10:33:03 pm
Quote from: focusgroup
May I suggest simply using your eyes people? !??!!


What? Are you mad! Photography isn't about what you SEE!!! It's about numbers! Sheesh.

Nick-T
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Bill Caulfeild-Browne on April 28, 2009, 10:42:10 pm
Quote from: Nick-T
What? Are you mad! Photography isn't about what you SEE!!! It's about numbers! Sheesh.

Nick-T


Aw c'mon - it's about GEAR! Sheesh....

Bill
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Ray on April 29, 2009, 12:48:00 am
Quote from: focusgroup
May I suggest simply using your eyes people? !??!!  Make up your own mind by looking at whatever medium you generally use or distribute.  Sheesh

I agree. Use your eyes. That's what I do. The problem is, we're all creatures of habit, to some degree at least. Different formats of cameras require different techniques and different lenses, so the whole issue becomes terribly confused when the better image from an MFDB might be due to the better performance of the lens at a particular aperture, or simply the lack of an AA filter which produces the semblance of higher resolution.

The main issue in this thread is that the MFDB does not appear to have a significantly higher DR and SNR than the 35mm format D3X, for whatever reason, and even at base ISO.

If anyone wants to dispute these DXO results, then show us the comparisons at equal ISO at least. I would not insist upon equal DoF and equal shutter speed, because then you would be stuffed. The D3X would triumph in everything except resolution. The sensor with the higher pixel count usually produces the higher resolution. That's completely understandable and I would not attempt to contest that point.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: ziocan on April 29, 2009, 01:19:31 am
Does it really matter?
I mean who bloody care about what DXO guys think.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on April 29, 2009, 11:54:12 am
Quote from: focusgroup
May I suggest simply using your eyes people? !??!!  Make up your own mind by looking at whatever medium you generally use or distribute.  Sheesh
May I ask what exactly "people" should use their eyes for? Honestly, do you have the slightest idea what the discussion is about?
Title: DXOmark ranks DB image quality well below DSLR!
Post by: ziocan on April 29, 2009, 12:21:01 pm
Quote from: Panopeeper
May I ask what exactly "people" should use their eyes for? Honestly, do you have the slightest idea what the discussion is about?
is about:
Stop! otherwise you go blind!
Title: DXOmark ranks DB image quality well below DSLR!
Post by: focusgroup on April 29, 2009, 01:59:20 pm
Quote from: ziocan
is about:
Stop! otherwise you go blind!

Honestly do you have a clue?  This is they typical bs people discuss when they have little else to do and are dilettantes.  I mean who gives a care what DXO says?  Would you really buy a kit or make any sort of significant investment using their limited testing as a valuable resource?  

Yes, I have a bit of experience with this subject.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: BJL on April 29, 2009, 03:18:27 pm
Quote from: billcb
Aw c'mon - it's about GEAR! Sheesh....
No no, the game of "forum fantasy fotography" it is not just about numbers or just about gear: it is about proving the superiority of your gear with a single number, and arguing which single numerical measure should be used: MP, ISO, DR, DxOMark ...
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Ray on April 30, 2009, 05:27:39 am
Quote from: BJL
No no, the game of "forum fantasy fotography" it is not just about numbers or just about gear: it is about proving the superiority of your gear with a single number, and arguing which single numerical measure should be used: MP, ISO, DR, DxOMark ...

No, no, no! It's about holding the manufacturers to acount. It's about getting precise and relevant information about the performance of equipment so that we end up buying what we thought we were buying.

MFDB owners who thought they were paying a huge premium for greater DR and lower noise than any 35mm format could offer, now have to reconsider in the light of the DXO results. However, DXO does not test resolution or the quality of specific lenses. There's no substitute for a really good lens, and a 39mp sensor will definitely produce higher resolution than a 21 or 24mp sensor, lens permitting.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: erick.boileau on April 30, 2009, 06:10:41 am
DXOmark results are a nonsense when you have in your computer thousands of photo with Canon 1Ds Mark III and P45 ...
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Guillermo Luijk on April 30, 2009, 08:02:14 pm
Quote from: focusgroup
I mean who gives a care what DXO says?
Among others, you. Otherwise you'll have to explain why you decided to invest your time in reading and posting on a thread entitled as 'DXOmark ranks DB image quality well below DSLR!'.

BR
Title: DXOmark ranks DB image quality well below DSLR!
Post by: BernardLanguillier on April 30, 2009, 08:50:23 pm
Quote from: erick.boileau
DXOmark results are a nonsense when you have in your computer thousands of photo with Canon 1Ds Mark III and P45 ...

Dxo Marks gives the P45+ 0.9 stop more DR than the 1ds3, don't you think that this is about right?

http://www.dxomark.com/index.php/eng/Image...d2)/Phase%20One (http://www.dxomark.com/index.php/eng/Image-Quality-Database/Compare-cameras/(appareil1)/291%7C0/(appareil2)/304%7C0/(onglet)/0/(brand)/Canon/(brand2)/Phase%20One)

In what sense does that make no sense?

Regards,
Bernard
Title: DXOmark ranks DB image quality well below DSLR!
Post by: selsoe on May 14, 2009, 10:21:05 am
Phase One P65+ just took the lead ahead of Nikon D3x:
http://www.dxomark.com/index.php/eng/DxOMark-Sensor (http://www.dxomark.com/index.php/eng/DxOMark-Sensor)
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Jonathan Wienke on May 14, 2009, 11:30:16 am
Quote from: lisa_r
I think it's interesting that anywhere above 50 ISO the DSLRs beat the backs in terms of D.R. Often by one or two full stops, even though the backs have larger pixels. I am really wondering where the truth lies.

It's because if you use "fake" ISO settings, which most MFDB do for their higher settings, every stop above the highest "real" ISO comes directly out of DR. So if the back's highest real ISO is 50, and you shoot at ISO 400, then you're throwing away 3 stops of highlights. This will reduce the usable DR in the captured image by 3 stops. When compared to a DSLR that has a real ISO 400, the MFDB has a 3-stop DR handicap, and may very well do worse.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: gss on May 14, 2009, 02:10:57 pm
Quote from: David Grover / Hasselblad
ISO is probably an ancient term which should not be used with digital, but it is what we know and is what is familiar to us photographers.
I would agree with this if you had said ASA; ISO is just an upstart.  
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Eric Myrvaagnes on May 14, 2009, 02:12:55 pm
I have just created the ultimate solution to this pressing question. I have just defined a new Universal Quality Index (or "UQI", pronounced "Yucky!") Keeping in mind the KISS Principle, I use a binary system. Any camera, lens, or other piece of photographic equipment, or photograph, photographer, photographer's assistant, painting, fine dining experience, fast food meal, comic strip, web site, etc., which I like gets a UQI value of 1. Anything i don't like gets a 0.

As an example, the Luminous Landscape website and forum both rate a 1. Some individual threads on the forum (this one, for example) do not.



Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on May 14, 2009, 07:27:00 pm
Quote from: Jonathan Wienke
It's because if you use "fake" ISO settings, which most MFDB do for their higher settings, every stop above the highest "real" ISO comes directly out of DR
Not so. This is the way DSLRs adopt ISO steps without hardware support; this is an inexcusable dumb rubbish from the DSLR manufacturers.

MFDBs, at least those I have analyzed, work differently: they do not ruin the original pixel values by senseless multiplication (or division); they simply state, that the numerical range of the pixel values is now lower than with another ISO, but this occurs without causing clipping of the pixel values.

Thus the DR of the MFDBs does not change with the fake ISO increase, in contrast to DSLRs.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: ziocan on May 14, 2009, 10:50:28 pm


geeeeeee!
Title: DXOmark ranks DB image quality well below DSLR!
Post by: lisa_r on May 14, 2009, 11:53:58 pm
Quote from: Panopeeper
Not so. This is the way DSLRs adopt ISO steps without hardware support; this is an inexcusable dumb rubbish from the DSLR manufacturers.

MFDBs, at least those I have analyzed, work differently: they do not ruin the original pixel values by senseless multiplication (or division); they simply state, that the numerical range of the pixel values is now lower than with another ISO, but this occurs without causing clipping of the pixel values.

Thus the DR of the MFDBs does not change with the fake ISO increase, in contrast to DSLRs.

What's the reason for the decrease in DR as the ISO increases as can be seen over at DXOmark? IT looks like as soon as you hit ISO 100 or above the DSLRs win in terms of D.R.. SLick on yhe dymanic range tab here:

http://dxomark.com/index.php/eng/Image-Qua.../(brand3)/Canon (http://dxomark.com/index.php/eng/Image-Quality-Database/Compare-cameras/(appareil1)/304|0/(appareil2)/318|0/(appareil3)/305|0/(onglet)/0/(brand)/Phase%20One/(brand2)/Phase%20One/(brand3)/Canon)


Title: DXOmark ranks DB image quality well below DSLR!
Post by: ejmartin on May 15, 2009, 01:32:09 am
Quote from: lisa_r
What's the reason for the decrease in DR as the ISO increases as can be seen over at DXOmark? IT looks like as soon as you hit ISO 100 or above the DSLRs win in terms of D.R.. SLick on yhe dymanic range tab here:

http://dxomark.com/index.php/eng/Image-Qua.../(brand3)/Canon (http://dxomark.com/index.php/eng/Image-Quality-Database/Compare-cameras/(appareil1)/304|0/(appareil2)/318|0/(appareil3)/305|0/(onglet)/0/(brand)/Phase%20One/(brand2)/Phase%20One/(brand3)/Canon)

The analog-to-digital converter (ADC) in the camera, which takes the analog signal from the sensor and digitizes it, has a maximum output value.  Every time the ISO is doubled, the amplification is doubled, and one more stop of exposure levels is pushed past the saturation point of the ADC.  Therefore the DR will decrease by one stop unless there is a compensating reduction in shadow noise.   In MFDB's there is typically no reduction in shadow noise with ISO increase, so DR decreases by about a stop for every stop increase in ISO (if ISO is implemented as a hardware amplification; if not there are some subtleties).

In Canon and most other CMOS sensor DSLR's, there is a reduction in shadow noise with the first few stops of ISO increase above base ISO (because the ISO amplifier and ADC are noisier than they ought to be, and increasing the ISO decreases their contribution to shadow noise), and so one sees the DR curve to be relatively flat for those first couple of stops.  Eventually the improvement in shadow noise ends, however, and the DR starts to drop.  Note, however, that if Canon were able to use cleaner electronic components for the ISO amplifier and ADC, their DR would be over two stops better than the MFDB for all ISO, not just high ISO; and would be about 14 stops at base ISO.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on May 15, 2009, 01:56:48 am
Quote from: lisa_r
What's the reason for the decrease in DR as the ISO increases as can be seen over at DXOmark? IT looks like as soon as you hit ISO 100 or above the DSLRs win in terms of D.R.
I don't know how they arrived at this result. However, the P45 Plus DOES have analogue ISO up to 800, at least I don't see any sign of digital manipulation. On the other hand, analogue supported ISO does not necessarily mean any noise reduction. For example the 5D2 has analogue gain up to 3200, but the loss of dynamic range from 1600 to 3200 is a full stop, because the analogue gain of 3200 is ineffective, compared to 1600.

I don't have P45 Plus raw files suitable for measurement, but guestimated from what I have, it would have been better not to adopt ISO 800 in the hardware. I don't know if ISO 400 is effective vs. ISO 200.

There is another side of this issue, for those using LR/ACR: Adobe did not notice, that the P45 Plus has hardware ISOs (I guess the P45 does not have) and the raw conversion automatically "makes up" for the "missing ISO" by applying +1 EV with ISO 400 and +2 EV with ISO 800 as "Exposure" adjustment (invisible on the slider). However, there is no missing ISO, thus ACR makes the images appear much brighter than they should be, it can even cause clipping, thereby reducing the apparent DR even more.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: douglasf13 on May 15, 2009, 02:25:00 am
Quote from: Panopeeper
Not for low ISOs, but most owners beleive it can. See a very technical study (http://www.cryptobola.com/PhotoBola/SonyA900/SonyA900_NR.htm) of this topic (although I thought you had seen this on GetDPI).

Carsten, talk to Iliah Borg about this, as his studies show something very different...and he has multiple A900 cameras at his disposal to test.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: BJNY on May 17, 2009, 03:53:33 pm
P65+ just received the highest score at dxomark.com database
beating previous leader Nikon D3x.

Probably pertains to P40+ as well since it's just a smaller wafer of same sensor?
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Ray on May 17, 2009, 10:04:32 pm
Quote from: BJNY
P65+ just received the highest score at dxomark.com database
beating ptevious leader Nikon D3x.

Interesting! What I see from the graphs is that the D3X still trounces the P65+ with regard to dynamic range. In fact, the only advantages of the P65+ that are apparent in the DXOmark results are better noise, better tonal range and better color sensitivity, at base ISO only and on a small print (8"x12").

At 8x12" print size and above base ISO, the P65+ is no better than the D3X in any of the qualities measured. It's either approximately equal, or in respect of DR significantly worse.

If one compares performance at the pixel level, the D3X appears better or equal in all respects, except for an anomaly at ISO 800. It appears that the extrapolated results for the P65+ at ISO 800 take a big leap in performance with regard to all parameters tested. What's going on here? Anyone?
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on May 17, 2009, 11:07:42 pm
This is all "kindergarten arithmetic". DxO, just like the vast majority of review sites, publish whatever garbage they want to, without any proof.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Carsten W on May 18, 2009, 04:12:42 am
Quote from: Ray
Interesting! What I see from the graphs is that the D3X still trounces the P65+ with regard to dynamic range. In fact, the only advantages of the P65+ that are apparent in the DXOmark results are better noise, better tonal range and better color sensitivity, at base ISO only and on a small print (8"x12").

Minor addition: what we see is that DxO's measurements of DR still have the D3x come out on top. I have not yet heard of someone who has seen results from both who would agree with that, so all this points out is that how they measure it is not how we see it. In other words, the results are useless.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Ray on May 18, 2009, 05:37:55 am
Quote from: carstenw
Minor addition: what we see is that DxO's measurements of DR still have the D3x come out on top. I have not yet heard of someone who has seen results from both who would agree with that, so all this points out is that how they measure it is not how we see it. In other words, the results are useless.

I have not yet seen any comparisons of real-world images, period. If someone is convinced their P45+ or P65+ delivers better DR than the D3X, then please show the comparisons. Hearing from such people is not sufficient. This is not audio. Seeing is believing.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Ray on May 18, 2009, 11:26:42 pm
I would like to add to this issue of the reliability of the DXO tests, that I personally have never found them to be at significant variance with my own tests.

There are two instances where I compared the performance qualities of different cameras for my own benefit, and then later compared my impressions with the results at DXOmark. I found a surprisingly accurate correlation between my results and DXOmark.

All reasonable people would therefore understand my skepticism at claims that the DXOmark results in respect of MFDBs are a nonsense.

For the benefit of those who have an open mind, the two comparisons I made were:

(1) The Canon 5D versus the Nikon D3 at high ISO, in respect of noise and DR.

(2) The Canon 50D versus the Canon 5D in respect of noise and DR after ISO adjustment had been made to equalise DoF and shutter speed. (Ie. 50D at F4 and ISO 100, compared with the 5D at F6.3 and ISO 320).

I therefore take with a pinch of salt any unsubstantiated claims from MFDB owners that the DXOmark results are nonsense. I think such claims are merely bluster. However, I have an open mind. Show me some comparisons, and demonstrate that such comparisons are genuine and have been taken with sound technique and flawless methodology, and I'll reconsider my position.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on May 19, 2009, 12:46:45 am
Quote from: Ray
(2) The Canon 50D versus the Canon 5D in respect of noise and DR after ISO adjustment had been made to equalise DoF and shutter speed. (Ie. 50D at F4 and ISO 100, compared with the 5D at F6.3 and ISO 320)
Ray, do you mind uploading one of the ISO 320 raw files? If you have one heavily overexposed (raw clipping), that would be great, but if not, then any one is ok.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: BernardLanguillier on May 19, 2009, 01:53:46 am
Quote from: carstenw
Minor addition: what we see is that DxO's measurements of DR still have the D3x come out on top. I have not yet heard of someone who has seen results from both who would agree with that, so all this points out is that how they measure it is not how we see it. In other words, the results are useless.

