True, but there's much more striping amongst the 5D2's so more variance from each other is expected. At any rate, it's pretty clear to me why I was getting better results when pushing 1Ds3 files.
I am also struggling with very monaxial colour in shadows when pushing the 5D2. Seems like it records one shade of purple and one shade of green and that's it. Feels like a 10 bit camera. The 1Ds3 and 5D suffered from this too, but not to the same extent.
brainiac wrote:
I am also struggling with very monaxial colour in shadows when pushing the 5D2. Seems like it records one shade of purple and one shade of green and that's it. Feels like a 10 bit camera. The 1Ds3 and 5D suffered from this too, but not to the same extent.
I get that with the 1Ds3 as well... but I really have to do some extreme pushing to make it visible. The 5D2 shows it much earlier. You don't get this with Nikon cameras... but Nikon cameras aren't as sharp.
Daan B wrote:
I get that with the 1Ds3 as well... but I really have to do some extreme pushing to make it visible. The 5D2 shows it much earlier. You don't get this with Nikon cameras... but Nikon cameras aren't as sharp.
It seems that there's only so much salami (read image entropy/IQ) and Canon have chosen to recalibrate the meat slicer. Undoubted someone without brianiac's experience and rabid testing thought they were making a real improvement (and it didn't take much to convince the ad guys). Realize that this is fast developing into a R Galibraith situation but brianiac is in the cross hairs, now, for a fault that most everybody at Canon can't see/understand.
Thanks Mr. b! So now that I bought a 5D M II I still can't sell my 1Ds Mk III? Not that I (REALLY) expected a $2700 Canon camera would best the (obsolete) $8000 flagship.
And by the way Flash sucks!
Daan B wrote:
Interesting to see if the ISO performance of Brainiac's 5D2 is exemplary for all 5D2's out there... And if not (meaning other 5D2's show different performance - let's say ISO100 cleaner than ISO160), what could be the explanation?
The 5D2 implements intermediate ISO's in the same way that the xxD series does -- by digital multiplication post-capture from an analog amplification using "even" ISOs 100-200-400-800 etc. Therefore the amplifier noise for the intermediate ISO will be the same as that of the analog ISO it was derived from, scaled up or down by a factor of about 1.25. But note that the signal is also scaled up or down by 1.25, so one doesn't gain anything in terms of S/N by using the intermediate ISOs. I have them turned off on my 40D and 1D3.
Thanks for that Emil. I had read that somewhere before. I'm sure it's the reason. To clarify, despite the fact that the 160 iso black frame looks good in these tests, according to Emil's link, the 160 iso DYNAMIC RANGE, which is what the real issue is, is no better than 100 iso, or indeed 125 iso, since those settings will have 1/3 and 2/3 of a stop more headroom at the top. Maybe.
p.17 #12 · Are you seeing 5D2 Criss-Cross Pattern?
brainiac wrote:
Thanks for that Emil. I had read that somewhere before. I'm sure it's the reason. To clarify, despite the fact that the 160 iso black frame looks good in these tests, according to Emil's link, the 160 iso DYNAMIC RANGE, which is what the real issue is, is no better than 100 iso, or indeed 125 iso, since those settings will have 1/3 and 2/3 of a stop more headroom at the top. Maybe.
I'm not sure how the 5D2 is set up explicitly, but if it follows the xxD series, then 125 is 100 pushed, 160 is 200 pulled, 250 is 200 pushed, 320 is 400 pulled, etc. Dynamic range on Canons tends to be largest at ISO 200 -- RAW saturation at ISO 100 occurs somewhat below the saturation level of the ADC, while at ISO 200 it fills the range; and read noise in RAW levels is about the same for 100 and 200.
I have to say I'm amused by all the black frames being posted exhibiting pattern noise. Did nobody read my earlier post? It's been a "feature" of Canon electronics for a long time.
However, it has no latitude for underexposure. The 1Ds3 is very very nearly as good at iso 6400, AND you can get away with underexposure when the lights are up-down and all over the place (as ususal).
p.17 #15 · Are you seeing 5D2 Criss-Cross Pattern?
brainiac wrote:
Because I am a terrible photographer.
