p.6 #1 · 5DSR Low ISO DR 1.5-2 stops better than 1DX ...
The worst that happens is the 5Ds has the same base ISO DR as previous Canon releases and most likely it'll have a little more due to the extra MP. The best that happens is that Canon surprises to the upside. Either way I think most in the market for this body will be buying regardless of either outcome. So relax and enjoy the ride
p.6 #2 · 5DSR Low ISO DR 1.5-2 stops better than 1DX ...
pureclassa wrote:
So multiple professionals who Canon has entrusted pre-production models of the 5DS with to field test are automatically all dead wrong in saying the sensor operates at significantly lower temps, has a CFA much more akin to the 1Ds3, and appears to have 1.5 to 2 stops more usable DR because the noise has been greatly reduced .... because of Chuck Westfall!! Why is Westfall so infallible in your book? The 2 sensors DO have similar DR, but apparently the 5DS is making far more efficient use of it.
The "1.5 to 2 stops more usable DR" is likely a result of Canon having addressed the banding issues of the 5D3, which is to be expected, as the banding in the 6D was also significantly less than the 5D3, and the banding in the 7D2 is significantly less than the 7D. However, the electronic noise is likely not so different.
As to the more accurate CFA, that makes sense. A tighter CFA will reduce the quantum efficiency which will adversely affect the noise performance, and would go a long way to explain the ISO cap. On the other hand, it is possible that Canon managed to find a way to increase the quantum efficiency through other means (sensorgen shows a decent improvement in quantum efficiency for the 7D2), which would offset the tighter CFA. But then why the ISO cap?
Either way, it looks to be a fine camera. Not as good as it could have been with Sony's Exmor tech, but a decent improvement over Canon's past sensors nonetheless. All the same, I'm waiting to see what a 6D2 might bring to the table, and am in no rush.
p.6 #3 · 5DSR Low ISO DR 1.5-2 stops better than 1DX ...
I trust Chuck Westall because I wrote and asked him if I needed to upgrade my 1.4xII to use with the 100-400 II and he replied that based on what he'd heard from Canon and his experience with the 1.4x II & III on the 70-200 2.8 II that I wouldn't see much difference. I asked the same question of one of Canon's Explorers of Light and he said yes, I'd see a big difference and please use the link on his web page when purchasing. I believe Chuck understands and accurately represents the technology although is careful about any related editorializing.
p.6 #4 · 5DSR Low ISO DR 1.5-2 stops better than 1DX ...
Yes, the truth is somewhere in between. My point was not to take one person's official statement and then throw out the findings of other people actually testing this camera.
Westfall commented, "Canon says the DR..." Now people testing it are saying "We're pulling out almost 2 usable stops more from shadows"
Something is going on here... All I'm asking is that we don't get so easily and immediately dismissive.
p.6 #5 · 5DSR Low ISO DR 1.5-2 stops better than 1DX ...
pureclassa wrote:
Yes, the truth is somewhere in between. My point was not to take one person's official statement and then throw out the findings of other people actually testing this camera.
Westfall commented, "Canon says the DR..." Now people testing it are saying "We're pulling out almost 2 usable stops more from shadows"
Something is going on here... All I'm asking is that we don't get so easily and immediately dismissive.
The key word here is "usable", and, as I suggested earlier, that most likely comes from Canon solving, or, at least, greatly mitigating, the banding issue in the 5D3 (which they did to a large extent with the 6D, and also the 7D2 vs 7D).
p.6 #6 · 5DSR Low ISO DR 1.5-2 stops better than 1DX ...
Jon Tainton wrote:
FAO jctriguy - well something of a result on a thread discussing the 5DS R, IDX, it somehow reached the depths of page 3 before the PRECIOUS DR charts/graphs and general Canon negativity rocked up.
What negativity you are talking about here. Where have you been lately, the whole world know about canon weak DR sensor since D3X released. My post provided 1DX, 5D2, 5D3, 7D2 data to support our 5Ds DR guessing game and some other known cams DR so that we don't get off too far from reality DR track like you do. A lot of canon shooter (me included) wished you were not wrong on that one, believe me.
p.6 #7 · 5DSR Low ISO DR 1.5-2 stops better than 1DX ...
mttran wrote:
I would guess a 5D2 DR at its best. At wide DR scenes, less shadows noise and better colors are only two things I hope for this coming 5Ds/r. Here ya go:
p.6 #9 · 5DSR Low ISO DR 1.5-2 stops better than 1DX ...
Since these samples are not the same subject so our eyes can only judge on the smoothness transition between highlights to shadows. What do you think what we have here? Which one has more digital look compared to other ?
p.6 #11 · 5DSR Low ISO DR 1.5-2 stops better than 1DX ...
mttran wrote:
Since these samples are not the same subject so our eyes can only judge on the smoothness transition between highlights to shadows. What do you think what we have here? Which one has more digital look compared to other ?
p.6 #12 · 5DSR Low ISO DR 1.5-2 stops better than 1DX ...
NCAndy wrote:
Until someone can produce a raw file I don't think we can tell all that much. What is that lavender haze on the left balcony in the 5Ds photo?
Yes, we need RAW files. I'm sick of jpegs processed with unknown parameters. Canon jpegs are often poor as well.
p.6 #13 · 5DSR Low ISO DR 1.5-2 stops better than 1DX ...
EB-1 wrote:
Canon jpegs are often poor as well.
+1
It always surprises me how underwhelming the released pics are ... ... when I know they can do better.
I recall a dialogue with Chuck Westfall when I got my first 1 Series camera after shooting consumer Nikon. I was having trouble with my PP (compared to what I'd been used to) and he pointed out how the Canon files are different, but they are more malleable for industry to dial in to their liking.