Hum... care to share any factual comparisons? I don't really see how shadows and color transitons could get significantly cleaner than those of the D3x and still retain a photographic nature. This is now just about the tone mapping application used.

(http://farm4.static.flickr.com/3544/3524673519_c3975976a1_o.jpg)

Cheers,
Bernard
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Ray on May 19, 2009, 10:04:27 am
Quote from: Panopeeper
Ray, do you mind uploading one of the ISO 320 raw files? If you have one heavily overexposed (raw clipping), that would be great, but if not, then any one is ok.

No worries! But give me a couple of days. Those test images were transferred to DVD discs some time ago and are stored at a different location, and I don't have my 5D with me at present.

Cheers!
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Amery on May 19, 2009, 04:02:11 pm
This guy did a brief comparison of the P65+, H3DII-50, H3DII-39, CF22, 5DII, D3X, D700, D90, and Fuji S5:

http://www.flickr.com/photos/dos-chin/sets...57614936120567/ (http://www.flickr.com/photos/dos-chin/sets/72157614936120567/)

The subject matter isn't a good indicator of DR performance and there aren't any high ISO shots. However the full size files are available for download and comparison(some jpeg compression to fit flickr file size limitations). Hope this helps.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Ray on May 19, 2009, 10:26:08 pm
Quote from: Amery
The subject matter isn't a good indicator of DR performance and there aren't any high ISO shots. However the full size files are available for download and comparison(some jpeg compression to fit flickr file size limitations). Hope this helps.

Thanks for the link, but it doesn't really help. The DXOmark results indicate that at base ISO (which appears to be actually ISO 44) the P65+ does display better tonality, greater color sensitivity and lower noise than the D3X, as one might expect, but interestingly only at base ISO. One imagines that such differences would be subtle and probably lost with the fairly severe amount of jpeg compression that's been applied to these Flickr images. (The 170mb of the P65+ has been compressed to just 6.65MB, approx a 25:1 compression.) The resolution advantage of the P65+ is not under dispute.

The really surprising aspect of the DXO results is that the DR of the D3X appears to be so much better than that of the P65+. Even when both images are downsampled to an 8"x12" print size, which of course improves both the DR and noise of the P65+ to a greater extent than it does for the the D3X, the D3X still retains a whopping 2 stops DR advantage at the 'real' ISO of 170, compared with the real ISO of 178 for the P65+.

The best-case scenario for the P45+, regarding dynamic range, is found at ISO 44 where DR is only a 2/3rds' stop worse than that of the D3X at ISO 78, at 8"x12" size.

It's these DR results that seem really surprising, considering that the P65+ is a larger sensor.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on May 19, 2009, 10:47:17 pm
Quote from: Ray
Even when both images are downsampled to an 8"x12" print size, which of course improves both the DR and noise of the P65+ to a greater extent than it does for the the D3X
This is plain rubbish. Downsampling does not increase the dynamic range the very least.

Quote
the D3X still retains a whopping 2 stops DR advantage at the 'real' ISO of 170, compared with the real ISO of 178 for the P65+
I belive only that comparison, which I cooked myself (paraphrasing WSC).

Without having seen any image created by P65 Plus, I risk saying, that this "2 stops DR advantage" is the the product of a comparison performed in drunken stupor.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: paratom on May 20, 2009, 03:37:38 am
Quote from: Panopeeper
This is plain rubbish. Downsampling does not increase the dynamic range the very least.


I belive only that comparison, which I cooked myself (paraphrasing WSC).

Without having seen any image created by P65 Plus, I risk saying, that this "2 stops DR advantage" is the the product of a comparison performed in drunken stupor.

I use a D3x and a Sinar54LV back on a Hy6/Rollei system.
I have tried to run comparison several times but in the end it is hard to judge. First-how to convert them and with which paramameters. Second the slightest difference in focus will allready make a difference.
I came to the conclusion that or really judging a system one has to use it for some time to get an overall impression.

My overall impression regarding IQ (at low ISO) of D3x and Sinar54LV is the following:
-Dynamic Range: is very good for me with both cameras - if I would have to bet which one is a little better here I would say the D3x
-now where I feel the Sinar back is better than the D3x is the following:
   - (micro) detail of Skin and comparable textures- here the images from the SinarBack just look smoother and more natural with more micro detail to me; Skin from the D3x looks good but still sometimes a little bit plastic look; Also Color in skin tones looks more natural for my taste from the MF back. Also I find the transition between sharp and unsharp overall better / smoother from the larger sensor (dont know why this is the case).

So for me and my taste overall IQ I get from the Sinar back is better. If I dont need to be fast and if I dont need higher ISO and if I dont mind the weight I prefer to use MF camera.
I also like the Hy6 user interface - very eays acceess to mirror up, exp bracketing and the mirror stays up and the posibility to use self timer AND mirror up at the same time (I think the D3x can not do it and I find this to be a problem).
Both cameras works good- the D3x the universal and flexible speed machine, the Hy6 with the last additional 2% of IQ.




Title: DXOmark ranks DB image quality well below DSLR!
Post by: Ray on May 20, 2009, 10:14:06 am
Quote from: Panopeeper
This is plain rubbish. Downsampling does not increase the dynamic range the very least.

Without having seen any image created by P65 Plus, I risk saying, that this "2 stops DR advantage" is the the product of a comparison performed in drunken stupor.

Gabor,
Downsampling the images to an 8"x12" size at 300ppi, which represents approximately an 8mp image, would appear to create a perception of increased DR, according to DXO.

At full pixel size, the DR of the P65+ sensor at base ISO of 44 is reported as 11.51EV. Downsampled to an 8x12 print size it jumps up to 12.97EV, almost 1.5 stops greater.

The D3X on the other hand, jumps from a DR of 12.84 to 13.65EV, an increase of less than 1 stop.

I would deduce that what's happening here is that the signal-to-noise is increased as a result of the downsampling. The reduction in apparent noise has the effect of increasing apparent DR. This is not difficult to understand, is it? Hasn't it already been established that downsampling an image reduces noise? Downsampling from 60mp to 8mp has to have a noise reduction effect. In fact, according to DXO, it reduces noise by a very significant 8.8dB.



Title: DXOmark ranks DB image quality well below DSLR!
Post by: Jonathan Wienke on May 20, 2009, 10:42:10 am
The increased DR due to downsampling is only valid if you are willing to accept the simultaneous loss of resolution. If I'm going to buy a 60MP MFDB, I really don't care what its DR is at 8MP, I care about the DR at 60MP.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: BJL on May 20, 2009, 11:57:38 am
Quote from: Jonathan Wienke
The increased DR due to downsampling is only valid if you are willing to accept the simultaneous loss of resolution. If I'm going to buy a 60MP MFDB, I really don't care what its DR is at 8MP, I care about the DR at 60MP.
I see two cases: those when the extra resolution is an IQ advantage and those where it is not. (Never mind that all my photography is in the second case when it come to 20MP plus sensors!)

In situations where the extra resolution is an advantage, the higher resolution option is superior at least in that respect. Equal print size comparisons are always fairest, with visible differences in resolution as well as in noise being part of the overall IQ comparison.

In situations where the extra resolution is not an advantage, it is perfectly reasonable for the higher pixel count image with "surplus resolution" to be processed down to the lower but completely acceptable resolution of the lower pixel count image, with NR processing, or downsampling, ... or maybe just printing/displaying on screen at small enough size.

What never makes sense is comparing sensor IQ on the basis of images displayed at different sizes. Doing that can prove that an image is worse than itself, because displaying it at sufficiently large size can make noise and DR limits more visible than in a smaller print. Hence, per pixel measures from images of different pixel counts are in themselves misleading for IQ comparisons. Measuring "noise per pixel", "DR per pixel" and "lines of resolution per linear pixel count" are a bit like measuring the "horsepower per cylinder" when comparing 4-cylinder and 8-cylinder engines.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Ray on May 20, 2009, 12:50:01 pm
Quote from: Jonathan Wienke
The increased DR due to downsampling is only valid if you are willing to accept the simultaneous loss of resolution. If I'm going to buy a 60MP MFDB, I really don't care what its DR is at 8MP, I care about the DR at 60MP.

Nothing to do with accepting the simultaneous loss of resolution. It's not an option that affects DR. 8mp delivers 8mp of resolution whether you accept it or not. At such resolution, the D3X delivers greater DR.

I care about the DR of all my prints whetever their size. I have quite a few prints that are very close to 8x12" in A4 size albums.

If DR is increased by close to a stop when the D3X image is downsampled by a factor of 3, then it might be reasonable to surmise that the D3X's DR would be reduced by a slightly lesser amount when the image is upsampled by 2.5x to the size of the P65+ file. Instead of 12.84EV at native resolution, it might be just 12EV at 60mp, which makes it still 1/2 a stop better than the P65+.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Dustbak on May 20, 2009, 02:55:31 pm
I have severe difficulties grasping the concept of increased DR with downsampling. So, suddenly detail appears in whites or blacks where there wasn't to begin with at a much larger resolution? Is there an explanation for this?

Also the idea of less noise with downsampling though I am willing to accept that sooner. It might be less visible but I am pretty sure the percentage of noise will be the same unless the downsampling is accompanied with noise reduction that tosses away bad pixels/noise first.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: BJL on May 20, 2009, 04:05:49 pm
Quote from: Dustbak
I have severe difficulties grasping the concept of increased DR with downsampling. So, suddenly detail appears in whites or blacks where there wasn't to begin with at a much larger resolution? Is there an explanation for this?
The explanation is that the noise floor level that sets the bottom of the dynamic range is based on the level of random fluctuations of noise, it is not the minimum output level above zero. Merging multiple pixels into one larger one increases the signal in proportion to the number of pixel merged, but the total noise increases by less, as it is a mixture of positive and negative deviations from the "true" signal, so there is typically some cancellation of the noise from the different pixels. So the signal to noise ratio improves.

Standard signal theory says that if the noise at each of the merged pixels is independent, the noise in a super-pixel produced from N pixels is about sqrt(N) times the typical single pixel noise level, while the signal is N times greater, and so the signal to noise ratio and dynamic range increases in proportion to sqrt(N).


But enough theory: there is a very familiar and obvious example in the form of black and white negatives. Examined closely enough under a microscope, the raw data is exposed and unexposed silver halide crystals, each either pure black or pure white. This means extremely very low DR and very strong visible noise at the silver grain "pixel" level. But prints effectively blur many of these "silver grain pixels" into a fine gray scale with a far greater dynamic range.  And famously, prints using the same emulsion in a larger format and so enlarged less have finer tonal gradations, a sign of higher DR and higher S/N ratios in the signal reaching the eye from the print.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on May 20, 2009, 11:51:59 pm
Quote from: BJL
Standard signal theory says that if the noise at each of the merged pixels is independent, the noise in a super-pixel produced from N pixels is about sqrt(N) times the typical single pixel noise level, while the signal is N times greater, and so the signal to noise ratio and dynamic range increases in proportion to sqrt(N)
Noise is not the only governing factor; in fact, this is the lesser issue. Our subject is photography; the dynamic range is not limited alone by noise but by detail reproduction as well. Or do you believe, that the dynamnic range of a camera can be increased by aggressive noise reduction?

However, image reduction destroys details. Therefor the decreased noise level is useless, except on clean, textureless surfaces - but that's, what noise reduction software can solve.

Title: DXOmark ranks DB image quality well below DSLR!
Post by: Ray on May 21, 2009, 01:22:26 am
Quote from: Panopeeper
Noise is not the only governing factor; in fact, this is the lesser issue. Our subject is photography; the dynamic range is not limited alone by noise but by detail reproduction as well. Or do you believe, that the dynamnic range of a camera can be increased by aggressive noise reduction?

However, image reduction destroys details. Therefor the decreased noise level is useless, except on clean, textureless surfaces - but that's, what noise reduction software can solve.

Gabor,
Whilst I agree with BJL's excellent explanation on this issue, I think there's another point that needs to be addressed with regard to resolution, which is also related to your points above. The P65+ is undoubtedly capable of greater resolution than the D3X, so one might tend to think that such greater resolution would apply across the whole 'subject-brightness-range' of the scene.

This is not necessarily true. If the scene being photographed has a high SBR (say 15EV), the P65+ will not deliver better detail than the D3X in the deepest shadows, whatever the print size (assuming equal print size for the comparison, of course). This is what the statement, 'the D3X has better DR than the P65+', actually means in practice.

We must all have observed that image detail is always significantly degraded in the deepest shadows of any scene that actually has deep shadows. It doesn't matter what camera you use, if the scene being photographed has a high SBR, then detail in the deepest shadows will likely be crap, even at base ISO. This is precisely why it's often necessary to bracket exposure and merge to HDR, or, when possible, use fill flash to illuminate the shadows, when the shadows are not too far away.

Another interesting issue that emerges from these DXOmark results is the apparent huge discrepancy between the nominal ISO sensitivities of the P65+ and the real and actual ISO sensitivity, as measured by DXO.

For example, ISO 400 on the P65+ is actually ISO 178. That's a huge discrepancy. Whereas ISO 200 on the D3X is actually ISO 170, a minor discrepancy. Such variance of course complicates the procedures for comparing the DR of these two cameras. Anyone who wishes to do serious comparisons of these two cameras needs to take this issue into consideration. If DXO is right, a big if perhaps, and what could be the subject of another thread. But bear in mind that DXO probably employs highly qualified scientists who know what they are doing. At the same time, they are in the business of marketing their own RAW converter, so their claims should be questioned.

If we assume that DXO is correct with regard to its ISO testing, then anyone comparing the P65+ at ISO 400 with the D3X, should set the D3X at ISO 200. This is not giving an advantage to the D3X. It's simply getting things right. Phase appears to have lied about their ISO settings. Many other manufacturers also appear to lie. We expect some variance and discrepancies. But more than a whole stop of variance??? Perhaps lying is not the right word. I don't want to start a legal issue. Perhaps there's some other technical issue which DXO is not addressing.

One way or another, this issue should be sorted. ISO is an international standard. It's supposed to mean something specific. We're very short on international standards. They should be adhered to as much as possible.

If we compare the DR of the D3X at ISO 170 with the DR of the P65+ at ISO 178 at the pixel level (in accordance with DXO results), we find that the D3X has a full 3 stops advantage. That's huge and very surprising. I really think that some of you busy professionals who might have access to both a P65+ and a D3X should take the time to check this out.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: BJL on May 21, 2009, 11:09:25 am
Quote from: Panopeeper
Noise is not the only governing factor ... the dynamic range is not limited alone by noise but by detail reproduction as well.
You seem to be redefining "dynamic range" to suit your purposes. I was explicitly talking only about dynamic range, not some more general concept of image quality including "detail reproduction". Detail is another dimension of image quality, not at all a part of the meaning of dynamic range.

And if you do wish to include detail in an IQ comparison between two cameras, then it even more makes sense to consider DR comparisons with equalized image detail, which means with downsampling, NR or such applied to the higher pixel count image within the limits of still retaining as much image detail as the lower pixel count camera offers.

By the way, downsampling from 60MP Bayer CFA raw output to say a 24.5MP RGB format will have more image detail than 24.5MP Bayer CFA raw, so it is not so clear how to equalize image detail. Maybe downsampling to half or less the lowest sensor pixel count equalizes detail.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Ray on May 21, 2009, 12:28:15 pm
Quote from: BJL
You seem to be redefining "dynamic range" to suit your purposes. I was explicitly talking only about dynamic range, not some more general concept of image quality including "detail reproduction". Detail is another dimension of image quality, not at all a part of the meaning of dynamic range.

