Pages: 1 [2] 3 4 5   Go Down

Author Topic: My journey into MF digital, starting with a P45+ on a Hasselblad 555ELD  (Read 48014 times)

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto

Hi,

The comparison I made was using Lightroom, also it was at actual pixels. I will revisit the issue when I am back from travel. Will retest with Capture One, too. The Sony I have has the same sensor as the Nikon D600.

Best regards
Erik


Are you absolutely, beyond any possible doubt, certain?

Because P45+ users used to claim that the P45+ had several stops more DR than the best DSLR 4 years ago, the D3x, that itself clearly has better DR than your Sony.

Either you or them must be wrong. ;)

Cheers,
Bernard

« Last Edit: June 22, 2013, 02:14:57 am by ErikKaffehr »
Logged
Erik Kaffehr
 

yaya

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1254
    • http://yayapro.com

I will revisit the issue when I am back from travel. Will retest with Capture One, too.

I vote for more travel and less testing  ;)
Logged
Yair Shahar | Product Manager | Phase One - Cultural Heritage
e: ysh@phaseone.com |

jerome_m

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 670

More travel and less testing may be good, but trying to understand one's camera capabilities is also a must.

About dynamic range, you published the following curves:





I'll use the red channel as an example, because it is not clipped. The histogram goes between +2 and -8 for the P45 and between +1.5 and -8 or -5 (depending how one interprets the artifacts) on the A99. I understand that the artifacts (the fact that the histogram lacks values) appear faster on the A99 because it uses less quantization bits. But why are the histograms so different?
Logged

BernardLanguillier

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 13983
    • http://www.flickr.com/photos/bernardlanguillier/sets/

Hi,

The comparison I made was using Lightroom, also it was at actual pixels. I will revisit the issue when I am back from travel. Will retest with Capture One, too. The Sony I have has the same sensor as the Nikon D600.

Ah yes, I haf forgotten that you had replaced your A900, it makes more sense now.

Cheers,
Bernard

EricWHiss

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2639
    • Rolleiflex USA

The measurement of DR is almost always controversial and the value useful to photographers is not the ISO defined version.  I recommend you try shooting a transmission step wedge and using imatest software to get data more reliable than comparing histograms.
 
Logged
Rolleiflex USA

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto

Hi,

I will do that pretty soon, but I wanted to see it from a more practical side, namely, how smooth deep shadow detail I can extract from an ETTR image. This was a side effect of some other testing I have made

Shooting a wedge under identical conditions will give interesting results. On the other hand that is exactly what DxO-mark, but they use a test target developed for analyzing dynamic range, so I assume I would just confirm their data.

Best regards
Erik


The measurement of DR is almost always controversial and the value useful to photographers is not the ISO defined version.  I recommend you try shooting a transmission step wedge and using imatest software to get data more reliable than comparing histograms.
 
« Last Edit: June 22, 2013, 08:19:52 am by ErikKaffehr »
Logged
Erik Kaffehr
 

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto

Hi,

The Hassy is not with me on the trip, I carry the Sony as I shot some birds and also some video. There are limits to weight...

I plan to have the Hassy on a trip to Dolomites later this year.

Testing is good to find out how to make best use of your stuff. A wasted test shot is much better than a missed opportunity.

Best regards
Erik

I vote for more travel and less testing  ;)
Logged
Erik Kaffehr
 

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto

Hi,

This are my present plans for testing...

I have been asked to shoot some long exposures and post raw images.

I will be on business travel and hope to sneak in some interesting shooting.

I will probably shoot a comparison between 24MP APS-C, 24 MP full frame and 39 MP MFD.

Stuffer wedge shots in a week or so. Shooting the wedge is quite demanding as you need to eliminate all light leaks.

Focusing accuracy with waist level viewfinder vs. Loupe finder.

MLU vs. No MLU

Hopefully, there will be about an even split between test shooting and real images.

Hopefully, I can also look into using C1 as an alternative to Lightroom.

Best regards
Erik
« Last Edit: June 22, 2013, 09:25:41 am by ErikKaffehr »
Logged
Erik Kaffehr
 

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto

Hi,

The reason I show the raw data is to indicate how far we are of ETTR. Note also that I use a logarithmic Y axis. The way I look at this I see the luminosity range of a channel from ETTR exposure down to say 500 pixels. I presume that low contributions are more coming from noise than pixels. What I think the raw images show is that both images are reasonably close to ETTR. Than I look at how much shadow details I can extract, and what amount of noise I get.

Best regards
Erik

More travel and less testing may be good, but trying to understand one's camera capabilities is also a must.

