Pages: 1 ... 5 6 [7] 8 9   Go Down

Author Topic: EOS 5Ds- why all this negativity?  (Read 57330 times)

AlterEgo

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1995
Re: EOS 5Ds- why all this negativity?
« Reply #120 on: February 15, 2015, 10:03:56 pm »

by 4-5dB, which is a bit more than a stop
1 stop ~= 6 dB
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
Re: EOS 5Ds- why all this negativity?
« Reply #121 on: February 16, 2015, 12:49:48 am »

1 stop ~= 6 dB

Maybe Bart can explain what's going on here. The Nikon D7000 was the first DSLR that was claimed to be ISO-less. By that, it was meant if one underexposes by one or more stops at any particular ISO, in order to use the desired shutter speed at the desired f/stop, then it made no difference to image quality if shooting in RAW mode. One could compensate for underexposure in ACR or Lightroom, raise shadows and blacks and get an image with no more noise than the same exposure used at a higher ISO.

The graphs for the D7000 on DXOMark are very linear and straight. If you check the values for DR and SNR at 18% at each ISO setting, you'll see there is approximately a 3dB fall in SNR that corresponds with each fall in ISO, and a fall of approximately 1 EV of DR for each fall in ISO.

I suspect that the figure of 6dB that is sometimes equated to a change of 1 stop of exposure applies only to the deep shadows. In other words, at very low signal levels a change of 1 EV is equivalent to a change of 6dB in SNR. However, in the mid-tones, or at 18% grey, a change of 3dB in SNR is equivalent to a change of 1 stop of exposure. I think I've got that right.  If not, I'm sure Bart will correct me. ;)
Logged

AlterEgo

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1995
Re: EOS 5Ds- why all this negativity?
« Reply #122 on: February 16, 2015, 01:44:11 am »

Maybe Bart can explain what's going on here. The Nikon D7000 was the first DSLR that was claimed to be ISO-less. By that, it was meant if one underexposes by one or more stops at any particular ISO, in order to use the desired shutter speed at the desired f/stop, then it made no difference to image quality if shooting in RAW mode. One could compensate for underexposure in ACR or Lightroom, raise shadows and blacks and get an image with no more noise than the same exposure used at a higher ISO.

The graphs for the D7000 on DXOMark are very linear and straight. If you check the values for DR and SNR at 18% at each ISO setting, you'll see there is approximately a 3dB fall in SNR that corresponds with each fall in ISO, and a fall of approximately 1 EV of DR for each fall in ISO.

I suspect that the figure of 6dB that is sometimes equated to a change of 1 stop of exposure applies only to the deep shadows. In other words, at very low signal levels a change of 1 EV is equivalent to a change of 6dB in SNR. However, in the mid-tones, or at 18% grey, a change of 3dB in SNR is equivalent to a change of 1 stop of exposure. I think I've got that right.  If not, I'm sure Bart will correct me. ;)

DR is about what is happening in deep shadows, not about midtones

http://www.dxomark.com/Reviews/Noise-characterization/Summary
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
Re: EOS 5Ds- why all this negativity?
« Reply #123 on: February 16, 2015, 05:16:55 am »

DR is about what is happening in deep shadows, not about midtones

http://www.dxomark.com/Reviews/Noise-characterization/Summary

Yes, of course it is. Thanks for the link to the DXO explanation which confirms what I wrote, that a 6dB change in SNR is equivalent to a 1 EV change (or a one stop change) in DR, whereas a 3dB change in SNR in the midtones , ie. SNR at 18% grey, is equivalent to a one stop change in exposure, approximately.

What perhaps needs to be clarified is that there is a gradual change from that 6dB figure as one moves from the deepest shadows to the highlights. In other words, the deepest shadows might have greater than a 6dB change for a 1 EV change in exposure. The lower mid-tones will have slightly less than a 6dB change, perhaps 5dB or 4dB, and the upper mid-tones slightly less than a 3dB change for each 1 EV change in exposure.