Well, you've lost me here, BJL   . Surely reproduction of detail is the crucial defining element of DR for the practicising photographer. If someone claims that the D3X has a higher DR than the P65+, then surely that means that the D3X, at any print size, will produce more detail in the deepest shadows than the P65+, when the scene photographed has a higher SBR than the DR capabilities of both cameras, eg. 15EV.

If one wishes to make the point that the obviously higher resolution of the P65+, across most of the tonal range of most scenes captured, is preferred to a perhaps slight resolution advantage of the D3X in the deepest shadows, then that is quite understandable. However, the resolution advantage that any high pixel-count camera has across the mid to high tones does not contribute to its DR assessment, does it?
Title: DXOmark ranks DB image quality well below DSLR!
Post by: ThierryH on May 21, 2009, 12:47:18 pm
I am lost since ages, by all of you and related to this subject!

Thierry

Quote from: Ray
Well, you've lost me here, ...
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on May 21, 2009, 01:04:16 pm
Quote from: Ray
If the scene being photographed has a high SBR (say 15EV), the P65+ will not deliver better detail than the D3X in the deepest shadows, whatever the print size (assuming equal print size for the comparison, of course). This is what the statement, 'the D3X has better DR than the P65+', actually means in practice
Your interpretation of English sentences is remarkable. For me, the meaning of one camera will not deliver better detail than the other... is, that the one is not better in that setting, not the other camera is better.

This only to illustrate your attitude towards this question, which I characterise as cooking a test, which has to deliver a given result.

Anyway, I don't see as proven, that the P65+ will not deliver better detail than the D3X in the deepest shadows. As of now, it is a claim, nothing else.

Quote
Another interesting issue that emerges from these DXOmark results is the apparent huge discrepancy between the nominal ISO sensitivities of the P65+ and the real and actual ISO sensitivity, as measured by DXO
...
Anyone who wishes to do serious comparisons of these two cameras needs to take this issue into consideration
You stress the "equalization" so much, that you are prepared to degrade a 65Mpix camera to a 12Mpix (8x12 with the very high printing density of 360ppi). On the other hand, you are insisting to balance the ISO differences - how do you balance the resolution difference? Downsampling is not "balancing" but "obfuscating".

Back to the issue: the P65+ is not a walkaround camera. You would not make shots with that like your #9199 (the greek taverne) with your 5D. It will be used mostly in settings, where the illumination won't limit the exposure seriously. In other words, it will be used at its optimal setting.

Thus measuring the dynamic range in a setting adjusted to the D3X's capability is ridiculous, even plainly amateurish.

I have only a Canon 40D, but the vast majority of my shots were made (not taken) at ISO 200 (or at ISO 100, only a tiny bit better than ISO 200), in order to utilize the maximum dynamic range. I would not waste a P65+ in settings, where it can not be utilized optimally.

Thus your (and DxO's) equalization of ISOs is a worthless excercise, waste of breath and bandwidth.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Ray on May 21, 2009, 01:04:55 pm
Quote from: ThierryH
I am lost since ages, by all of you and related to this subject!

Thierry

     

Don't worry! Be happy!
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on May 21, 2009, 01:20:46 pm
Quote from: BJL
You seem to be redefining "dynamic range" to suit your purposes
You seem to believe, that the current topic is engineering. In photography, both the noise and the detail reproduction are components of the image quality.

Once more, if detail reproduction were to be ignored, then the dynamic range of any camera could be increased to any degree simply by noise reduction in software - is this not obvious? Would this interpretation of dynamic range have any useful meaning?

Quote
downsampling from 60MP Bayer CFA raw output to say a 24.5MP RGB format will have more image detail than 24.5MP Bayer CFA raw, so it is not so clear how to equalize image detail. Maybe downsampling to half or less the lowest sensor pixel count equalizes detail.
Perhaps to the greatest common denominator? Let's downsample all images to one megapixel. P&S cameras are too welcome to participate in the shootout.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Ray on May 21, 2009, 10:56:27 pm
Quote from: Panopeeper
Your interpretation of English sentences is remarkable.

Thank you.

Quote
For me, the meaning of one camera will not deliver better detail than the other... is, that the one is not better in that setting, not the other camera is better.

DXO doesn't measure resolution differences between cameras. These are my sensible deductions from the SNR and DR figures they supply. I'm surprised that you should revert to the simplistic notion, if Camera A delivers better resolution than Camera B, then it must do so in all circumstances. Surely you can see that that is not true? What's the matter with you?

A good example would be comparing the 15mp G10 with the 3mp D30 or the 6mp D60 or even the 8mp 20D. Which camera delivers the highest resolution? The G10 of course. Does it do so in all circumstances? Of course not. Even the D30 with 1/5th the pixel count of the G10 will deliver better and more detailed image quality than the G10 in the deepest shadows of any scene of high SBR. It will deliver better detail in those shadows because it has a higher DR and lower noise. Isn't that obvious?

Quote
Anyway, I don't see as proven, that the P65+ will not deliver better detail than the D3X in the deepest shadows. As of now, it is a claim, nothing else.

It's not even a claim. DXO doesn't adress resolution differences. It's a reasonable deduction based upon a concept of what dynamic range means in practice.

Quote
You stress the "equalization" so much, that you are prepared to degrade a 65Mpix camera to a 12Mpix (8x12 with the very high printing density of 360ppi). On the other hand, you are insisting to balance the ISO differences - how do you balance the resolution difference? Downsampling is not "balancing" but "obfuscating".

It's not me who's downgrading the image to 8x12. This a 'normalisation' procedure that DXO have adopted, presumably because it makes a lot of sense to compare such factors as SNR and DR on equal size prints. If it were up to me, I'd offer a few more normalisation options, such as the popular A3+, A2 and even 24"x32". By the way, the printing density they've chosen is 300ppi, not 360 as you've stated.

Those who've invested in a P65+ will be pleased to know that at 8x12" size their DB outperforms the Canon 10D in all parameters tested, although, with regard to DR it only just beats the 10D by an insignificant margin at ISO 360 (the P65's highest 'real' ISO). The 10D image has presumably been upsampled to reach an 8x12' size at 300ppi.

Quote
Back to the issue: the P65+ is not a walkaround camera. You would not make shots with that like your #9199 (the greek taverne) with your 5D. It will be used mostly in settings, where the illumination won't limit the exposure seriously. In other words, it will be used at its optimal setting.

DXO does not address such factors as usability under various circumstances. Nor have I. However, we know from many user comments on this forum, that some photographers do want to use their MFDBs at higher than base ISO and have often deplored the fact, and complained about it, that performance at high ISO usually leaves a lot to be desired, compared with a Nikon or Canon DSLR.

When a camera's ISO readout tells me it's ISO 400, I expect it to be reasonably close to that figure, and not out by more than a whole stop. This is information that I personally would certainly care about. It looks as though the P65+ is being marketed as a camera with ISO settings up to ISO 800 when in fact its maximum ISO setting is only ISO 400 (or 360 to be precise, according to DXOmark).

Quote
I have only a Canon 40D, but the vast majority of my shots were made (not taken) at ISO 200 (or at ISO 100, only a tiny bit better than ISO 200), in order to utilize the maximum dynamic range. I would not waste a P65+ in settings, where it can not be utilized optimally.

Thus your (and DxO's) equalization of ISOs is a worthless excercise, waste of breath and bandwidth.

You've got a serious problem with your logic there, Gabor. Because you rarely use your 40D above ISO 200, then anyone using a P65+ would also rarely want to use their camera above ISO 200 and therefore mentioning the fact that ISO 200 on the P65+ is really only ISO 89 is a waste of bandwidth!!! Dear me! You're not on medication, are you?

I can think of lots of reasons why a user of a P65+ might want to use a high ISO, especially in landscape photography when an extensive DoF might sometimes be desired in conjunction with a reasonably fast shutter speed to freeze any slight movement of drifting ice, moving foliage in the breeze, or anything else in the scene which is not perfectly static. As you no doubt already know, and certainly should know, in order to get the DoF of a 40D at F8, you need to use the P65+ at an F stop somewhere between F16 and F22, depending on aspect ratio.

Do you like arguing just for the sake of arguing? Or, are you genuinely serious about the points you've just made?
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Dale Allyn on May 22, 2009, 03:29:22 am
This is an hilarious thread (and entertaining). And I mean no disrespect to those posting... but... (and I'm sure this has been said many times before)... one doesn't need Dx0 or on-line measurebating to determine if the P65+ or the D3x or Canon 5DII is best for one's needs. In fact, most of this (interesting) technical crap is so academic as to be at risk of influencing a decision into the wrong direction.

Demoing a P65+ is the obvious course for those considering it (or any other digital back) – not consulting Dx0 or any other similar site. I'll even submit that if one can't see the differences in the files, then there's no need to consider any of the new equipment. As Ray says "don't worry, be happy". Same goes for the D3x and others. Each can be rented or demoed. Charts, graphs, DR, do not determine the image. In fact, often a 10 or 12 stop image might be quite UNappealing in comparison to those with "sexy light" holding only six stops.

I own only the lowly, and much forgotten, P25+ (in my MF kit). I love it. I have compared the files to those of other cameras/backs and I still like it. Some cameras (or backs) best it, others do not. If I could afford it I would consider upgrading to something with higher pixel count, but not without significant consideration. It's the image, not the numbers. For my needs, a full-frame Phase One "P50+" with maximum exposure length of about five minutes, and 7µ pixel dimension, might be perfect (to facilitate making even larger prints). For my good friend (a working pro) the 5DII is a much better choice, another, the 1Ds3, etc.

The images simply do not LOOK the same from these different equipment. What's better or best? That depends on your needs and tastes. For example, for many DR has little importance once they learn what their equipment will record. That comes from... wait for it... ... taking pictures. As discussed, the sensor is part of it; the conversion software is part of it; the operator is part (much) of it.

All meant in fun, folks.  I waited all these many pages before venting.  We all have different needs and interests. I find that many people considering my images don't care about any of the stuff that we consider when we choose our backs, bodies, printers, etc. I always learn something from this type of thread, but perhaps not what some might expect or wish to "teach".  I like the technical and theoretical discussions too (up to a point), but please let's not throw "photography" under the bus for "computing".

I know, I know... I stepped over the line, or at least, off-topic, by dragging photography in to this.

Best,

Dale
Title: DXOmark ranks DB image quality well below DSLR!
Post by: ThierryH on May 22, 2009, 03:42:05 am
Very well said, Dale, it makes all so much sense.

Best regards,
Thierry

Quote from: DFAllyn
This is an hilarious thread (and entertaining). And I mean no disrespect to those posting... but... (and I'm sure this has been said many times before)... one doesn't need Dx0 or on-line measurebating to determine if the P65+ or the D3x or Canon 5DII is best for one's needs. In fact, most of this (interesting) technical crap is so academic as to be at risk of influencing a decision into the wrong direction.

Demoing a P65+ is the obvious course for those considering it (or any other digital back) – not consulting Dx0 or any other similar site. I'll even submit that if one can't see the differences in the files, then there's no need to consider any of the new equipment. As Ray says "don't worry, be happy". Same goes for the D3x and others. Each can be rented or demoed. Charts, graphs, DR, do not determine the image. In fact, often a 10 or 12 stop image might be quite UNappealing in comparison to those with "sexy light" holding only six stops.

I own only the lowly, and much forgotten, P25+ (in my MF kit). I love it. I have compared the files to those of other cameras/backs and I still like it. Some cameras (or backs) best it, others do not. If I could afford it I would consider upgrading to something with higher pixel count, but not without significant consideration. It's the image, not the numbers. For my needs, a full-frame Phase One "P50+" with maximum exposure length of about five minutes, and 7µ pixel dimension, might be perfect (to facilitate making even larger prints). For my good friend (a working pro) the 5DII is a much better choice, another, the 1Ds3, etc.

The images simply do not LOOK the same from these different equipment. What's better or best? That depends on your needs and tastes. For example, for many DR has little importance once they learn what their equipment will record. That comes from... wait for it... ... taking pictures. As discussed, the sensor is part of it; the conversion software is part of it; the operator is part (much) of it.

All meant in fun, folks.  I waited all these many pages before venting.  We all have different needs and interests. I find that many people considering my images don't care about any of the stuff that we consider when we choose our backs, bodies, printers, etc. I always learn something from this type of thread, but perhaps not what some might expect or wish to "teach".  I like the technical and theoretical discussions too (up to a point), but please let's not throw "photography" under the bus for "computing".

I know, I know... I stepped over the line, or at least, off-topic, by dragging photography in to this.

Best,

Dale
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Ray on May 23, 2009, 12:01:53 am
Quote from: DFAllyn
Demoing a P65+ is the obvious course for those considering it (or any other digital back) – not consulting Dx0 or any other similar site.
Dale

Dale,
If you consider that demoing a P65+ is the obvious course of action for those considering it, or any other digital back, is there any reason why such an approach should not apply to all cameras? If you think this is true, then, according to you, we could dispense with all technical tests of cameras and just rely upon word of mouth or simply adopt a 'follow the leader' approach.

When the cost of the equipment is of a minor concern, you don't even need to demo the equipment. Just buy what the leading photographers in your field are using, and you'll be right, or simply buy the most expensive equipment available on the assumption that the higher the price, the better the equipment.

Speaking from personal experience, demoing equipment should only be necessary in the absence of thorough technical reviews. When Nikon caused a big stir with its first FF 35mm DSLR, the D3, which seemed to have better SNR and DR than any Canon DSLR then available, there was initially simply no reliable information that quantified the improvements over the Canon equipment. Most reviewers just compared in-camera jpegs. The Nikon shots clearly had had a lot of in-camera chroma-noise reduction applied, which Canon wasn't doing in-camera. The Nikon shots at ISO 3200 therefore were significantly cleaner than the Canon shots, but not after one applied an appropriate amount of chroma-noise reduction to the Canon images using a program like Noise Ninja. The Canon 5D shots then sometimes looked better than the D3 shots, but on balance about equal. The argument used by some that any additional noise reduction applied in post processing to the Canon images should also be applied to the Nikon images, was a false argument. You cannot apply chroma noise reduction twice and expect a further improvement. The Nikon images simply had the benefit of in-camera chroma noise reduction. The Canon images didn't have that benefit.

If the reviewers had been doing their job properly, they would have reported that for jpeg shooters the D3 had the time-saving feature of in-camera chroma noise reduction which Canon DSLRs lacked, and that in order to get a high-ISO 5D image looking almost as good (if not as good) as a D3 image it was necessary to apply chroma noise reduction in post-processing.

In fact, what we got was wild exaggeration about the D3's high ISO performance. Such exaggeration was made easier by the fact that the D3 boasted a number of very high ISO settings from ISO 6400 to ISO 256,000. The high numbers seemed to have their intended psychological effect and soon the rumour was widespread that the D3 had up to 2 stops better high-ISO performance than any Canon DSLR on the market.

Since I place a high value on high-ISO performance (I like to do street photography without flash) I decided to demo a D3 and compare it with my 5D. It wasn't easy. I was in Bangkok at the time. No-one was hiring out this latest Nikon DSLR; they were in such short supply. The best I could do was struggle through the dense traffic in Bangkok to the main Nikon agent and use their D3 demo model to compare with my 5D, photographing dark corners inside the shop.

The results were surprising. From ISO 3200 to ISO 256,000 (underexposing the 5D to simulate the higher ISOs), the difference in DR and noise was of the order of 1/3rd of a stop. Sometimes a little more. Sometimes a little less. That's a far cry from the 'up-to-two-stops' claims from some who had actually used and demoed the camera but who had not done any thorough comparisons. It's clear to me that simply demoing a camera, giving it a twirl and a spin in the haphazard manner that most photographers would apply, is not necessarily sufficient to get an accurate impression of its performance in relation to other systems.

When I returned later to Brisbane, Australia, I tried to hire a D3 with the Nikkor 14-24 lens, which I was very interested in. No chance. The major camera hire company in Brisbane was still fulfilling back orders for the D3, and the Nikkor 14-24 was never going to be available for hire becase its protruding front element prevented the fitting of any protective filter.