About dynamic range, you published the following curves:





I'll use the red channel as an example, because it is not clipped. The histogram goes between +2 and -8 for the P45 and between +1.5 and -8 or -5 (depending how one interprets the artifacts) on the A99. I understand that the artifacts (the fact that the histogram lacks values) appear faster on the A99 because it uses less quantization bits. But why are the histograms so different?
Logged
Erik Kaffehr
 

jerome_m

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 670

I noticed the logarithmic Y axis. It explains why we have "combs" in the lower part of the signal.

Still, it seems reasonable to think that the actual image data is represented by the envelope of the histogram. That envelope goes to lower values on the P45+ than on the Sony.
Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto

Yeah,

Question is how much is fake and how much is real. What I can see is that I tried to lift some shadow detail, and that detail is smooth in the Alpha 99 image but has a lot of noise in the P45+ image. Now, it could be that the lens on the Sony has more vailing flare, making shadow smoother, or something else. As exposure is pretty much ETTR I would expect that the image having better DR would have cleaner shadows.

Just to make clear, noise would distribute the signal over many values. Say that lowest real signal would correspond to the data number of 16 and we have a readout noise of 16 electrons, that would spread the signal down to perhaps 5. My guess is is that readout noise on the P45 sensor is about 16. Phase One gives a DR of 12 EV in the spec sheet. So their data says that dynamic range goes from +3EV to -9EV as +3EV represents saturation.

The "comb shape" of the histogram is an artifact of the histogram and not the sensor data. The histogram bunches data in 32 wide channels, but shows the channels with fixed with.

Best regards
Erik

I noticed the logarithmic Y axis. It explains why we have "combs" in the lower part of the signal.

Still, it seems reasonable to think that the actual image data is represented by the envelope of the histogram. That envelope goes to lower values on the P45+ than on the Sony.
« Last Edit: June 23, 2013, 12:02:05 am by ErikKaffehr »
Logged
Erik Kaffehr
 

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto

Some updates:

1) Comment on shadow/dark detail being less noisy on Capture One than LR4 added (needs to be investigated

2) Real worlds samples added

3) Format comparison added (P45+, full frame 24MP and 24 MP APS-C). Somewhat perplexing results, needs more checking.

Best regards
Erik
Logged
Erik Kaffehr
 

Doug Peterson

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4210
    • http://www.doug-peterson.com

All these lab tests really hurt my head.

I'd really encourage you to use more tests like the one you did at the top of the Dynamic Range page. But I'd suggest including Capture One - it's not coincidence that the overwhelming majority of the users on this forum and GetDPI who are asked where to process Phase One files say "Capture One".

And don't be constrained to default settings in the software either. Feel free to tweak noise reduction and sharpening on both images to make each one sing as best as you can for your aesthetic - that's the way you'd use the camera in real life, why do any differently in the tests. There is some ostensible academic value in isolating the variables, but there is little practical value in it.

In particular the color tests you've done don't really correlate to creative image making. Such delta-e color accuracy charting is of great value for art reproduction. But the flash and daylite profiles in Capture One aren't meant for art reproduction. They are meant to balance accuracy and pleasantness of color. If you wanted to minimize delta-e believe me you could hit some insanely tight performance requirements (our Department of Cultural Heritage does it all the time as measured by FADGI and METAMORFOZE compliance). But this won't necessarily make your pictures prettier (better than even odds in my experience that it won't). The Colorchecker Passport is a nice consumer-grade profiling system but
1) it's usually brought up in conversation when the default LR profile is pretty awful as it is in this case for the P45+
2) it increases accuracy, but with no human touch to the pursuit of making pleasant color (as is the case when the color gurus at Capture One sit down and tweak profiles for dozens of hours per back to really make the color sing in as broad of situations as is possible)
3) it isn't nearly as accurate as a more professional profiling system (more patches, more constraints on production, more fine tuning steps)

I really think you're limiting yourself using LightRoom. The dark frame data isn't used (important for deep shadow recovery and long exposures), the algorithms aren't as deeply catered for this back, and the overall math is, IMO, behind C1 for image quality.

Same thing with Raw Analyzer. It's not going to use the dark frame data to bring as much life/accuracy to the shadows as C1 will. Academically interesting, but not very relevant to how far into the shadows you can get printable/pleasant shadows.

Still always nice to see a digital back based on a sensor from 2005 take on a dSLR from late 2012 and hold it's own even in third party software.

And all that said, thanks for sharing your results. I know how inherently imperfect, frustrating, and criticism all testing is. Thank you very much for doing the enormous work of doing your tests and sharing the results.
« Last Edit: June 24, 2013, 08:50:02 pm by Doug Peterson »
Logged

BernardLanguillier

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 13983
    • http://www.flickr.com/photos/bernardlanguillier/sets/

Still always nice to see a digital back based on a sensor from 2005 take on a dSLR from late 2012 and hold it's own even in third party software.