There is also a curious characteristic of Canon cameras with respect to DR at low ISOs, typically between ISO 100 and 400. There's no change in DR for each doubling in ISO.  ;D

If one is comparing the midtones of the 5DS with the midtones of the 7D2 on the basis that the pixels of the two cameras are the same quality, then the midtones of the 5Ds should be slightly more than 3dB better because the sensor gathers 2.5x the amount of light, which is equivalent to more than a 1 EV increase.


http://www.dxomark.com/Reviews/Noise-characterization/Summary


Shadows (): the SNR increases 6dB for every EV and loses 6dB for each doubling of the ISO setting.
Midtones (): the SNR increases 3dB for every EV and decreases by 3dB for each doubling of the ISO setting.
Highlights (): the SNR is constant and does not depend on the ISO.
« Last Edit: February 16, 2015, 05:38:24 am by Ray »
Logged

Bart_van_der_Wolf

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 8913
Re: EOS 5Ds- why all this negativity?
« Reply #124 on: February 16, 2015, 07:34:40 am »

Given that this is all a lot of speculation based on some cursory remarks that the noise of the 5DS would be comparable to that of a 5D Mark III or  a 7D Mark II, this is as much as one can tell. They all have different sensel pitches, so most likely they use (similar but) different technology. The DR (screen) of the mentioned cameras at nominal ISO 100 according to DxOmark, 10.97 and 11.11 respectively, suggests that we may expect something in the order of 11.1 for the 5DS.

Normalization as a function of megapixel size will add an amount of DR (stops) in the order of +Log(Sqrt(N1/N0)) / Log(2) , where N0 is the number of pixels for the reference camera and N1 is the number of pixels of the camera under investigation. Therefore, due to down-sampling potential of the larger MP sensor, the relative gain is 0.59 or 0.67 stops of DR, so a normalized total of 11.6 to 11.8 stops of DR.

IOW, native sensor DR is not something to write home about, but that's not too much of an issue for studio settings where the lighting can be controlled. For stationary subjects one could average several images (4 images will reduce all random noise, in both shadows and highlights, to 50%, i.e. add more than 1 stop of DR). For landscape use, the tried and trusted techniques (ETTR, Bracketing, Exposure fusion/HDR compositing) remain useful.

Having said that, I do expect that these 50.3MP files will have very good post-processing quality, due to the significantly higher system SNR caused by a higher MTF (due to denser sampling of the optical signal) from the given lenses. The high MP count will in many cases allow to down-sample, or require less upsampling, which will improve the received output quality.

Cheers,
Bart
Logged
== If you do what you did, you'll get what you got. ==

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
Re: EOS 5Ds- why all this negativity?
« Reply #125 on: February 16, 2015, 09:34:25 am »

Given that this is all a lot of speculation based on some cursory remarks that the noise of the 5DS would be comparable to that of a 5D Mark III or  a 7D Mark II, this is as much as one can tell. They all have different sensel pitches, so most likely they use (similar but) different technology. The DR (screen) of the mentioned cameras at nominal ISO 100 according to DxOmark, 10.97 and 11.11 respectively, suggests that we may expect something in the order of 11.1 for the 5DS.


Hi Bart,
I'm confused. Are you using a Mac by any chance?  ;D   On my PC, the DXO graphs show the 7D2 as having  a DR of 11.78 EV and the 5D3 a DR of 11.74 at base ISO, essentially the same. The DR at ISO 200 and 400 is also the same, although both cameras are down an insignificant 1/3rd of a stop at ISO 400. Refer attached image.

I see no reason not to expect the 5DS to have one full stop better DR than the 7D2, ie. 12.78 EV at ISO 100 and 12.39 EV at ISO 400, approximately. That's still well behind the Nikon D800E of course, which has a claimed DR of 14.33 EV at the nominal ISO of 100, but not so far behind at ISO 400; maybe just 1/2 a stop behind.

I think it's also reasonable to expect that the DR of the 5DSR will be better than that of the D800E above ISO 1600 because the DR of the 5D3 is already about the same as the DR of the D800E from ISO 3200 onwards, and only 1/4th of a stop behind at ISO 1600.

Cheers!
Logged

Bart_van_der_Wolf

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 8913
Re: EOS 5Ds- why all this negativity?
« Reply #126 on: February 16, 2015, 10:15:10 am »

Hi Bart,
I'm confused. Are you using a Mac by any chance?  ;D   On my PC, the DXO graphs show the 7D2 as having  a DR of 11.78 EV and the 5D3 a DR of 11.74 at base ISO, essentially the same.

Hi Ray,

That is when comparing down-sampled output at some size of print resolution, which says nothing about the sensors themselves. That's why one needs to use the 'screen' settings on the charts, instead of the 'print' settings. The 'print' settings have already been normalized, to an arbitrary size, thus making all subsequent normalizations 'iffy' at best. IMHO, it's better to take the native sensor quality ('Screen' settings) and draw conclusions based on that, optionally with normalization to a common size if one wants to compare apples and oranges.

The 5D Mark II also had a DR of 11.16, so basically the same as the later models, although the sensels got smaller. That suggests that the fabrication technology is relatively constant, despite features getting smaller. The 5DS has marginally larger sensels than the 7DII, so a different design of which we yet have to see solid test data.