The fact is, hiring the latest equipment to make your own tests is not only expensive and time-consuming, it's sometimes not even possible. This is why DXOmark is so useful. Instead of spending half a day battling through Bangkok traffic, or going to the expenses of having equipment for hire shipped from the other end of the country because it's not available at your location of residence, you can sit comfortably on your computer stool and get accurate information on most aspects of camera performance from competent testers with a scientific background who know what they are doing.

For information on the more personal aspects of ergonomics and handling, one can refer to field reports from users of the equipment, such as our host, Michael Reichmann.

If  DXOmark had been available 18 months ago, I could have saved myself hours of stuffing around. When I examine their graphs carefully, comparing the DR and SNR of the D3 versus the 5D, I find their results at ISO 3200 are almost exactly the same as mine. So close in fact, it's uncanny.

DXOmark results at ISO 3200 place the D3 as having 0.42EV higher DR than the 5D. However, they also show that the 5D at ISO 3200 is slightly more sensitive than the D3 (real ISO 2710 for the 5D compared with real ISO 2566 for the D3). The difference is so slight it's insignificant in practical terms, but it's sufficient to bring that O.42EV difference in DR closer to my O.33EV estimate, which is why I say the closeness of the result is uncanny. All my comparisons were made at equal f stop and equal shutter speed. My 100% crops are available on this forum, somewhere, if the search engine is functioning properly.

There's another interesting result from DXOmark regarding these two camera. Whilst everyone seemed to be trumpeting the high-ISO capabilities of the D3 in its early days, that we now realise were much exaggerated, it seems that the more significant performance advantage of the D3 is at base ISO, at ISO 200. Here, DR is a whole stop better than the 5D, even comparing the D3 at ISO 200 with the 5D at ISO 100, or more precisely, the D3 at ISO 162 with the 5D at ISO 92.

This is something I didn't have time to test during the brief time I had to demo the D3. This advantage is also something that wasn't made explicit in the early reviews. For practicing photographers to test different systems thoroughly and at different ISOs by demoing hired equipment at great expense to themselves is not only a huge job but simply inefficient.

Organisations like DXO who share their results with the public should be congratulated. Those who state that DXO results are a load of codswallop are merely displaying their scientific illiteracy. Perhaps they are proud of such illiteracy. Reading a graph is not that difficult, is it?


Title: DXOmark ranks DB image quality well below DSLR!
Post by: ejmartin on May 23, 2009, 01:11:44 am
Quote from: Dustbak
I have severe difficulties grasping the concept of increased DR with downsampling. So, suddenly detail appears in whites or blacks where there wasn't to begin with at a much larger resolution? Is there an explanation for this?

Also the idea of less noise with downsampling though I am willing to accept that sooner. It might be less visible but I am pretty sure the percentage of noise will be the same unless the downsampling is accompanied with noise reduction that tosses away bad pixels/noise first.

I think people are much too fixated on the notion of resampling images to a common size to discuss noise and dynamic range.  The point is that noise is not a single number, it has a spectrum, it is a function of spatial frequency, just as the MTF of a lens varies with spatial frequency.  Noise that is independent from pixel to pixel generates a noise spectrum that (after averaging over angular orientation) is linearly rising with spatial frequency.  For instance, here are noise spectra of the Canon 40 (red) and 50D (blue):

http://theory.uchicago.edu/~ejm/pix/20d/po...epower-norm.png (http://theory.uchicago.edu/~ejm/pix/20d/posts/tests/Noise/40d-50d_noisepower-norm.png)

The 40D, with less resolution, cannot reach the highest spatial frequencies in line pairs per picture height available to the 50D; consequently its noise spectrum cuts off somewhat earlier.  The pixel level noise is the area under the curve; since the 50D curve extends further, it has more pixel level noise.  However, if we fix a scale by picking a particular spatial frequency that both cameras can reproduce, the noise is the same.  The 50D is not a noisier camera than the 40D.

What does downsampling do?  Proper downsampling simply removes all spatial frequencies above the Nyquist frequency of the target image, and chops off the portion of the noise spectrum above that frequency.  I downsampled the 50D image to the 40D pixel dimensions, and this is what happened to the noise spectrum:

http://theory.uchicago.edu/~ejm/pix/20d/po...epower-norm.png (http://theory.uchicago.edu/~ejm/pix/20d/posts/tests/Noise/40d-50dresamp_noisepower-norm.png)

Red and Blue as before; Orange is the 50D downsampled with PSCS3 bicubic, black is downsampling with ImageMagick's Lanczos resampling.  The downsampling, especially with Lanczos, quite faithfully reproduces the 40D noise spectrum.  Of course, the downsampling also removed any image detail at spatial frequencies above the Nyquist frequency of the 40D.

However, one needn't have done the downsampling; it was enough to know that the noise power spectra have the same slope to know that the finer resolution camera is not noisier than its lower resolution cousin; one doesn't need to resample to compare noise, one simply needs to level the playing field by fixing a reference spatial frequency and comparing the noise there, much as one can compare MTF's of different lenses at a fixed spatial frequency to see how much detail they render.  This is my interpretation of what DxO does in their "print" tab; they are implicitly choosing a spatial frequency by fixing an output size and doing a mathematically ideal resampling to a reference Nyquist frequency associated to that output size at a standard resolution in dpi (which translates to a given Nyquist in lph).  The resampling is superfluous, all that matters is the comparison at a fixed spatial frequency.

Now, dynamic range is tightly correlated to noise; the technically savvy photographer's working definition of DR is the range of illumination levels having an acceptably large S/N in an image capture.  Since noise is a function of spatial frequency, so is DR (BJL explained this in a somewhat more intuitive way).  Again, downsampling does not increase DR; rather it is changing the spatial scale at which DR is being measured.  DR at a fixed scale largely doesn't care about downsampling, just as the noise of the 50D at a fixed spatial frequency below the target Nyquist didn't change when the image was downsampled.  To compare the DR of two cameras without fixing a common scale or spatial frequency at which to do the comparison, is a largely meaningless exercise.  However, the finer resolution camera need not be downsampled to make the comparison, rather one needs to measure the dependence of DR on spatial frequency.



Title: DXOmark ranks DB image quality well below DSLR!
Post by: Ray on May 23, 2009, 02:04:31 am
Quote from: ejmartin
I think people are much too fixated on the notion of resampling images to a common size to discuss noise and dynamic range.

Well, I'd have to strongly disagree on this point, Emil. The rest of your post makes sense (sort of   ) but is too technical for me to dispute with confidence.

Let's start from facts that we can all agree upon. When we view an image or make a print from a file, that image or print has of necessity a specific size which is absolute. It's not possible to view a print or image which does not have a very precise (absolutely precise) size, unless we apply Einstein's Theory of Relativity and discuss the size of the print from the perspective of a spaceman encircling the earth at a specific velocity. We might not be able to measure the size absolutely precisely, and two images that appear to be the same size might vary by a small fraction of a millimetre. But for practical purposes, their sizes may be considered as absolutely identical.

It makes no sense whatsoever, to compare any technical photographic attributes visually (and I emphasise the word visually) on different size prints, and then make an assessment of the performance of the camera. At postage stamp size, all cameras are the same. At a larger size, differences may become apparent.

In the absence of comparisons of equal size prints (or displayed images), it may be possible to extrapolate the data and deduce what the results might be on a different size print. This is what one has to do when comparing the DR and noise of the D3X when the image is interpolated to the P65+ native size.

However, for the viewer and buyer of any photograph, the qualities of the photograph are entrenched in the physical dimensions of the print. That's what counts and that's all that matters.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Dustbak on May 23, 2009, 02:22:07 am
Quote from: ejmartin
I think people are much too fixated on the notion of resampling images to a common size to discuss noise and dynamic range.  The point is that noise is not a single number, it has a spectrum, it is a function of spatial frequency, just as the MTF of a lens varies with spatial frequency.  Noise that is independent from pixel to pixel generates a noise spectrum that (after averaging over angular orientation) is linearly rising with spatial frequency.  For instance, here are noise spectra of the Canon 40 (red) and 50D (blue):

http://theory.uchicago.edu/~ejm/pix/20d/po...epower-norm.png (http://theory.uchicago.edu/~ejm/pix/20d/posts/tests/Noise/40d-50d_noisepower-norm.png)

The 40D, with less resolution, cannot reach the highest spatial frequencies in line pairs per picture height available to the 50D; consequently its noise spectrum cuts off somewhat earlier.  The pixel level noise is the area under the curve; since the 50D curve extends further, it has more pixel level noise.  However, if we fix a scale by picking a particular spatial frequency that both cameras can reproduce, the noise is the same.  The 50D is not a noisier camera than the 40D.

What does downsampling do?  Proper downsampling simply removes all spatial frequencies above the Nyquist frequency of the target image, and chops off the portion of the noise spectrum above that frequency.  I downsampled the 50D image to the 40D pixel dimensions, and this is what happened to the noise spectrum:

http://theory.uchicago.edu/~ejm/pix/20d/po...epower-norm.png (http://theory.uchicago.edu/~ejm/pix/20d/posts/tests/Noise/40d-50dresamp_noisepower-norm.png)

Red and Blue as before; Orange is the 50D downsampled with PSCS3 bicubic, black is downsampling with ImageMagick's Lanczos resampling.  The downsampling, especially with Lanczos, quite faithfully reproduces the 40D noise spectrum.  Of course, the downsampling also removed any image detail at spatial frequencies above the Nyquist frequency of the 40D.

However, one needn't have done the downsampling; it was enough to know that the noise power spectra have the same slope to know that the finer resolution camera is not noisier than its lower resolution cousin; one doesn't need to resample to compare noise, one simply needs to level the playing field by fixing a reference spatial frequency and comparing the noise there, much as one can compare MTF's of different lenses at a fixed spatial frequency to see how much detail they render.  This is my interpretation of what DxO does in their "print" tab; they are implicitly choosing a spatial frequency by fixing an output size and doing a mathematically ideal resampling to a reference Nyquist frequency associated to that output size at a standard resolution in dpi (which translates to a given Nyquist in lph).  The resampling is superfluous, all that matters is the comparison at a fixed spatial frequency.

Now, dynamic range is tightly correlated to noise; the technically savvy photographer's working definition of DR is the range of illumination levels having an acceptably large S/N in an image capture.  Since noise is a function of spatial frequency, so is DR (BJL explained this in a somewhat more intuitive way).  Again, downsampling does not increase DR; rather it is changing the spatial scale at which DR is being measured.  DR at a fixed scale largely doesn't care about downsampling, just as the noise of the 50D at a fixed spatial frequency below the target Nyquist didn't change when the image was downsampled.  To compare the DR of two cameras without fixing a common scale or spatial frequency at which to do the comparison, is a largely meaningless exercise.  However, the finer resolution camera need not be downsampled to make the comparison, rather one needs to measure the dependence of DR on spatial frequency.


Thank you, I believe you have just explained what I intuitively felt should be case. I think
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Dale Allyn on May 23, 2009, 03:13:38 am
Quote from: Ray
Dale,
If you consider that demoing a P65+ is the obvious course of action for those considering it, or any other digital back, is there any reason why such an approach should not apply to all cameras? If you think this is true, then, according to you, we could dispense with all technical tests of cameras and just rely upon word of mouth or simply adopt a 'follow the leader' approach.

[snip]...

Ray:

Our opinions differ.  And, it seems, that our "relationships" with our photographic equipment differ as well. That's certainly OK with me.

I do read reviews, and I find that certain sources of said reviews carry more weight for me than do others. I did not suggest that one not read, listen, or look at many resources (including some on-line fora) before buying a back or camera body, etc. I DID say that surrendering to algorithms was not in the best interest of one's photography. I guess I should have added "IMO" at the end of every sentence (or at least, my post), but I don't see others engaging in this practice, so assumed it was understood.

Just to be clear, my "day" job is science oriented as I spend time studying subjects through Leica stereo-zoom microscopes over custom made dark-field illumination assemblies, or through visible light spectroscopes, via UV-VIS spectrophotometers or Fourier transform infrared spectroscopy, etc. (and some computer coding as a distraction). Big deal – so can any college student. My point is that I love science and technology. The point of my post was meant as a little reality bump, and a reminder that photogrpahy as a presentational art is not simply about algorithms. At least not for some of us who passionately appreciate that art. I think that you like it too, but I also feel that many here also enjoy the "peacock festival" of matching wits over the technologies of the whole thing. That's fine, of course. I was posting as a venting, and to have a bit of fun to lighten things up a bit.

As for "demoing equipment", I have read your posts regarding MF vs FF DSLRs and how you don't buy into the process of shooting samples and deciding what's best based on one's observations (I apologize if I have incorrectly summarized your ideas here), and I must say that I don't buy it. But that's OK, again, our opinions differ.

I stand by my remark that putting too much trust in sites like Dx0 or other similar "tools" might lead one to a purchase one later regrets or at the least, does not understand. In my case, I have shot the tools that are reasonable considerations for my purposes; I can see the differences in each; and I have purchased what will work for me (within my budget). And before you say that I live where demoing is easy, I'll say that I live in a town of 12,000 people in which there is not one useful camera shop.

As for your remark about testing equipment for street photography in BKK, etc.: there are many ways to "demo" equipment. I live in BKK part-time (I maintain an apartment there) and there are good opportunities for many in that part of the world to see equipment, but hiring (renting) is not always easy. That does not stop someone from shooting in the presence of a salesman, etc. I guess our expectations before acquiring equipment differ greatly. I want comfortable ergonomics, intelligent menus, good image quality based on looking at files from personal tests and those of others whom I trust. (Actually, IMO, most of us never come close to pushing the limits of our equipment, but are limited by our own skills, talents and vision.) The information on sites like Dx0 can be useful, but I trust them as much as I trust our 'elected' officials. Well, that's a bit harsh, but I don't afford them a great deal of trust because I have found that most with a dog in this race have an agenda (or at least, priorities which differ from mine) ... but that's just me.

All in all, good fun. We should all read what we want, test what we want, and shoot what we want. I hope your images fill your heart. I'm not interested in "sport arguing", or forum paint ball battles.

Best,

Dale
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Carsten W on May 23, 2009, 04:19:13 am
Quote from: Ray
Organisations like DXO who share their results with the public should be congratulated. Those who state that DXO results are a load of codswallop are merely displaying their scientific illiteracy. Perhaps they are proud of such illiteracy. Reading a graph is not that difficult, is it?

You say many interesting things and make some good points, but whenever you start arguing against others, you have a really strong tendency re-phrase what they say so that it is no longer representative, and then argue against that instead of the real point being made. In this case, no one here is saying that they can't read graphs. There are some real issues with the DxO measurements, but you ignore those.

When you were arguing with Gabor, you kept misrepresenting what he was saying, and then arguing against this modified version. In fact, he was just pointing out that considering only DR and ignoring resolution doesn't give useful information. You can use noise reduction software to increase *apparent* DR, but the loss in ability to reproduce detail won't make it worth the exercise to most people.

The DxO site lists three values and then gives a single summary number. I once set up three equations in three unknowns to figure out what the weights were (of the form Ax + By + Cz = D), and chose the Leica M8, Nikon D3 and Canon 1Ds3 or something like that, and one of the weights came out negative. DxO really needs to be more open about how they put together these numbers. They make ratings and claims but sometimes what they post makes no sense. If they were open about it, people could point out what doesn't make sense and what needs to be improved. As it is, you get a few numbers describing the cameras but you don't know what they mean, really. That isn't useful.

With respect to the D3x versus MFDBs, I don't yet believe that what they are posting is true in any real-world sense. MFDB results still have that clarity and colour that 35mm cameras from Canon and Nikon can't touch. This isn't anywhere in those numbers, yet they come up with other results which contradict what people see with their own eyes. It isn't enough to say "we measured it" to convince people. They have to be open with the process so that people can reproduce their results.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: billthecat on May 23, 2009, 05:28:02 am
I have a 5D2 and a ZD. On DXO the 5D2 rates a tad higher than the ZD on everything at base ISO. But when I check out my photos taken at the same time, the ZD dynamic range feels much greater, especially in the shadows.