Yep, it can be assumed that the top contenders on both sides are a bit better than that, but DR is pretty much an issue of the past, unless you are invested in Canon lenses.  ;)

http://www.dxomark.com/index.php/Cameras/Compare-Camera-Sensors/Compare-cameras-side-by-side/(appareil1)/831%7C0/(brand)/Sony/(appareil2)/792%7C0/(brand2)/Nikon/(appareil3)/746%7C0/(brand3)/Phase%20One

Cheers,
Bernard

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto

Hi,

Capture 1: http://echophoto.dnsalias.net/ekr/Articles/MFDJourney/Samples/Castle2/20130624-CF043185_C1.jpg

LR 4.4: http://echophoto.dnsalias.net/ekr/Articles/MFDJourney/Samples/Castle2/20130624-CF043185.jpg

Raw image: http://echophoto.dnsalias.net/ekr/Articles/MFDJourney/Samples/Castle2/RawImages/20130624-CF043185.iiq

I own both Capture 1 and Lightroom, but Lightroom has been my main tool since 2006. So I do some testing with Capture 1 but I'm not really comfortable with it and workflow matters a lot.

I have actually seen that C1 gives cleaner shadows, and will add some C1 processed images later on, like end of this week.

Personally, I never shoot people, so skin reproduction plays little role for me.

The reason I use Raw Analyser is mainly that it shows me the raw data. With Raw Analyser I can see how near I am saturation on each channel far more reliably than looking at histograms on the back or histograms in LR or C1. The histograms in LR 4 (PV 2012) hide clipping. So if we look at the DR tests the Raw analyser tells me that:

- The P45+ image is near saturation, has some clipped pixels.
- The Sony image could take 0.5 stop more exposure
- The scene illumination range is around 11EV, both P45+ and Sony image have very few pixels below -8EV and both go to +3EV. So exposure is nearly optimal for highlights.

Finally, I am not really doing tests. For me it is about learning my equipment. Each time I get something new I put it trough a lot of tests. It is a good way to learn.

Best regards
Erik



All these lab tests really hurt my head.

I'd really encourage you to use more tests like the one you did at the top of the Dynamic Range page. But I'd suggest including Capture One - it's not coincidence that the overwhelming majority of the users on this forum and GetDPI who are asked where to process Phase One files say "Capture One".

And don't be constrained to default settings in the software either. Feel free to tweak noise reduction and sharpening on both images to make each one sing as best as you can for your aesthetic - that's the way you'd use the camera in real life, why do any differently in the tests. There is some ostensible academic value in isolating the variables, but there is little practical value in it.

In particular the color tests you've done don't really correlate to creative image making. Such delta-e color accuracy charting is of great value for art reproduction. But the flash and daylite profiles in Capture One aren't meant for art reproduction. They are meant to balance accuracy and pleasantness of color. If you wanted to minimize delta-e believe me you could hit some insanely tight performance requirements (our Department of Cultural Heritage does it all the time as measured by FADGI and METAMORFOZE compliance). But this won't necessarily make your pictures prettier (better than even odds in my experience that it won't). The Colorchecker Passport is a nice consumer-grade profiling system but
1) it's usually brought up in conversation when the default LR profile is pretty awful as it is in this case for the P45+
2) it increases accuracy, but with no human touch to the pursuit of making pleasant color (as is the case when the color gurus at Capture One sit down and tweak profiles for dozens of hours per back to really make the color sing in as broad of situations as is possible)
3) it isn't nearly as accurate as a more professional profiling system (more patches, more constraints on production, more fine tuning steps)

I really think you're limiting yourself using LightRoom. The dark frame data isn't used (important for deep shadow recovery and long exposures), the algorithms aren't as deeply catered for this back, and the overall math is, IMO, behind C1 for image quality.

Same thing with Raw Analyzer. It's not going to use the dark frame data to bring as much life/accuracy to the shadows as C1 will. Academically interesting, but not very relevant to how far into the shadows you can get printable/pleasant shadows.

Still always nice to see a digital back based on a sensor from 2005 take on a dSLR from late 2012 and hold it's own even in third party software.

And all that said, thanks for sharing your results. I know how inherently imperfect, frustrating, and criticism all testing is. Thank you very much for doing the enormous work of doing your tests and sharing the results.
« Last Edit: June 25, 2013, 01:20:06 am by ErikKaffehr »
Logged
Erik Kaffehr
 

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto

Hi,

I am not concerned so much about DR, it is seldom a problem for me, less than one promille of my images are DR-limited, I guess.