Cheers,
Bart
« Last Edit: February 16, 2015, 10:19:17 am by BartvanderWolf »
Logged
== If you do what you did, you'll get what you got. ==

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
Total sensor SNR ("DR") vs local SNR in the mid-tones
« Reply #127 on: February 16, 2015, 02:36:21 pm »

Maybe Bart can explain what's going on here. ... The graphs for the D7000 on DXOMark are very linear and straight. If you check the values for DR and SNR at 18% at each ISO setting, you'll see there is approximately a 3dB fall in SNR that corresponds with each fall in ISO, and a fall of approximately 1 EV of DR for each fall in ISO.

I suspect that the figure of 6dB that is sometimes equated to a change of 1 stop of exposure applies only to the deep shadows. In other words, at very low signal levels a change of 1 EV is equivalent to a change of 6dB in SNR.

I'm no Bart, but we must distinguish two things here:
a) "DR" in the sense of total SNR; the rest of maximum possible signal at a given Exposure Index ["ISO"] setting (limited by highlight clipping) down to the noise floor
b) SNR at a given level of signal, like the level that gives 18% of maximum level, which is then the ratio of that signal to the noise level in parts of the image at that signal level.

Putting aside the fact that I think sense (a) is a bit silly, since it is mostly measuring amplifier clipping of highlights due to excessive amplification, and could be avoided in a truly "ISO-less" sensor by applying that same analog amplification at all EI settings and then adjusting in the digital domain, here goes:
In sense (a), each doubling of ISO setting corresponds to a halving of the maximum photosite output level before clipping, while leaving the noise floor unchanged, so reducing the SNR by a fact of 2, or 6dB.

For the 18% tone level comparison (b) doubling the IS again halving the signal level (photoelectron count) corresponding to that 18% level, but at all but very high EI settings, the noise at 18% level is dominated by shot noise, not sensor dark noise, and halving the signal also reduces this shot noise by a factor of sqrt(2), so that the SNR, being roughly "mid-tone signal to mid-tone shot noise" is reduced by only a factor of sqrt(2), which on the logarithmic dB scale is 3dB instead of 6dB.

By the way, I would expect that as you push to extremely high EI settings, the increase in 18% SNR for each doubling of EI gradually increases from 3dB towards 6dB, a sign that sensor dark noise has become significant even in the "18% mid-tone" region.  Is that what you see?
Logged

nemophoto

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1021
    • Nemo Niemann Photography
Re: EOS 5Ds- why all this negativity?
« Reply #128 on: February 16, 2015, 02:38:29 pm »


But let's wait for the 5DS files (which is the real topic of this thread), rather than one from a camera from 2007 (that's still doing fine with the proper amount of TLC).

Cheers,
Bart

Interesting you say this Bart. With ever better RAW converters, I've been able to massage some very nice images out of elderly digital files from the original 1D and 1Ds.

And that said, I think the 5Ds will be just fine. How can people bitch about something they've never used or even seen? For my money, though the camera body is not what I would have wished (always being a 1-series guy), as long as the camera delivers the same quality and dynamic range of the 5D3, I will be happy. It remains to be seen what Sony can actually come up with if they have a res that high. It basically comes down to the physics of the photosites and the software that interprets the signal. Sure, Sony MAY be able to come up with something better, but I personally HATE the direction of their cameras. Call me old fashion -- I like an optical viewfinder 99% of the time. So could they have a better sensor at that res? Possibly. Would I care? Not really.

Nemo
Logged

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
Re: EOS 5Ds- why all this negativity?
« Reply #129 on: February 16, 2015, 02:49:06 pm »

The 5D Mark II also had a DR of 11.16, so basically the same as the later models, although the sensels got smaller.
One possible explanation is that Canon photo-sites have a total SNR (full well capacity to dark noise level) of well over 11.2 stops, but the Canon sensor design with off-chip ADC and its long path of:
- signal transfer from photo-site to sensor edge (probably with variable charge amplification at this stage in recent Canon sensors)
- change or voltage transport along the sensor edge and off the chip, through possible further amplification and then to the ADC
imposes that DR limit -- so changes in "photosite DR" do not effect the final DR.


The evidence suggesting this is that, from what I have read:

- Canon's DSLR photosites do not have a substantially smaller full well capacity that Sony's (base ISOs-speeds are not substantially higher, etc.)

- Canon's DSLR photosites do not have a substantially higher dark noise floor that Sony's, as shown by their good low-light (high exposure index) performance.