It feels like the ZD has many stops more dynamic range than the 5D2 but DXO doesn't show that.

Bill
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Ray on May 23, 2009, 07:46:56 am
Quote from: DFAllyn
As for "demoing equipment", I have read your posts regarding MF vs FF DSLRs and how you don't buy into the process of shooting samples and deciding what's best based on one's observations (I apologize if I have incorrectly summarized your ideas here), and I must say that I don't buy it. But that's OK, again, our opinions differ.

Dale,
You certainly have incorrectly summarised my ideas on demoing equipment. I'll say it again, it should only be necessary in the absence of thorough technical reviews from people who know what they are doing.

Quote
I stand by my remark that putting too much trust in sites like Dx0 or other similar "tools" might lead one to a purchase one later regrets or at the least, does not understand. In my case, I have shot the tools that are reasonable considerations for my purposes; I can see the differences in each; and I have purchased what will work for me (within my budget). And before you say that I live where demoing is easy, I'll say that I live in a town of 12,000 people in which there is not one useful camera shop.

It might indeed. If DXO test results are incorrect because their methodology is flawed, or for whatever reason, then one could be led astray in basing a purchase on their test results. I agree completely.

Now, let me ask you, Dale. Do you have any evidence (other than hearsay) which would suggest that DXOmark results are flawed, or to put it bluntly, simply wrong or wildly inaccurate? One can speculate till the cows come home on what might be the case. However, as a scientist, you should know that only hard facts count, not speculation. I can only repeat that all my recent, careful comparisons between cameras match DXO results very closely. I have no reason to suspect their results are dodgy.


Title: DXOmark ranks DB image quality well below DSLR!
Post by: Ray on May 23, 2009, 07:59:07 am
Quote from: carstenw
There are some real issues with the DxO measurements, but you ignore those.

Carstenw,
What are they? If you don't specify them, it's understandable I would ignore them. I'm not a mind-reader.

Quote
When you were arguing with Gabor, you kept misrepresenting what he was saying, and then arguing against this modified version. In fact, he was just pointing out that considering only DR and ignoring resolution doesn't give useful information. You can use noise reduction software to increase *apparent* DR, but the loss in ability to reproduce detail won't make it worth the exercise to most people.

I've never ignored resolution. Check my posts again. I've stated more than once that it's understood the P65+ delivers higher resolution than the D3X, on balance. I merely make the point that in the deep shadows of a high SBR scene (high subject-brightness-range scene), that the superior resolution of the P65+ breaks down to the point where the D3X has at least equal resolution and possibly superior resolution, at an extreme pixel-peeping level, as a result of it's superior DR. But only in the deepest shadows of course. If this is incorrect, show me the real-world comparisons that refute it. I have an open mind.

Quote
The DxO site lists three values and then gives a single summary number. I once set up three equations in three unknowns to figure out what the weights were (of the form Ax + By + Cz = D), and chose the Leica M8, Nikon D3 and Canon 1Ds3 or something like that, and one of the weights came out negative. DxO really needs to be more open about how they put together these numbers. They make ratings and claims but sometimes what they post makes no sense. If they were open about it, people could point out what doesn't make sense and what needs to be improved. As it is, you get a few numbers describing the cameras but you don't know what they mean, really. That isn't useful.

Is this the source of the confusion? The single number assessments under the 'overview' heading? If you are at all serious you will look at the individual graphs that detail performance at various ISO settings. The single numbers in the overview are weighted. That means there's a subjective element as to the significance of various levels of performance that have been summarised.

For example, it has been established by me, and later confirmed by DXOmark, (that sounds arrogant, but I did show my results first) that the D3 has approximately a 1/3rd to 1/2 a stop DR advantage over the 5D at high ISO. So how come there's such a huge difference in the single number rating for low-light ISO? The D3 is rated at 22.9 and the 5D at 13.68. That's a huge difference. It doesn't seem to corelate with the graphic results under the DR heading. What does DXO mean by the expression 'low-light ISO'? It clearly doesn't mean, 'high-ISO performance'. I've got no idea what subjective elements have influenced these single number ratings. I recommend ignoring them.

(Well, I do have an idea, but no more than an idea. DXO have not produced any extrapolated results for the 5D at ISO 6400 and beyond. The high single-number rating for low-light ISO of the D3 possibly takes into consideration that the D3 has an ISO 6400 setting, an ISO 12,800 and an ISO 25,600 setting, whereas the 5D doesn't have any of those. However, it has perhaps escaped the DXO team that the D3 at ISO 25,600 still has no more than 1/2 a stop greater DR than the 5D underexposed 3 stops at ISO 3200. Perhaps they never tested this. No-one's perfect).



Title: DXOmark ranks DB image quality well below DSLR!
Post by: Ray on May 23, 2009, 08:08:39 am
Quote from: billthecat
I have a 5D2 and a ZD. On DXO the 5D2 rates a tad higher than the ZD on everything at base ISO. But when I check out my photos taken at the same time, the ZD dynamic range feels much greater, especially in the shadows.

It feels like the ZD has many stops more dynamic range than the 5D2 but DXO doesn't show that.

Bill

Now that's definitely something worth exploring. How come you get the 'feeling' that there's more DR in your ZD shots but cannot demonstrate it? Perhaps we're into deep mysticism here.

Without appearing to sound facetious, there must be a rational explanation. I think you should explore this in more detail.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: ejmartin on May 23, 2009, 08:22:52 am
Quote from: billthecat
I have a 5D2 and a ZD. On DXO the 5D2 rates a tad higher than the ZD on everything at base ISO. But when I check out my photos taken at the same time, the ZD dynamic range feels much greater, especially in the shadows.

It feels like the ZD has many stops more dynamic range than the 5D2 but DXO doesn't show that.

Bill

I haven't looked at images from the ZD, but there is one aspect of the 5D2 that doesn't show up in DR, SNR, etc: The 5D2 has substantial pattern noise (as do all Canon DSLR's, especially non-1 series).  There is a recent thread over at FM discussing the issue:

http://www.fredmiranda.com/forum/topic/748830 (http://www.fredmiranda.com/forum/topic/748830)
http://www.fredmiranda.com/forum/topic/748830/15 (http://www.fredmiranda.com/forum/topic/748830/15)

The pattern noise is quite visually apparent, because it is correlated from pixel to pixel along lines, and our visual perception is designed to pick up on lines and edges.  It is noise because the amplitude fluctuates not from pixel to pixel, but line to line.  As a contribution to the overall noise std dev of a pixel, the pattern noise is a small percentage.  But by averaging along a line the pattern rises well above the overall noise since the pattern noise reinforces along the line while the 2d random noise which is the biggest contributor to the std dev averages out.  Our vision automatically performs this integration and so the pattern noise stands out, even though it makes very little contribution to the std dev of noise and therefore won't appear in the sorts of quantitative tests that DxO does, which are based on pixel std dev's.  Passing through a RAW converter, the demosaic algorithms are designed to preserve edges and patterns and so can enhance the pattern noise.  NR is ineffective.

What is sad is that there are some easy ways to deal with the pattern noise (Nikon has made strides here).  Best is to use cleaner amplifier electronics, since the pattern noise comes from the ISO amplifier/ADC circuitry downstream of the sensor.  One could also provide a border of masked off pixels on the sensor border, use those to detect the line noise, and subtract it off.  Canon doesn't seem interested, since they have let this problem fester for years.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Ray on May 23, 2009, 10:19:29 am
Quote from: ejmartin
The 5D2 has substantial pattern noise (as do all Canon DSLR's, especially non-1 series).

5D pattern noise. Ain't that beautiful! I've just created a tapestry with a camera   . For the record, I think that shot might have been taken at ISO half a million or more. It's difficult to be precise   .

[attachment=13925:1628.jpg]
Title: DXOmark ranks DB image quality well below DSLR!
Post by: billthecat on May 23, 2009, 11:14:37 am
My ZD has more DR than my 5D2. It looks rather obvious. I didn't try to test it by looking at numbers and so forth and if you do it takes a lot of effort I assume to do it right. And when you bring up the shadows in the ZD it works much better than the 5D2.

I was taking some difficult shots in a forest with directly sunlight on a face but the rest of the body in shadows. The ZD did quite well with the scene. The ZD feels more like a nice quality motion picture with the 5D2 feeling like a TV show video by comparison. That is sort of how I relate to it.

It is sad that the 5D2 has that pattern noise down bellow. I assume that doesn't show up with DXO.

I almost always use the ZD at ISO 50.

Bill

Title: DXOmark ranks DB image quality well below DSLR!
Post by: Ray on May 23, 2009, 11:23:56 am
Quote from: billthecat
My ZD has more DR than my 5D2. It looks rather obvious. I didn't try to test it by looking at numbers and so forth and if you do it takes a lot of effort I assume to do it right. And when you bring up the shadows in the ZD it works much better than the 5D2.

I was taking some difficult shots in a forest with directly sunlight on a face but the rest of the body in shadows. The ZD did quite well with the scene. The ZD feels more like a nice quality motion picture with the 5D2 feeling like a TV show video by comparison. That is sort of how I relate to it.

It is sad that the 5D2 has that pattern noise down bellow. I assume that doesn't show up with DXO.

I almost always use the ZD at ISO 50.

Bill

Unless you can show us 100% crops of the same scene with the same lighting, the same FoV and the same shutter speed, we can't be sure about the validity of your impressions. You need to expose your methodology for us to make an assessment, otherwise it's all speculation.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Plekto on May 23, 2009, 11:36:23 am
Quote from: DFAllyn
I stand by my remark that putting too much trust in sites like Dx0 or other similar "tools" might lead one to a purchase one later regrets or at the least, does not understand. In my case, I have shot the tools that are reasonable considerations for my purposes; I can see the differences in each; and I have purchased what will work for me (within my budget).

But the thing is that Dx0 is at least giving us SOME scientific data.  The camera makers flatly lie and throw tons of FUD and marketing claims at us.  I don't know of any other site that tests cameras based upon scientific methods.

So it is useful for at least that much, no matter what one may think of it one way or the other.

P.S. I still wish Dx0 would test the Sigma/Foveon sensors.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: billthecat on May 23, 2009, 11:45:53 am
Ray,

It's too much work for me to setup such a comparison. But perhaps I will if I have some free time this 3day weekend. Currently you have a subjective testimonial that Bill says the ZD has more DR than the 5D2 especially in the shadows. Easton the car photographer also said the same thing a while back.

Bill

Quote from: Ray
Unless you can show us 100% crops of the same scene with the same lighting, the same FoV and the same shutter speed, we can't be sure about the validity of your impressions. You need to expose your methodology for us to make an assessment, otherwise it's all speculation.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Carsten W on May 23, 2009, 12:29:29 pm
Quote
There are some real issues with the DxO measurements, but you ignore those.

Quote from: Ray
Carstenw,
What are they? If you don't specify them, it's understandable I would ignore them. I'm not a mind-reader.

I listed some in my post. The single-most damning flaw is that they don't publish their exact process, so that their numbers are only black magic to others.

Quote
I've never ignored resolution. Check my posts again.

I didn't say that you did. I said that Gabor was saying that resolution was important.

Quote
Is this the source of the confusion? The single number assessments under the 'overview' heading? If you are at all serious you will look at the individual graphs that detail performance at various ISO settings. The single numbers in the overview are weighted.

Man, you are one of the worst readers I have ever seen!  What do you think I set up three variables in three unknowns for? I even said so: to find out what the weights are. The weights made no sense, and it throws even more doubt onto their claims when one can't even figure out how they combine their three individual ratings into their one final rating.

I'll go back and figure out the rating weights again, and see if they have fixed anything.

Quote
For example, it has been established by me, and later confirmed by DXOmark, (that sounds arrogant, but I did show my results first) that the D3 has approximately a 1/3rd to 1/2 a stop DR advantage over the 5D at high ISO. So how come there's such a huge difference in the single number rating for low-light ISO? The D3 is rated at 22.9 and the 5D at 13.68. That's a huge difference. It doesn't seem to corelate with the graphic results under the DR heading. What does DXO mean by the expression 'low-light ISO'? It clearly doesn't mean, 'high-ISO performance'. I've got no idea what subjective elements have influenced these single number ratings. I recommend ignoring them.

Right, and so our opinions get closer. There are too many question-marks over their numbers. Some seem to match what we ourselves feel (like your D3 results), and others are way off. Their process is too secretive. I recommend using the DxO site only as another anecdotal data point, until they get more serious about this and publish proper methods. I am all for giving credit to DxO for doing the community a great service, but only once they have given up their secretiveness and list reproducible results.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Carsten W on May 23, 2009, 12:46:09 pm
Quote from: Ray
Unless you can show us 100% crops of the same scene with the same lighting, the same FoV and the same shutter speed, we can't be sure about the validity of your impressions. You need to expose your methodology for us to make an assessment, otherwise it's all speculation.

That is right, but why don't you also think that this is true for DxO?
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on May 23, 2009, 02:50:10 pm
Quote from: billthecat
It's too much work for me to setup such a comparison. But perhaps I will if I have some free time this 3day weekend. Currently you have a subjective testimonial that Bill says the ZD has more DR than the 5D2 especially in the shadows. Easton the car photographer also said the same thing a while back.

It is imaginable, that the ZD had greater DR than the 5D2, though it is improbable. However, Bill said :-) It feels like the ZD has many stops more dynamic range than the 5D2 but DXO doesn't show that , and I dear say flatly, that this is not so.

Pls keep in mind, that the ZD creates only 8 levels for the tonality over the 9th stop of the DR. Theoretically, if the noise is very-very low, one could say there are four levels for the 10th stop, two levels for the 11th stop, and one level for everything above. In other words, if the sensor of the ZD were so good, the ZD would be bit depth challenged, like the 5D classic.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: ejmartin on May 23, 2009, 03:34:16 pm
Quote from: Panopeeper
It is imaginable, that the ZD had greater DR than the 5D2, though it is improbable. However, Bill said :-) It feels like the ZD has many stops more dynamic range than the 5D2 but DXO doesn't show that , and I dear say flatly, that this is not so.

Pls keep in mind, that the ZD creates only 8 levels for the tonality over the 9th stop of the DR. Theoretically, if the noise is very-very low, one could say there are four levels for the 10th stop, two levels for the 11th stop, and one level for everything above. In other words, if the sensor of the ZD were so good, the ZD would be bit depth challenged, like the 5D classic.


I don't think bit depth is the issue.  Here are deep shadows of my 1D3 at ISO 100:

(http://theory.uchicago.edu/~ejm/pix/20d/tests/noise/blend2-100only-gmcc.jpg)

The 1D3 only has about 11.7 stops of DR at the pixel level; 14-bit tonal depth is wasted on it.  In the above image, the bottom left square of the GM chart is about 9 stops down from RAW saturation.  There is pattern noise all over the place, and it keeps the shadows from being clean to an extent that does not show up in standard measures of DR using std dev of pixels, because it is correlated noise.  For another set of examples involving the 5D2, see the FM thread I linked to earlier, or

http://www.diglloyd.com/diglloyd/free/Push...acks/index.html (http://www.diglloyd.com/diglloyd/free/PushingTheBlacks/index.html)

If Canon cleaned up their low ISO amplification a bit, my image could look like this:

(http://theory.uchicago.edu/~ejm/pix/20d/tests/noise/blend2-gmcc.jpg)

This is the same scene with the same illumination, shot at the same Tv/Av, but at ISO 1600.   The shadows could be so much cleaner if Canon was willing to put in the effort.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on May 23, 2009, 09:33:54 pm
Quote from: ejmartin
The 1D3 only has about 11.7 stops of DR at the pixel level; 14-bit tonal depth is wasted on it
On what criterion is this figure based on?
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Ray on May 23, 2009, 10:20:13 pm
Quote from: carstenw
The single-most damning flaw is that they don't publish their exact process, so that their numbers are only black magic to others.