Doug is quite right about the KAF sensor from 2006 being impressive, the problem is that we have see so little development, see below.

On the other hand, DxO data indicates that my Sony Alpha 99 has about 1EV more DR than the P45+ and about 0.5EV more than the quite recent IQ 180. The data for the IQ180 and the P45+ essentially overlap, but the IQ 180 goes down 29 ISO and gains advantage over the P45+.

Best regards
Erik


Yep, it can be assumed that the top contenders on both sides are a bit better than that, but DR is pretty much an issue of the past, unless you are invested in Canon lenses.  ;)

http://www.dxomark.com/index.php/Cameras/Compare-Camera-Sensors/Compare-cameras-side-by-side/(appareil1)/831%7C0/(brand)/Sony/(appareil2)/792%7C0/(brand2)/Nikon/(appareil3)/746%7C0/(brand3)/Phase%20One

Cheers,
Bernard

Logged
Erik Kaffehr
 

design_freak

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1128

I vote for more travel and less testing  ;)
+1000
Logged
Best regards,
DF

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto

Hi,

Testing and shooting is a great way to learn about your system. Learn handling, find issues and learn to handle pecularities. For instance, moving from FF DSLR to MFDB raises a new challenge in that depth of field is shorter. Testing is a good way to learn living with the reduced depth of field. It is nice to work out one shooting technique and make early mistakes at home than learning by doing on travel and missing opportunities.

Yeasterday, I ruined one of my shots by not stopping down to optimal aperture on of one of my lenses. I wanted a short DoF, and ended up with a useless picture. Lesson learned. Today I shot a similar subject with two lenses (120/4 and 150/4), so next time I will know more about how they perform and can make an educated choice.

Admittedly, I am also quite curious, coming with my background as a engineer and science student.

Best regards
Erik

+1000
Logged
Erik Kaffehr
 

jerome_m

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 670

In particular the color tests you've done don't really correlate to creative image making. Such delta-e color accuracy charting is of great value for art reproduction. But the flash and daylite profiles in Capture One aren't meant for art reproduction. They are meant to balance accuracy and pleasantness of color. If you wanted to minimize delta-e believe me you could hit some insanely tight performance requirements (our Department of Cultural Heritage does it all the time as measured by FADGI and METAMORFOZE compliance). But this won't necessarily make your pictures prettier (better than even odds in my experience that it won't). The Colorchecker Passport is a nice consumer-grade profiling system but
1) it's usually brought up in conversation when the default LR profile is pretty awful as it is in this case for the P45+
2) it increases accuracy, but with no human touch to the pursuit of making pleasant color (as is the case when the color gurus at Capture One sit down and tweak profiles for dozens of hours per back to really make the color sing in as broad of situations as is possible)
3) it isn't nearly as accurate as a more professional profiling system (more patches, more constraints on production, more fine tuning steps)

I am citing this post because it should not go unnoticed.
Logged

Guy Mancuso

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1133
    • http://forum.getdpi.com/forum/index.php

Erik

Couple quick notes the P45plus is a older back with pretty much a outdated sensor comparing it to the new breeds of the Dalsa lets say and even your new Sony and Nikons. So the DR will probably will not be as wide open lets say as the newer sensors thats a given as I tested this a long time ago with the P65 Plus and P40 Plus. It still is a great sensor and it does have that Kodak color and saturation that it is known for compared to the Dalsa which has more DR and a more neutral color palette. Not that one is better than the other but there is a difference and some folks like one over the other for sure. I had the P25, P30 Plus backs and used the P45plus back pretty often. They are pretty much the same Kodak breeds of sensors but than I moved on to the Dalsa via the P40, IQ 160 and IQ 140 which I liked better but that is my preference in the end was the Dalsa. What the P45 plus is good at is very long exposures good ISO noise levels up to maybe ISO 200 but after that shadow noise will come into play with the higher ISO's. So if you use it within its limitations its a great back. Use it outside of them you may have some disappointment. But I encourage you to use C1 as in the past Lightroom sucked on those backs, not sure about today and there newer versions since I dont even have it on my machine anymore. But if I am bias than C1 is my raw processing machine and I use that on every cam it supports or i wont buy a system it does not support. You will get better results that I can tell you as the Phase engineers specifically work C1 for there backs which makes perfect sense.

BTW I was passing by and glad to see you finally jumped in. MF is fun and the results are worth it IMHO. Also go have some fun, testing gets pretty damn boring even for a engineer. Be well
Logged
[url=http://forum.getdpi.com/forum/showt
Pages: 1 [2] 3 4 5   Go Up