In other words, I think that Canon's photo-sites themselves are fine; the disadvantage of current Canon designs is extra read noise at low EI settings (when less of the early on-sensor charge amplification is applied) arising during to the long, old-fashioned analog signal path from photo-site to ADC, compared to the modern approach of column-parallel ADC's.
« Last Edit: February 16, 2015, 03:21:01 pm by BJL »
Logged

AlterEgo

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1995
Re: EOS 5Ds- why all this negativity?
« Reply #130 on: February 16, 2015, 03:05:25 pm »

Sure, Sony MAY be able to come up with something better, but I personally HATE the direction of their cameras. Call me old fashion -- I like an optical viewfinder 99% of the time. So could they have a better sensor at that res? Possibly. Would I care? Not really.

with Sony sensors you have 2 more alternatives (at least) with OVF  = Nikon and now finally Ricoh/Pentax... now lenses for those 2 mounts are (might be for you) a different story
Logged

Bart_van_der_Wolf

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 8913
Re: EOS 5Ds- why all this negativity?
« Reply #131 on: February 16, 2015, 03:19:23 pm »

In other words, I think that Canon's photo-sites themselves are fine; the designs's disadvantage is extra read noise at low EI settings (when less of the early on-sensor charge amplification is applied) due to the long, old-fashioned long path from photo-site to ADC compared to the modern approach of column-parallel ADC's.

That could indeed be the explanation.

Cheers,
Bart
Logged
== If you do what you did, you'll get what you got. ==

dwswager

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1375
Re: EOS 5Ds- why all this negativity?
« Reply #132 on: February 16, 2015, 05:18:27 pm »

50MPs is basically the selling point for this Camera.  But both Sony and Nikon are likely to match it in pretty short order.

So considering there is not a significant DR upgrade in the offering to match the Sony sensor, is Canon going to throw a bone or 2 to make their camera different?  I find it baffling that RAW histograms or a ETTR metering mode isn't included for the landscape guys.  Without some differentiation, if Nikon releases a 50MP D910 that has 3 more stops DR and less shadow noise, then only for those people locked into Canon does the 5Ds make any sense. 

My point is, already knowing you are going to loose the DR/Shadow noise contest to your competitors for the entire development cycle of this camera, would it not make sense to offer something else to 1) keep people in the fold and 2) attract new customers.  Increased color fidelity, it it does actually test out certainly is nothing to sneeze at I guess.
Logged

shadowblade

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2839
Re: EOS 5Ds- why all this negativity?
« Reply #133 on: February 16, 2015, 05:59:48 pm »

50MPs is basically the selling point for this Camera.  But both Sony and Nikon are likely to match it in pretty short order.

So considering there is not a significant DR upgrade in the offering to match the Sony sensor, is Canon going to throw a bone or 2 to make their camera different?  I find it baffling that RAW histograms or a ETTR metering mode isn't included for the landscape guys.  Without some differentiation, if Nikon releases a 50MP D910 that has 3 more stops DR and less shadow noise, then only for those people locked into Canon does the 5Ds make any sense. 

My point is, already knowing you are going to loose the DR/Shadow noise contest to your competitors for the entire development cycle of this camera, would it not make sense to offer something else to 1) keep people in the fold and 2) attract new customers.  Increased color fidelity, it it does actually test out certainly is nothing to sneeze at I guess.

As a landscape photographer, I'd just get the Sony A7r or A9 and put Canon lenses onto that.

Without any improvement in DR, the only reason to get the 5Ds is as a high-pixel-density camera for wildlife photography, since it has a high-performance AF system and enough resolution to function as a full-frame and crop sensor at the same time.
Logged

BernardLanguillier

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 13983
    • http://www.flickr.com/photos/bernardlanguillier/sets/
Re: EOS 5Ds- why all this negativity?
« Reply #134 on: February 16, 2015, 06:04:37 pm »

So considering there is not a significant DR upgrade in the offering to match the Sony sensor, is Canon going to throw a bone or 2 to make their camera different?  I find it baffling that RAW histograms or a ETTR metering mode isn't included for the landscape guys.  Without some differentiation, if Nikon releases a 50MP D910 that has 3 more stops DR and less shadow noise, then only for those people locked into Canon does the 5Ds make any sense. 

It's more a when than an if really. The 2 years old D7100 pixels already have significantly more DR than the current FF Canon sensors at lower ISO.

Now it remains to be seen whether Nikon will finally decide to release their lng patented next gen T/S lenses.