DXO are in the business of marketing their own RAW converter. I imagine that the exact processes they use to get their measurements will be patented. The'tool' they use to get their measurements is called a "DXO Analyzer". The information they gather about a sensor is crucial to the design of their RAW converter. Do camera manufacturers publish full details of their sensor technology? C'mon! Be sensible!  

However, DXO do publish broad descriptions of their methodology, under the heading 'Technologies'. There, for example, they describe the two different ways of measuring ISO sensitivity (in accordance with ISO standard 12232) and tell you which method they use, which is 'sensitivity for full saturation'. For those who are interested in maths, there are a also few formulas there to get excited about.

Quote
I didn't say that you did. I said that Gabor was saying that resolution was important.

Everyone reading this thread knows that resolution is important. I knew that resolution is important when I was a small boy running around in short pants. Why criticise a review for the aspects of camera performance that are not addressed. DXO also do not address price, weight and waterproofing, which are also important factors. I don't need to be told that the P65+ has higher resolution than the D3X. I can work that out for myself. However, I cannot work out for my self that the D3X has a higher DR than the P65+ without going to a lot of trouble and expense in hiring equipment. This is the sort of information that is therefore valuable.

Sensor resolution very closely follows pixel count. I'm not aware of any sensor with a significantly higher pixel count than another which also does not deliver higher resolution than the other. When sensors have a similar pixel count, there will likely be some variation in resolution depending on the strength of the AA filter, or whether the sensor has one or not. However, sensor resolution is also affected by the brand of RAW converter used, and perhaps more significantly, the 'system' resolution of any 'real-world' image, which most practicing photographers would consider more important, will depend very much on the quality of the lens used. There are probably very good reasons why DXO have not addressed resolution.

Quote
Man, you are one of the worst readers I have ever seen!  What do you think I set up three variables in three unknowns for? I even said so: to find out what the weights are. The weights made no sense, and it throws even more doubt onto their claims when one can't even figure out how they combine their three individual ratings into their one final rating.

I'll go back and figure out the rating weights again, and see if they have fixed anything.

It won't necessarily help you if you figure out the rationale behind their weighting, or the algorithm used to determine a single-figure value. The essential point about a weighting process is that there is a subjective opinion at its core. If you discover the method by which they arrive at those single-figure assessments, you will at least be in a position to determine if DXO are consistent in the application of their ratings, but you may disagree completely with the subjective decision behind their approach.

That is why I ignore their 'overview' ratings, and I would recommend you do too. The results that count are under the five headings, ISO Sensitivity, SNR 18%, DR, Tonal Range, Color Sensitivity. DXO provide full definitions of those terms. I can't understand why you are complaining   .

If you think any of their figures (apart from the subjective weightings) are way out, please demonstrate the fact and provide the evidence. That's not too much to ask is it?  
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Plekto on May 23, 2009, 11:32:04 pm
I think it really comes down to the fact that a lot of people were lied to by the DB makers for years(eg - fake ISOs above ~400) and don't want to deal with the results of the bit camera makers' DSLRs.  It's not like Sony is a tiny company, after all.  One would expect the giants to eventually overtake the small DB makers or come very close - and for less cost, too.  If for no other reason than market share and advertising, which allows them to sell 10x as many units.

This happens in every case.  Take the Audi R8.  It's a mainstream auto maker that suddenly has a car that's nipping at the heels of the Ferrari and similar boutique cars.  Well, it's not too surprising if you think about just their R&D budget and number of employees.

Dx0 is good for at least that much in any case, because it clearly shows glitches and lying.  For example:

http://www.dxomark.com/index.php/eng/Image...se-One/P45-Plus (http://www.dxomark.com/index.php/eng/Image-Quality-Database/Phase-One/P45-Plus)
Click on the ISO Sensitivity tab.  Whatever their testing methods, there plainly is something the maker isn't telling us going on.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: ejmartin on May 23, 2009, 11:44:16 pm
Quote from: Panopeeper
On what criterion is this figure based on?

Max engineering DR at ISO 200; 5 ADU read noise out of 16383 max signal.  16383/5~11.7 stops.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: ThierryH on May 23, 2009, 11:46:55 pm
Why has one to read such misleading and wrong information about digital back manufacturers?!

I can at least stand here for Sinar (one of the digital back manufacturer) and affirm that this is absolut bullshit and disrespectful at the least. There might be information which doesn't come accross, there might be information that is not understood, there might be information that is not explained, there might even be information which is not given, because not relevant, but to say that digital back manufacturers are lying and purposely lying to their customers or potential customers is a serious accusation.

Sinar has answered and still is answering and addressing questions or issues without lying to endusers.

Unfortunately one can write whatever one wants under the cover of anonymity, without consequence.

For the record.

Best regards,
Thierry

Quote from: Plekto
... I think it really comes down to the fact that a lot of people were lied to by the DB makers for years(eg - fake ISOs above ~400) ...
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on May 24, 2009, 12:03:41 am
Quote from: Plekto
I think it really comes down to the fact that a lot of people were lied to by the DB makers for years(eg - fake ISOs above ~400) and don't want to deal with the results of the bit camera makers' DSLRs
Keep in eye, that DSLR manufacturers are doing the same, with the same success. Read the diverse forums and see, how many owners believe to shoot with ISO 6400 and above with the 5D2, 12800 and 25600 with the D3, etc., while they are in fact only underexposing.

Quote
I think it really comes down to the fact that a lot of people were lied to by the DB makers for years(eg - fake ISOs above ~400) and don't want to deal with the results of the bit camera makers' DSLRsDx0 is good for at least that much in any case, because it clearly shows glitches and lying.  For example:

http://www.dxomark.com/index.php/eng/Image...se-One/P45-Plus (http://www.dxomark.com/index.php/eng/Image-Quality-Database/Phase-One/P45-Plus)
Click on the ISO Sensitivity tab.  Whatever their testing methods, there plainly is something the maker isn't telling us going on.
I'm afraid DxO is wrong in this case (too). According to the few P45+ files I have, the higher ISOs are *not* fake. I wonder if they saved the cost of renting a camera for test purposes, thinking that the P45 Plus is acting like the P45.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on May 24, 2009, 12:46:42 am
Quote from: ThierryH
There might be information which doesn't come accross, there might be information that is not understood, there might be information that is not explained, there might even be information which is not given, because not relevant, but to say that digital back manufacturers are lying and purposely lying to their customers or potential customers is a serious accusation
Thierry, forget about the term used by Plekto, let's deal only with the factual background.

The eMotion 54 LV is said to have ISO up to 400, according to you in an earlier post. However, the fact is, that the e54 does not have different ISOs at all; ISO 100 and above are plain underexposures, and the so-called ISO is nothing more than metadata telling to the raw processor to boost the intensity. Many customers (I dare say the vast majority of the customers) don't know this. Don't you think that if they knew this, they would be more conservative in their using of ISO 100 and above?
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on May 24, 2009, 01:34:13 am
Quote from: ejmartin
Max engineering DR at ISO 200; 5 ADU read noise out of 16383 max signal.  16383/5~11.7 stops.
How do you calculate it for Nikon cameras?

(Btw, you forgot to substract 1020, but the difference is not much.)
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Ray on May 24, 2009, 02:00:32 am
Quote from: Panopeeper
I'm afraid DxO is wrong in this case (too). According to the few P45+ files I have, the higher ISOs are *not* fake. I wonder if they saved the cost of renting a camera for test purposes, thinking that the P45 Plus is acting like the P45.

Gabor,
How do you reconcile the above statement with the following statement from Emund Ronald about the high ISO performance of his new P45+?

Quote
Although indications are one can go pretty far down and have usable pictures, it would seem that ISO 1600 is better reached by underexposing ISO 100 than by underexposing from ISO 400.

http://luminous-landscape.com/forum/index....c=21763&hl= (http://luminous-landscape.com/forum/index.php?showtopic=21763&hl=)
Title: DXOmark ranks DB image quality well below DSLR!
Post by: ejmartin on May 24, 2009, 02:17:14 am
Quote from: Panopeeper
How do you calculate it for Nikon cameras?

For the read noise, one can either access the masked pixels and look at their std dev, or fit a straight line to the noise variance vs exposure, whose intercept is the read noise variance (take the sqrt to get the std dev).  For the latter method, it is best to take differences of pairs of raw files in order to eliminate systematic errors.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on May 24, 2009, 02:18:25 am
Quote from: Ray
How do you reconcile the above statement with the following statement from Emund Ronald about the high ISO performance of his new P45+?
I don't. However, it seems (as far as I can measure it from the few raw files I have), that the higher ISOs of the P45+ could be fake with the same result, i.e. increasing the ISO does not reduce the noise. Like ISO 3200 with the 5D2: it is not fake but useless, except for JPEG (one could say it is an analog fake, while 6400 is fake analog .
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on May 24, 2009, 02:27:55 am
Quote from: ejmartin
For the read noise, one can either access the masked pixels and look at their std dev
But they don't record any (except for the D300). Did you think I asked specifically the Nikons by chance?

Quote
or fit a straight line to the noise variance vs exposure, whose intercept is the read noise variance (take the sqrt to get the std dev).  For the latter method, it is best to take differences of pairs of raw files in order to eliminate systematic errors.
I find this not reliable enough even with the latter method, particularly with 12bit depth. A small difference in the "read noise" causes a large difference in the engineering DR.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Carsten W on May 24, 2009, 05:59:50 am
Quote from: Ray
DXO are in the business of marketing their own RAW converter. I imagine that the exact processes they use to get their measurements will be patented. The'tool' they use to get their measurements is called a "DXO Analyzer". The information they gather about a sensor is crucial to the design of their RAW converter. Do camera manufacturers publish full details of their sensor technology? C'mon! Be sensible!  

Okay, if they don't want to publish their methods, then they also cannot expect to be used as anything more than another data point. In this case, their values should in no way change the way people do research before buying a new camera, and in that case, what good are the numbers at all?

Quote
Everyone reading this thread knows that resolution is important.

Then why doesn't DxO attack it in some way? This I don't understand. One hypothetical example: Let's say that a 15MP has slightly higher dynamic range than the P65+. By downsampling the P65+ image to 15MP, you can then boost the DR of the image, getting superior results. In other words, just looking at DR in isolation is not so useful. You must look at DR and resolution together, or at least, at the same time, to get the whole picture for your personal needs.

Quote
It won't necessarily help you if you figure out the rationale behind their weighting, or the algorithm used to determine a single-figure value. The essential point about a weighting process is that there is a subjective opinion at its core. If you discover the method by which they arrive at those single-figure assessments, you will at least be in a position to determine if DXO are consistent in the application of their ratings, but you may disagree completely with the subjective decision behind their approach.

Sure, there is subjectivity everywhere, and that is fine, but there is too much black magic on the DxO site to properly separate fact from opinion, which reduces the usefulness of their published numbers. They need to be more open here, even if they don't reveal everything.

Quote
If you think any of their figures (apart from the subjective weightings) are way out, please demonstrate the fact and provide the evidence. That's not too much to ask is it?  

This is a forum, not a panel of scientists. We are all trying to provoke the others into doing all the work  I currently have 6 cameras, but only one is digital, so I cannot compare anything.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: ThierryH on May 24, 2009, 07:56:36 am
Gabor,

we understand each other, when speaking about that.

To add to your remark: yes, the eMotion backs are "pushing", as do other backs from other manufacturers: that is not a secret, and never was ever lied on concerning this. Now, coming to the naming of a pushed file: how would you call it?: +1 for ISO 200?, +2 for ISO 400?, etc ...? This has been explained many times, on different forums, without making a secret out of it.

But I guess that is not really relevant. What is relevant is the final image quality, and in this respect, I would not believe that a photographer would use higeher ISOs less when having the quality he wants.

Anyway, it is a long road between telling a lie and what is described here.

Best regards,
Thierry

Quote from: Panopeeper
Thierry, forget about the term used by Plekto, let's deal only with the factual background.

The eMotion 54 LV is said to have ISO up to 400, according to you in an earlier post. However, the fact is, that the e54 does not have different ISOs at all; ISO 100 and above are plain underexposures, and the so-called ISO is nothing more than metadata telling to the raw processor to boost the intensity. Many customers (I dare say the vast majority of the customers) don't know this. Don't you think that if they knew this, they would be more conservative in their using of ISO 100 and above?
Title: DXOmark ranks DB image quality well below DSLR!
Post by: ejmartin on May 24, 2009, 08:33:44 am
Quote from: Panopeeper
But they don't record any (except for the D300). Did you think I asked specifically the Nikons by chance?

Are you sure about this?  Bill Claff's Nefutil program
http://home.comcast.net/~NikonD70/Download...til/NefUtil.htm (http://home.comcast.net/~NikonD70/Downloads/NefUtil/NefUtil.htm)
uses the masked pixels to evaluate read noise.  He reports results from many Nikons at
http://home.comcast.net/~NikonD70/Investig...acteristics.htm (http://home.comcast.net/~NikonD70/Investigations/Sensor_Characteristics.htm)
in the column labelled "NP" (for Nikon Proprietary).

Quote
I find this not reliable enough even with the latter method, particularly with 12bit depth. A small difference in the "read noise" causes a large difference in the engineering DR.

You need images taken specifically for this purpose, with a sufficient amount of data near black (but not clipped) in order for the method to be accurate.  Analyzing web-posted RAW files is not up to the task.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on May 24, 2009, 11:58:17 am
Quote from: ejmartin
Are you sure about this?  Bill Claff's Nefutil program
http://home.comcast.net/~NikonD70/Download...til/NefUtil.htm (http://home.comcast.net/~NikonD70/Downloads/NefUtil/NefUtil.htm)
uses the masked pixels to evaluate read noise
How do you think Rawnalyze could display an image without my knowing about the masked pixels?

D40, D50, D70, D3X: none.

The D60 and the D80 have four rows and two columns of masked pixels; they are useless for measuring noise (the standard deviation varies between the channels for example from 1.23 to 1.95 in the same image - this in the numerical range 0-4095).

The D100 has two dirty columns; the D200 has four, the D2X eight.

The D3 has two rows, but the standard deviations between the channels are 70% apart.

If Claff uses those pixels for measurement, he can save the bandwidth.

Only the D300's 32 columns are suitable for this purpose.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Plekto on May 24, 2009, 01:40:01 pm
Quote from: Panopeeper
Keep in eye, that DSLR manufacturers are doing the same, with the same success. Read the diverse forums and see, how many owners believe to shoot with ISO 6400 and above with the 5D2, 12800 and 25600 with the D3, etc., while they are in fact only underexposing.

That was also (or at least I thought it was obvious) my point.  They *all* lie and fudge their data.  Some more than others, of course.  But when you see glitches in the data or things that stand out badly, well, you know something is wrong.  What that is, I'm making no judgment on(and not trying to start some quasi-religious war here about DBs vs DSLRs).   A good example of this in a DSLR would be how the A900 has "issues" with NR not really being off when it says it is.  So it does happen to DSLRs as well.  

Oh - and the link I posted was for a specific model that I happened to run across.  Could have been any manufacturer, really.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: paratom on May 25, 2009, 06:01:55 am
Quote from: Panopeeper
Thierry, forget about the term used by Plekto, let's deal only with the factual background.

The eMotion 54 LV is said to have ISO up to 400, according to you in an earlier post. However, the fact is, that the e54 does not have different ISOs at all; ISO 100 and above are plain underexposures, and the so-called ISO is nothing more than metadata telling to the raw processor to boost the intensity. Many customers (I dare say the vast majority of the customers) don't know this. Don't you think that if they knew this, they would be more conservative in their using of ISO 100 and above?