(Paraphrased from my untrustworthy memory.)
p.6 #14 · 5DSR Low ISO DR 1.5-2 stops better than 1DX ...
Lol this is old and people keeps/still/want complaining about the jpeg posted on their website all the time when Canon has a new DSLR launched and its funny
p.6 #15 · 5DSR Low ISO DR 1.5-2 stops better than 1DX ...
mttran wrote:
What negativity you are talking about here. Where have you been lately, the whole world know about canon weak DR sensor since D3X released. My post provided 1DX, 5D2, 5D3, 7D2 data to support our 5Ds DR guessing game and some other known cams DR so that we don't get off too far from reality DR track like you do. A lot of canon shooter (me included) wished you were not wrong on that one, believe me.
Sigh. You couldn't provide links to any of your positive FM posts about Canon then? Say 5 in your last 1,000
p.6 #16 · 5DSR Low ISO DR 1.5-2 stops better than 1DX ...
alundeb wrote:
I havent seen DxO referring directly to that ISO standard, but they do indeed test dynamic range according to the definitions in ISO 15739:2013, unlike many other online sources.
The key elements in that definition is that they test on raw data, and they define dynamic range as the ratio of the highest luminance to the luminance where the signal to noise ratio is 1:1. That is a little different from clarkvision.com and others who use the ratio of the highest signal to the read noise. The latter method uses read noise with zero luminance. That gives a little bit lower floor than with SNR 1:1, because when you still have some luminance you also have som photon shot noise in addition to the read noise. ...Show more →
Thank you Anders, for your reply and insight into sensor testing.
As you correctly stated everyone agrees that RAW data needs to be tested. So, I was surprised to learn that there is no ISO standard for a RAW data file and instead discovered it is a TIFF/EP file ISO 12234-1:2012. It is some years since I had need to work around ISO Standards, but wouldn’t ISO 15739:2013 require an ISO standard TIFF/FE image file to test and measure sensor performance? I'm guessing this is why RAW image file data is used instead of TIFF/FE files.
The declaration by DxO of RAW files being 'pre cooked' was interesting and there is some discussion of ‘baked’ ‘pre cooking’ of RAW data online, chiefly among those who test for DR, Noise performance, bit depth, colour and those with an interest in RAW conversion software, it makes for interesting reading and insight. It seems that Canon’s RAW file output is now considered the most ‘virgin’ RAW data.
p.6 #17 · 5DSR Low ISO DR 1.5-2 stops better than 1DX ...
Jon Tainton wrote:
As you correctly stated everyone agrees that RAW data needs to be tested. So, I was surprised to learn that there is no ISO standard for a RAW data file and instead discovered it is a TIFF/EP file ISO 12234-1:2012. It is some years since I had need to work around ISO Standards, but wouldn’t ISO 15739:2013 require an ISO standard TIFF/FE image file to test and measure sensor performance? I'm guessing this is why RAW image file data is used instead of TIFF/FE files.
Most raw file formats are actually TIFF (Tagged image file format) files by nature, with extension of new tags. There is room for custom tags in a TIFF file. What makes it appear that the raw file format changes with each new camera, is only the inclusion of new tags. A raw file from a "new, unsupported" cameras can nearly always be opened in old analysis software like IRIS, and we have sufficient standard tags to find the actual raw data in those files. That, combined with the pixels that are masked on the sensor and receive no light, is what makes it possible for us to immediately determine the read noise when we get access to a raw file.
p.6 #18 · 5DSR Low ISO DR 1.5-2 stops better than 1DX ...
alundeb wrote:
Most raw file formats are actually TIFF (Tagged image file format) files by nature, with extension of new tags. There is room for custom tags in a TIFF file. What makes it appear that the raw file format changes with each new camera, is only the inclusion of new tags. A raw file from a "new, unsupported" cameras can nearly always be opened in old analysis software like IRIS, and we have sufficient standard tags to find the actual raw data in those files. That, combined with the pixels that are masked on the sensor and receive no light, is what makes it possible for us to immediately determine the read noise when we get access to a raw file. ...Show more →
Anders, thanks again for taking time to explain the nature of most RAW file formats and data extraction, it is appreciated.
In your opinion, how easy is it to detect if RAW data has been 'pre cooked'? and since this is the Canon forum, have you any opinion on the online perception that Canon's DSLR's RAW data file is currently the most 'virgin' RAW?
p.6 #19 · 5DSR Low ISO DR 1.5-2 stops better than 1DX ...
Regarding the "pre cooked" comment, if I recall correctly, Canon itself admitted to use NR on RAW files on a recent camera.
Can't remember which one it was though.
p.6 #20 · 5DSR Low ISO DR 1.5-2 stops better than 1DX ...
Most recent bodies (both Canon and Nikon) do impulse noise reduction in raws. That's the salt 'n pepper specs you can see at very High ISOs that looks really ugly. You can see it on the 5DM2 and D3/D700 generation of bodies and generations before. The D3s was the first to start removing it on the Nikon side and I think the 1DX/5DM3 on the Canon side. The noise occurs infrequently enough to not show up in SNR calculations and so its removal in raws is also not detected.
DxO's "raw cooking" detection looks for pixel correlation, where noise from pixel to pixel does not follow the expected random poisson distribution. I'm not sure how advanced DxO's algorithm is but I do know they failed to detect the raw noise reduction that's employed on the A7s starting at ISO 51,200.
!! Why is Westfall so infallible in your book? The 2 sensors DO have similar DR, but apparently the 5DS is making far more efficient use of it.