Cheers!
Bernard

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
Re: EOS 5Ds- why all this negativity?
« Reply #135 on: February 16, 2015, 08:30:01 pm »

Hi Ray,

That is when comparing down-sampled output at some size of print resolution, which says nothing about the sensors themselves. That's why one needs to use the 'screen' settings on the charts, instead of the 'print' settings. The 'print' settings have already been normalized, to an arbitrary size, thus making all subsequent normalizations 'iffy' at best. IMHO, it's better to take the native sensor quality ('Screen' settings) and draw conclusions based on that, optionally with normalization to a common size if one wants to compare apples and oranges.

Hi Bart,
Yes, of course. I should have realised that. How silly of me.  ;D 

I guess I get so used to looking at the 'print' values on the DXOMark charts because I find them more useful for comparison purposes, and because I believe they more accurately relate to the performance of the entire sensor. It is generally the entire sensor that I use when taking a photo, so that's what interests me most.

When comparing the image quality in prints, I also insist on comparing prints of the same size. It makes no sense to do otherwise.

However, what I find quite remarkable in this comparison of the 7D2  and the 5D3 at the pixel level, is how close the performance of the individual pixels are with respect to DR, yet the 5D3 pixels are significantly bigger. Before this announcement of the new 50mp camera came to my attention, I wasn't aware that Canon had made such significant strides in the development of the 7D2. I'm now a Nikon shooter and pay less attention to Canon developments, unless they grab my attention because they are so mindboggling.  ;)

Some years ago I was using the Canon 5D a lot, and my main gripe was not so much the grain-like noise in the shadows, but the banding. Grain tends to be more acceptable for anyone who has shot with film, but banding seems totally unnatural. When I see banding in the shadows, I feel like I have a defective camera.

I'm pleased to learn that Canon have removed all banding and pattern noise in the 7D2 (or at least significantly reduced it) so it is reasonable to expect that the 5DS will have the same improvements.

If Nikon were to announce a 60mp full-frame based upon D7100 pixels, I would want to be assured they had removed the banding that is quite prominent in D7100 shadows.

Cheers!

Logged

AlterEgo

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1995
Re: EOS 5Ds- why all this negativity?
« Reply #136 on: February 16, 2015, 09:03:08 pm »

If Nikon were to announce a 60mp full-frame based upon D7100 pixels
it is not a Sony sensor, is it ? it is from Toshiba (24mp = D5200, D7100)... do you expect Toshiba to leap to 60mp FF ?

I would want to be assured they had removed the banding that is quite prominent in D7100 shadows.
and are Sony Semi sensors known for banding like some of their competition are ?
« Last Edit: February 16, 2015, 09:04:44 pm by AlterEgo »
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
Re: Total sensor SNR ("DR") vs local SNR in the mid-tones
« Reply #137 on: February 16, 2015, 09:15:38 pm »

By the way, I would expect that as you push to extremely high EI settings, the increase in 18% SNR for each doubling of EI gradually increases from 3dB towards 6dB, a sign that sensor dark noise has become significant even in the "18% mid-tone" region.  Is that what you see?

Hi BJL,
What I see on the DXOMark graphs is that the 18% SNR difference for the 5D3, between ISOs 51,200 and 102,400 is still only 2.9dB. I rarely use ISO settings greater than 3200, so any increase in the 3dB reduction in SNR, likely only for underexposures at maximum ISO settings, is not even nearly a problem for me.  ;)

I'm just hoping that DXO get onto the job quickly, of testing the 5DS as soon as it's available. I'm disappointed they haven't yet tested the Samsung NX1, despite it having been available for a number of months now.
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
Re: EOS 5Ds- why all this negativity?
« Reply #138 on: February 16, 2015, 09:29:30 pm »

it is not a Sony sensor, is it ? it is from Toshiba (24mp = D5200, D7100)... do you expect Toshiba to leap to 60mp FF ?

Why not? Sony leapt from 16mp to 36mp, and Canon are now about to leap from 20mp to 50mp. The question I ask myself is why did Nikon choose a Toshiba sensor for the D7100? Did Sony not have the manufacturing capability?


Logged

BernardLanguillier

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 13983
    • http://www.flickr.com/photos/bernardlanguillier/sets/
Re: EOS 5Ds- why all this negativity?
« Reply #139 on: February 16, 2015, 09:53:56 pm »

The question I ask myself is why did Nikon choose a Toshiba sensor for the D7100? Did Sony not have the manufacturing capability?

- Price?
- The willingness to have less dependency on Sony?
- Part of a broader collaboration framework?

Cheers,
Bernard
Pages: 1 ... 5 6 [7] 8 9   Go Up