Isnt it more important how the result looks like vs how it is produced (in hardware vs in software)?
For me 50 and 100 ISO look fine, 200 ok, 400 not really great.
I rather use my eyes than technical explanation. I wouldnt call that"lying" if ISO differences are generated in software.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: ejmartin on May 25, 2009, 09:46:10 am
Quote from: Plekto
I think it really comes down to the fact that a lot of people were lied to by the DB makers for years(eg - fake ISOs above ~400) and don't want to deal with the results of the bit camera makers' DSLRs.  It's not like Sony is a tiny company, after all.  One would expect the giants to eventually overtake the small DB makers or come very close - and for less cost, too.  If for no other reason than market share and advertising, which allows them to sell 10x as many units.

This happens in every case.  Take the Audi R8.  It's a mainstream auto maker that suddenly has a car that's nipping at the heels of the Ferrari and similar boutique cars.  Well, it's not too surprising if you think about just their R&D budget and number of employees.

Dx0 is good for at least that much in any case, because it clearly shows glitches and lying.  For example:

http://www.dxomark.com/index.php/eng/Image...se-One/P45-Plus (http://www.dxomark.com/index.php/eng/Image-Quality-Database/Phase-One/P45-Plus)
Click on the ISO Sensitivity tab.  Whatever their testing methods, there plainly is something the maker isn't telling us going on.

I wouldn't call software implementation of ISO "lying".  

The only reason to have hardware ISO gain is if the dynamic range of the electronics downstream of the sensor in the imaging pipeline is smaller than the DR of the sensor itself.  Then it is that electronics that limits the DR of the camera, and one is forced to choose what portion of the sensor DR is digitized and stored in the RAW data.  This is the way it works in DSLR's, especially CMOS, where the sensor has about 14 stops of DR in many current models but the camera only delivers about 11-12.  Then it makes sense to have variable hardware gain ISO.

If on the other hand one is using sufficiently clean electronics that the entire DR of the sensor is digitized with ample bit depth, there is absolutely no reason why one needs to use separate hardware gains for each ISO; the entire DR that the sensor is capable of is already captured, so one is not going to get anything more than that by changing the gain in the hardware.  Higher ISO is underexposing the sensor no matter what, in the sense that the max signal falls far below the well capacity of the photosites.  If the DR of the ADC sufficiently exceeds the DR of the sensor, it is irrelevant whether the amplification of that low signal is done in the analog or the digital domain.  In fact, it is better to do it in the digital domain, since one isn't throwing away highlight information that the sensor captured by amplifying it beyond the range of the ADC, which is what hardware-based higher ISO gain does.

If MFDB manufacturers are using sufficiently high quality components that they don't need to use hardware based ISO gain, good for them; it means that they are getting the most out of their sensor.  It looks like this is the case with the P45+ above ISO 100, based on the link to DxO.  I wish Canon/Nikon would do the same, then they could unlock the 14 stops of DR that their sensors seem to be capable of, and that their amplifier/ADC's can't so far deliver.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Ray on May 26, 2009, 01:18:58 am
Quote from: ejmartin
If on the other hand one is using sufficiently clean electronics that the entire DR of the sensor is digitized with ample bit depth, there is absolutely no reason why one needs to use separate hardware gains for each ISO; the entire DR that the sensor is capable of is already captured, so one is not going to get anything more than that by changing the gain in the hardware.  Higher ISO is underexposing the sensor no matter what, in the sense that the max signal falls far below the well capacity of the photosites.  If the DR of the ADC sufficiently exceeds the DR of the sensor, it is irrelevant whether the amplification of that low signal is done in the analog or the digital domain.  In fact, it is better to do it in the digital domain, since one isn't throwing away highlight information that the sensor captured by amplifying it beyond the range of the ADC, which is what hardware-based higher ISO gain does.

If MFDB manufacturers are using sufficiently high quality components that they don't need to use hardware based ISO gain, good for them; it means that they are getting the most out of their sensor.  It looks like this is the case with the P45+ above ISO 100, based on the link to DxO.  I wish Canon/Nikon would do the same, then they could unlock the 14 stops of DR that their sensors seem to be capable of, and that their amplifier/ADC's can't so far deliver.

That's an excellent point to make, Emil. However, the DXO results would suggest that neither the P45+ nor the P65+ is using sufficiently clean electronics to match the lower loss of DR, as ISO is increased, that is apparent with the D3X. How do you get the impression that this is not the case?

For example, at base ISO, the D3x has a full stop better DR than both the P45+ and P65+ (slightly more than a full stop, but let's not quibble).

At ISO 800, the D3X has a full 2.5 stops greater DR than the P45+. Nikon's analog, pre-A/D boost, is having the desired effect.

Here's the DXOmark link to the graphs I'm looking at.

http://www.dxomark.com/index.php/eng/Image...d3)/Phase%20One (http://www.dxomark.com/index.php/eng/Image-Quality-Database/Compare-cameras/(appareil1)/304%7C0/(appareil2)/287%7C0/(appareil3)/318%7C0/(onglet)/0/(brand)/Phase%20One/(brand2)/Nikon/(brand3)/Phase%20One)

Edit:I forgot to mention for the benefit of those who never get past the single figure "Overview". Click on the 'Dynamic Range' heading.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: ejmartin on May 26, 2009, 10:42:47 am
Quote from: Ray
That's an excellent point to make, Emil. However, the DXO results would suggest that neither the P45+ nor the P65+ is using sufficiently clean electronics to match the lower loss of DR, as ISO is increased, that is apparent with the D3X. How do you get the impression that this is not the case?

For example, at base ISO, the D3x has a full stop better DR than both the P45+ and P65+ (slightly more than a full stop, but let's not quibble).

At ISO 800, the D3X has a full 2.5 stops greater DR than the P45+. Nikon's analog, pre-A/D boost, is having the desired effect.

Here's the DXOmark link to the graphs I'm looking at.

http://www.dxomark.com/index.php/eng/Image...d3)/Phase%20One (http://www.dxomark.com/index.php/eng/Image-Quality-Database/Compare-cameras/(appareil1)/304%7C0/(appareil2)/287%7C0/(appareil3)/318%7C0/(onglet)/0/(brand)/Phase%20One/(brand2)/Nikon/(brand3)/Phase%20One)

Edit:I forgot to mention for the benefit of those who never get past the single figure "Overview". Click on the 'Dynamic Range' heading.

I didn't say their DR was better at base ISO, I simply said that they are getting close to the full DR of the sensor they are using.

The ISO chart indicates that there is no analog gain for the P45+ beyond ISO 100; that is what the flat ISO graph means.  The P65+ ISO rises from 100 to 3200, indicating that it does have hardware gains for each of these ISO's.  So let's concentrate on the P45+.  

DxO had a choice to make in reporting the DR of the P45+: Do they report the actual DR, max signal level over read noise?  This does not change when the "ISO" is set above 100, since the ISO is metadata and the same analog gain as ISO 100 is used.  However, a photographer would typically double the ISO to 200 when the ambient illumination level drops by two, so middle gray is going to be located at half the RAW level for a typical ISO 200 exposure compared to a typical ISO 100 exposure.  The engineering DR at ISO 200 is the same, but only by virtue of there being an extra stop of highlight headroom above middle gray of typical exposures; the amount of "footroom" in shadows below middle gray drops by a stop.  In other words, what changes is the typical placement of middle gray in the available DR.  So DxO could report the engineering DR, but that would be misleading for photographic purposes, so I think they decided in circumstances where the ISO is done in software to simply lower the graphed DR value by one stop for every stop increase in ISO above the highest hardware-based ISO.

Now, the CCD in the P45+ is a 12-bit device.  The DR of the P45+ at base ISO is quite close to 12 stops.  This says that they are getting just about everything out of the sensor that could be got, and therefore there is no reason to have hardware gain.  I'm not sure what bit depth of ADC they are using, though the output files are 16-bit, they may be oversampling the DR to eliminate quantization error, which would become more apparent as the RAW values are multiplied digitally to generate "higher ISO" if they didn't have enough bit depth.

Now, the D3x is another story.  The sensor has more DR than the camera is delivering, and so it makes sense to have hardware based ISO gain.  Here's how it works for my Canon 1D3:

(http://theory.uchicago.edu/~ejm/pix/20d/tests/noise/DRwindow1d3.png)

Here I have plotted the S/N as a function of absolute exposure, for various ISO.  At the upper end of exposure, each stop increase in ISO pushes another stop of sensor data past the range of the ADC, and so a stop of highlights is lost.  But because the amplifier isn't clean enough at low ISO, increasing the ISO gain boosts the signal relative to the amplifier/ADC noise, and so shadow S/N improves with increasing ISO up to about ISO 800, and a bit more at ISO 1600 (I didn't plot the ISO 3200 curve since it lies on top of the ISO 1600 curve, apart from an extra lost stop of highlights).  Thus the DR curve of the 1D3 is relatively flat from ISO 100-400, because although each bump in ISO drops a stop of highlights, the decrease in amplifier/ADC noise relative to signal adds back in almost a full stop at the shadow end.  If the amplifier/ADC noise were less than the noise of the photosite array, then the latter noise would dominate, it would be amplified as much as the signal with increasing ISO; shadow S/N would not improve with increasing ISO, and DR would drop in proportion to ISO when using analog gain.  And then there would be no point to having analog ISO gain; simply have an ADC with sufficient bit depth, and do ISO in software.  You're not going to get less noise than the photosites themselves are generating, so why throw away potentially useful highlight data by having variable ISO gain in hardware?

Note that if the amplifier/ADC were less noisy than the photosite array, the S/N curve would be the envelope of all the above S/N plots -- that is, it would look like the ISO 1600 curve at low exposure, and the ISO 100 curve at high exposure.  The full DR would be about 14 stops.  I did a little analysis and simulation of what the images from such a camera could look like:

http://theory.uchicago.edu/~ejm/pix/20d/te.../noise-p3a.html (http://theory.uchicago.edu/~ejm/pix/20d/tests/noise/noise-p3a.html)
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Ray on May 27, 2009, 06:14:43 am
Quote from: ejmartin
I didn't say their DR was better at base ISO, I simply said that they are getting close to the full DR of the sensor they are using.

Emil,
That's right. You didn't. The reason I mentioned the better DR of the D3X at base ISO is, despite the P45+ and P65+ having better quality components to process the signal from the sensor, and therefore not needing the analog boost prior to A/D conversion which Nikon and Canon employ, the D3X still has that full stop DR advantage at base ISO and a full 2.5 stops DR advantage at ISO 800.

Reading your explanation, I get the impression that it would be possible for Nikon in a future upgrade to the D3X, to use more expensive, more robust, heavier duty and lower noise A/D converters and other components in the processing chain, to produce a DR result which would be 2.5 stops greater than the P45+ at base ISO as well as at ISO 800.  (Although, after the furore regarding the price of the D3X, Nikon might be having second thoughts about an even more expensive upgrdade employing more expensive components. That's unfortunate for Nikon users I guess, but serves them right   )

Can you confirm that this is essentially what you are saying? Because, if you are, that's quite amazing.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: ejmartin on May 27, 2009, 09:48:33 am
Quote from: Ray
Emil,

Reading your explanation, I get the impression that it would be possible for Nikon in a future upgrade to the D3X, to use more expensive, more robust, heavier duty and lower noise A/D converters and other components in the processing chain, to produce a DR result which would be 2.5 stops greater than the P45+ at base ISO as well as at ISO 800.  (Although, after the furore regarding the price of the D3X, Nikon might be having second thoughts about an even more expensive upgrdade employing more expensive components. That's unfortunate for Nikon users I guess, but serves them right   )

Can you confirm that this is essentially what you are saying? Because, if you are, that's quite amazing.


The sensor photosites can't know what amplification of their signal is going to be performed downstream, therefore their contribution to the noise is independent of ISO.  The read noise has to be at least as large as the photosite noise, so an upper bound on the photosite noise is the lowest read noise, which when referred to photoelectron equivalents occurs at high ISO (typically 3-5 electrons, depending on the model).  So if the manufacturer were able to maintain that read noise at low ISO, the ISO 100 DR would be four stops more than the ISO 1600 DR, assuming that the well capacity is not exceeded at RAW saturation at ISO 100 (full well is slightly below RAW saturation on Canons at ISO 100, so one wouldn't quite get four stops, more like 3.8-3.9).  The extra stops come from the highlight end because the hardware ISO gain at ISO 1600 pushes four stops of highlight data, which would be captured at ISO 100, past the saturation point of the ADC.

Instead, low ISO read noise is about 15-25 electrons in current models, losing well over two stops of the DR that the sensor is recording.  The inferred DR of the sensor obtained by dividing the full well at ISO 100 by the read noise at high ISO is what Roger Clark calls "sensor DR" (as opposed to camera DR).  You can find some data on his website clarkvision.com.

In my article on noise that I linked to above, I proposed a scheme for recovering the extra DR.  I since found out that the idea is patented (one of the patent holders is the fellow who invented the active pixel sensor for CMOS).
Title: DXOmark ranks DB image quality well below DSLR!
Post by: BJL on May 27, 2009, 12:10:34 pm
Quote from: Ray
Reading your explanation, I get the impression that it would be possible for Nikon in a future upgrade to the D3X, to use more expensive, more robust, heavier duty and lower noise A/D converters and other components in the processing chain, to produce a DR result which would be 2.5 stops greater than the P45+ at base ISO as well as at ISO 800.

The D3X uses a Sony EXMOR sensor whose output is already digital, with ISO gain amplification and A/D conversion done on-chip at the bottom of each column of pixels. So if Nikon can do anything it is by helping Sony with its sensor designs, or moving to its own designs like that for the D700 and D3, which use off-chip A/D conversion. From what I have seen, reducing amplifier noise is likely to be the most important part of improving dark noise and DR, though the thousands of tiny 12-bit on-chip A/D convertors of EXMOR sensors could also be a significant limit.

(Aside: it seems likely to me that recent Canon CMOS sensors also apply ISO gain amplification as charge gain in the transfer of signal from photosite to column bottom, with the off-chip 14-bit A/D convertors unlikely to be a significant factor in DR. Emil Martin seems to believe that the A/D convertor is the main noise villain in Canon SLR's, but his evidence does does favor that hypotheses over my proposal about on-chip ISO gain amplifier noise. [Edit: Emil talks in this thread about combined amplifier/ADC noise, and I have no disagreement with that; his analysis both here and with the relationship between spatial resolution and DR is the best technical input we have in discussions like this.])

The idea of amp. noise as a major dark noise source can explain the slower degradation of DR and noise levels with increasing ISO speed with modern types of CMOS sensors than with CCD's, due to CMOS sensors applying gain earlier, and so better protecting the weaker signals at higher ISO from some noise sources, like noise during transportation across edge of the sensor.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on May 27, 2009, 07:36:57 pm
Quote from: BJL
The D3X uses a Sony EXMOR sensor whose output is already digital, with ISO gain amplification and A/D conversion done on-chip at the bottom of each column of pixels
Is this the very same sensor, which is used in the Sony A900? If so, why does the A900 create only 12bit raw data, while the D3X creates 12 or 14bit?
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Ray on May 27, 2009, 10:33:18 pm
Quote from: ejmartin
So if the manufacturer were able to maintain that read noise at low ISO, the ISO 100 DR would be four stops more than the ISO 1600 DR, assuming that the well capacity is not exceeded at RAW saturation at ISO 100 (full well is slightly below RAW saturation on Canons at ISO 100, so one wouldn't quite get four stops, more like 3.8-3.9).

Quote from: BJL
Emil Martin seems to believe that the A/D convertor is the main noise villain in Canon SLR's, but his evidence does does favor that hypotheses over my proposal about on-chip ISO gain amplifier noise. [Edit: Emil talks in this thread about combined amplifier/ADC noise, and I have no disagreement with that; his analysis both here and with the relationship between spatial resolution and DR is the best technical input we have in discussions like this.])

The idea of amp. noise as a major dark noise source can explain the slower degradation of DR and noise levels with increasing ISO speed with modern types of CMOS sensors than with CCD's, due to CMOS sensors applying gain earlier, and so better protecting the weaker signals at higher ISO from some noise sources, like noise during transportation across edge of the sensor.


The long an short of it appears to be that, with current Bayer-type CMOS sensors, we can expect only marginal improvements in high-ISO DR in subsequent models, but there's substantial scope for DR improvement at base ISO.

That the D3X has better DR than the P65+, even at 8x12 print size, is remarkable. But also remarkable is that the D3X has almost 2 stops better DR than the 5D2. If we can expect up to an additional 1.25 stops of DR at base ISO from future Nikon DSLRs, Canon will need to provide an additional 3 stops of DR at base ISO to catch up.

I kew there was something holding me back and preventing me from jumping in and buying a 5D2   . There's no doubt that the 5D2 is a significant improvement over the 5D, but there are too many issues which Canon haven't got right, including a less-than-stellar DR at base ISO, a rather limited autobracketing fuction, and a hobbled video capacity with limited manual control. There has to be something better on the horizon.

(Perhaps an additional 1.25 stops of DR for Nikon is unrealistic. If we make it an additional 1/2 a stop, Canon still needs an additional 2.25 stops to catch up.)
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on May 27, 2009, 10:46:05 pm
Quote from: Ray
a hobbled video capacity with limited manual control. There has to be something better on the horizon.
Like a firmware update, to be released on 2009-06-02?
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Ray on May 27, 2009, 10:50:01 pm
Quote from: Panopeeper
Like a firmware update, to be released on 2009-06-02?

That's news to me. What are the details?

I see. Dpreview, as usual, have the details. That's certainly good news. All we need now is an enhanced autobracketing feature that enables more than 3 shots, a greater range than +/- 2 stops, and the facility to autobracket ISO in manual mode. A faster frame rate would be useful too.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on May 27, 2009, 11:03:58 pm
Quote from: Ray
That's news to me
Really? Your complaints were decisive in the development.

Quote
What are the details?

http://luminous-landscape.com/forum/index....showtopic=35059 (http://luminous-landscape.com/forum/index.php?showtopic=35059)
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Ray on May 27, 2009, 11:16:38 pm
Quote from: Panopeeper
Really? Your complaints were decisive in the development.

I certainly hope so   . If the Aussie dollar keeps rising, I might eventually buy a 5D2, if nothing better appears on the market before my next photographic trip. A good video capability will be very useful to capture authentic folk dancing in the villages of Nepal, in the evenings when lighting is not so good.

I'm a bit worried about that rather poor DR at base ISO, though. I might have to take my D700 along as well.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on May 27, 2009, 11:31:13 pm
Quote from: Ray
I'm a bit worried about that rather poor DR at base ISO, though. I might have to take my D700 along as well.
How do you see the DR of your 5D?
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Ray on May 28, 2009, 02:20:04 am
Quote from: Panopeeper
How do you see the DR of your 5D?

Even worse, but I've moved on from the 5D. On a future photographic trip when weight is a consideration, I'll be carrying the 50D and the D700. I'd like to be able to 'get by' with just the 5D2, if I were to buy one. But I'm concerned there would be too many disadvantages compared with the 50D/D700 combination.

First, for long telephoto purposes, the 5D2 is hardly better than the 20D.

Secondly, there's no Canon lens on a par with the Nikkor 14-24.

Thirdly, neither the autobracketing flexibility of the 5D2 nor the frame rate is on a par with those of the D700.

Fourthly, whilst the DR at base ISO is not significantly worse than that of the D700, I'd be a bit concerned about the waterproofing issue with the 5D2 that became apparent on Michael's recent Antarctic trip.

It's a difficult decision with a lot of compromises and trade-offs.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: billthecat on May 28, 2009, 05:15:25 am
I ran a little test between the ZD back and my 5D2. The DR looked similar. The 5D2 noise down deep showed up as expected, but the ZD also got messy when bringing up the shadows. I used Lightroom for processing the RAW files of each camera. I used a hot light and used many exposures, but kept exposures bellow a second as the ZD gets messy quickly on long exposures. ZD ISO 50 and 5D2 ISO 100.

In my personal experience I tend to like the ZD when DR is important. It appears that the ZD doesn't have a DR advantage from this one test. I assume that the ZD has image quality advantages over the 5D2 that has made it feel like it has better DR to me.

Note: DXO rates the DR of the 5D2 a tad higher than the ZD. From photos I have taken I felt like the DR was much higher in the ZD.

Bill
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on May 28, 2009, 10:37:42 am
Quote from: Ray
Fourthly, whilst the DR at base ISO is not significantly worse than that of the D700, I'd be a bit concerned about the waterproofing issue with the 5D2 that became apparent on Michael's recent Antarctic trip.
The DR of the D700 with ISO 200 is 1/2 EV better than that of the 5D2 with ISO 100. If you have to go with ISO 200, the difference is 2/3 EV.

On the other hand, the 5D2 is 1/2 EV better than the 5D (I can compare only ISO 100 until you send me the other 5D shots).

1/2 EV is a lot. On the other hand, the difference between the D700 and the 50D is much greater: about 1 EV; again, the D700 with ISO 200, the 50D with ISO 100. (I don't engage in the speculation about converting pixels in dynamic range.) IMO the combination D700 and 50D is optimal: one for the DR and clean image, the other for the reach.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: BJL on May 28, 2009, 11:53:45 am
Quote from: Panopeeper
Is this the very same sensor, which is used in the Sony A900? If so, why does the A900 create only 12bit raw data, while the D3X creates 12 or 14bit?
The answer seems not to be public information.

One proposed explanation is that the A/D convertors do repeated sampling and averaging to reduce the RMS noise level. This goes with the lower frame rate in 14-bit mode (am I right that the D3X has a far lower frame rate in 14-bit mode?)

Another proposed explanation is that the D3X has a modified sensor which bypasses the on-board A/D convertors in 14-bit mode, using off-board 14-bit ADCs instead. Even if that is done, I would expect the on-chip ISO gain amplification to still be used, as that is likely part of the process of transferring the signal from photosite to column bottom. So if anything, amp. noise would then be more dominant, with the A/D noise presumably being less in 14-bit mode.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on May 28, 2009, 11:01:27 pm
Quote from: BJL
One proposed explanation is that the A/D convertors do repeated sampling and averaging to reduce the RMS noise level. This goes with the lower frame rate in 14-bit mode (am I right that the D3X has a far lower frame rate in 14-bit mode?)
I heard this speculation, already regarding the D3. However, there is a problem with it: analysis of the raw data shows, that the 14bit data is the "natural" version and the 12bit data is digitally derivated from the 14bit. The same is true re the D3X.

I think the question is logical: if Sony is making such a sensor, why don't they make use of it?
Title: DXOmark ranks DB image quality well below DSLR!
Post by: ejmartin on May 29, 2009, 07:27:36 am
Quote from: Panopeeper
I heard this speculation, already regarding the D3. However, there is a problem with it: analysis of the raw data shows, that the 14bit data is the "natural" version and the 12bit data is digitally derivated from the 14bit. The same is true re the D3X.

I think the question is logical: if Sony is making such a sensor, why don't they make use of it?

The D3 is a totally different sensor design, for which it would be natural for 14 to be the native bit depth.  Do you have evidence that this is so for the D300 and D3x?  The RAW files I have for the D300 suggest otherwise; they have no gaps or spikes in the G channel.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: bjanes on May 29, 2009, 08:38:47 am
Quote from: ejmartin
What does downsampling do?  Proper downsampling simply removes all spatial frequencies above the Nyquist frequency of the target image, and chops off the portion of the noise spectrum above that frequency.  I downsampled the 50D image to the 40D pixel dimensions, and this is what happened to the noise spectrum:

http://theory.uchicago.edu/~ejm/pix/20d/po...epower-norm.png (http://theory.uchicago.edu/~ejm/pix/20d/posts/tests/Noise/40d-50dresamp_noisepower-norm.png)

Red and Blue as before; Orange is the 50D downsampled with PSCS3 bicubic, black is downsampling with ImageMagick's Lanczos resampling.  The downsampling, especially with Lanczos, quite faithfully reproduces the 40D noise spectrum.  Of course, the downsampling also removed any image detail at spatial frequencies above the Nyquist frequency of the 40D.

However, one needn't have done the downsampling; it was enough to know that the noise power spectra have the same slope to know that the finer resolution camera is not noisier than its lower resolution cousin; one doesn't need to resample to compare noise, one simply needs to level the playing field by fixing a reference spatial frequency and comparing the noise there, much as one can compare MTF's of different lenses at a fixed spatial frequency to see how much detail they render.  This is my interpretation of what DxO does in their "print" tab; they are implicitly choosing a spatial frequency by fixing an output size and doing a mathematically ideal resampling to a reference Nyquist frequency associated to that output size at a standard resolution in dpi (which translates to a given Nyquist in lph).  The resampling is superfluous, all that matters is the comparison at a fixed spatial frequency.

Now, dynamic range is tightly correlated to noise; the technically savvy photographer's working definition of DR is the range of illumination levels having an acceptably large S/N in an image capture.  Since noise is a function of spatial frequency, so is DR (BJL explained this in a somewhat more intuitive way).  Again, downsampling does not increase DR; rather it is changing the spatial scale at which DR is being measured.  DR at a fixed scale largely doesn't care about downsampling, just as the noise of the 50D at a fixed spatial frequency below the target Nyquist didn't change when the image was downsampled.  To compare the DR of two cameras without fixing a common scale or spatial frequency at which to do the comparison, is a largely meaningless exercise.  However, the finer resolution camera need not be downsampled to make the comparison, rather one needs to measure the dependence of DR on spatial frequency.

Thus far in this thread, I have seen no mention of the Sensor+ (http://www.phaseone.com/Content/p1digitalbacks/Pplusseries/SensorPlus/SensorPlus2.aspx) pixel binning feature of the new Phase One P65 architecture. If one downsamples 4:1 by pixel averaging, the signal to noise improves as the square root of the downsampling ratio, or by 2:1. However, with the Sensor+ technology (which must be done on chip), 4 pixels are combined into one super-pixel which can be read with only one read noise, resulting in an improvement in the S:N of 4:1. Software downsampling would involve averaging of 4 pixels with 4 read noise contributions.

I would think that dynamic range would be improved, but Phase One claims a DR of 12.5 stops for both methodologies, but the ISO is quadrupled in the Sensor+ mode. I would think that the individual pixels would behave similarly whether or not binning is employed and that the ISO required for saturation of the pixels would not change. Does this mean that full well is not reached with Sensor+? How does binning affect the slope of the noise power spectrum?
Title: DXOmark ranks DB image quality well below DSLR!
Post by: ejmartin on May 29, 2009, 10:06:18 am
Quote from: bjanes
Thus far in this thread, I have seen no mention of the Sensor+ (http://www.phaseone.com/Content/p1digitalbacks/Pplusseries/SensorPlus/SensorPlus2.aspx) pixel binning feature of the new Phase One P65 architecture. If one downsamples 4:1 by pixel averaging, the signal to noise improves as the square root of the downsampling ratio, or by 2:1. However, with the Sensor+ technology (which must be done on chip), 4 pixels are combined into one super-pixel which can be read with only one read noise, resulting in an improvement in the S:N of 4:1. Software downsampling would involve averaging of 4 pixels with 4 read noise contributions.

I would think that dynamic range would be improved, but Phase One claims a DR of 12.5 stops for both methodologies, but the ISO is quadrupled in the Sensor+ mode. I would think that the individual pixels would behave similarly whether or not binning is employed and that the ISO required for saturation of the pixels would not change. Does this mean that full well is not reached with Sensor+? How does binning affect the slope of the noise power spectrum?

They may be claiming 12.5 stops; DxO says a bit over 11.5.  I agree that if their binning technology were clean, they should get two stops of DR.  Going by DxO, they seem indeed to be getting about two stops in the DR chart, and that is borne out in the SNR plots where the ISO 400 and Sensor+ ISO 1600 graphs lie almost on top of one another.  Note that the DR improvement is coming in the shadow end, nothing is changing in the photosites, rather it's the readout of the photosites that is changing so full well is unaffected; so I agree that full well will not be achieved with sensor plus.

BTW, binning will help with read noise but has no effect on shot noise; there the expected gain is 2:1 for 2x2 binning, and it doesn't matter whether you do it in hardware or software.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: BJL on May 29, 2009, 11:05:51 am
Quote from: bjanes
... with the Sensor+ technology (which must be done on chip), 4 pixels are combined into one super-pixel which can be read with only one read noise, resulting in an improvement in the S:N of 4:1.
That sounds right as far as the dark noise arising from transportation, amplification, and A/D conversion, but not photosite noise (dark current?). But these are probably the dominant sources of dark noise, so gaining about two stops of DR at equal ISO speed seems reasonable.

Quote from: bjanes
I would think that dynamic range would be improved, but Phase One claims a DR of 12.5 stops for both methodologies, but the ISO is quadrupled in the Sensor+ mode.
Roughly, quadrupling ISO removes two stops of DR because the signal strength is reduced by 1/4 while the dark noise level stays the same, so it makes sense that the effects on DR of 4-1 binning and quadrupling of ISO balance out. There should be a DR gain at equal ISO speed (400 or higher) with and without binning.

Why does Sensor+ not allow use of the lower ISO speeds? It seems that either the signal path after binning is limited in charge/current capacity to what a full well gives without binning, or the amplifier (charge to voltage conversion, ISO gain etc.) is limited to this DR. The blame cannot come after the amplification and charge to voltge conversion, because the amplification could be adjusted to keep the output within what subsequent components can handle, and it is hard to see 14-bit or 16-bit ADCs in backs this expensive imposing this 12.5 stop DR limit.


So once again, I see evidence suggesting that a major DR limitation is that the DR of the amplifiers is less than that of both the photosites or the ADC.
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Panopeeper on May 29, 2009, 11:44:16 am
Quote from: ejmartin
The D3 is a totally different sensor design, for which it would be natural for 14 to be the native bit depth.  Do you have evidence that this is so for the D300 and D3x?  The RAW files I have for the D300 suggest otherwise; they have no gaps or spikes in the G channel.
All the Nikon CMOS behave mostly the same way; this includes some pecularities as well. The D300 is different only in the handling of the green channels. There are no spikes in the 12bit version of a channel, because the numerical range is divided by exactly four.

The green clipping is not 16383, it is not even a constant value; this is so with the D3 and D3X as well. The red and blue are stretched to 16383, thus the gaps in 14bit mode; when these are mapped over the 4096 range, the spikes show up. The D3 and the D3X do show the division in the green channels as well.

Anyway, the red and blue are more suitable for the observation. See the attached fine histograms from a 14bit and a 12bit file. Note, that the 12bit version shows only one of the green channels (the other looks more weird due to the clipping point's pecularities). The following list of pixel numbers with specific values explain the spikes; I calculated the red one, it perfectly matches with the 10-out-of-11 population of the 14bit red values.

Red      Blue
2948    2632+    Level 599
3201    1961
3962+    1749
3184    1852
2958    1791
3727+    2331+
2760    1781
2810    1586
2942    1703
3695+    1577
2729    2116+
2734    1761
3608+    1642
2667    1627
2943    1590
3577+    2152+
2658    1775
2606    1685
2545    1660
3534+    1541
2901    2299+
2645    1760
3543+    1746
2547    1778
2557    1501
3620+    2157+
2723    1760
2572    1716
2651    1684
3592+    1440
2671    2116+
2877    1623
3580    1643
1605-    1557
3726+    1332
3792+    1940
2828    1512    Level 635
Title: DXOmark ranks DB image quality well below DSLR!
Post by: Ray on May 29, 2009, 12:07:52 pm
Quote from: Panopeeper
IMO the combination D700 and 50D is optimal: one for the DR and clean image, the other for the reach.

That's what I think also. I just wish Nikon would hurry up and develop a lens similar to the Canon 24-105/F4 IS, but better. I used that lens a lot on the 5D and I appreciate the IS because so much of my photoraphy is without a tripod. The Nikkor 24-70/2.8 is no doubt a sharp lens, but we're into the age of image stabilisation, aren't we? Are you listening, Nikon?