BTW, any event photographer can tell you that it is impossible to keep up with adjusting every picture for those wearing black or white (as in this case), and no camera light meter in the world behaves reliably under lights which move, fluctuate in intensity, and change colour rapidly, as they were designed to do. As a result, chimping is crucial, but you can't do it for everything when the action is fast and you are fighting through a crowd. Consequently some proportion of spontaneous pictures will be ill-exposed, no matter how methodical a photographer you are, and it is invariably the best shot that has the worst exposure. In these circumstances the most forgiving camera wins.
p.17 #16 · Are you seeing 5D2 Criss-Cross Pattern?
ejmartin wrote:
I have to say I'm amused by all the black frames being posted exhibiting pattern noise. Did nobody read my earlier post? It's been a "feature" of Canon electronics for a long time.
Hi Emil, I read your post and really appreciate your knowledge and ability to explain this matter. But why is this "feature" much more prominent in the 5D2 than say the 40D?
While you are here, do you have a comment to the DXOmark test results, and why they don't indicate this level of pattern noise?
p.17 #17 · Are you seeing 5D2 Criss-Cross Pattern?
Braniac: have you tried DXO pro for converting the Canon 5DII raw file? I find it does a much better job than lightroom with Canon raw files, though I'm still new to the 5DII (having used it a lot with the original 5D). DXO also has camera and iso specific noise reduction. I used it with iso 6400 iso files for the 5DII, usually using about 1/2 of DXO's default noise reduction at that iso, and didn't notice any of the shadow pattern noise you've identified. You might try the free download and see how it does.
p.17 #18 · Are you seeing 5D2 Criss-Cross Pattern?
I notice the exact same pattern on my 5Dm2. However, I once I apply NR and knock down magenta, the pattern goes away.
Alternatively, I apply 'fill light' in LR VERY judiciously, and stick with exposure, recovery, and blacks.
One last tip is that I expose past the right now. Always, as in, up to 2/3 stop. It ALWAYS comes back for me in LR, because for whatever reason, everything in LR (with no presets applied) comes in darker than the JPG previews. I was kind of annoyed like you are til I figured that out. Now I have shadow detail again. Don't know what you're processing with, but shoot in RAW+JPG, and you may see what I mean.
Then again, I could be a raving lunatic with a poorly calibrated monitor and some weird setting in LR that I don't know about.
p.17 #19 · Are you seeing 5D2 Criss-Cross Pattern?
alundeb wrote:
Hi Emil, I read your post and really appreciate your knowledge and ability to explain this matter. But why is this "feature" much more prominent in the 5D2 than say the 40D?
While you are here, do you have a comment to the DXOmark test results, and why they don't indicate this level of pattern noise?
I haven't done any testing of the 50D or 5D2, so I can't say for sure why it seems to be more of a problem with the 5D2. One possibility is that the data rates for 21MP are forcing a faster throughput for the amplifier and ADC, and that results in more RF noise that appears as pattern noise when the pixels are read out sequentially.
As for why DXO data don't indicate the level of pattern noise, the contribution of pattern noise to the noise of any individual pixel is actually rather small, negligible in the standard deviation of pixel values relative to other noise sources that are uncorrelated from pixel to pixel. The reason it is so visually apparent even though its magnitude is small is that it is coherent across many pixels in a row or column, while the random noise averages out; our perception is designed to detect edges and patterns and so readily integrates the data across pixels to see the pattern. So DXO data and any other tester measuring only the standard deviation of pixel values will not pick up the pattern noise, even though it may be quite apparent to the viewer. There are other measures besides the standard deviation that can quantify the pattern noise, but I haven't done any analysis as yet.
p.17 #20 · Are you seeing 5D2 Criss-Cross Pattern?
Brainiac and anyone else, please send your findings and images to Chuck Westfall via the Digital Journalist website (I have his email also and can dig that up), and I will do the same.
I am sure this is a firmware issue. Prints I have made from the 5D2 sample images supplied by Canon do not show these deficits (pre-1.0.6 firmware), and neither do prints made with the 1.0.6 firmware.
I am convinced the 5D2 is on par with the 1DS3 otherwise, and this is not "normal" performance.
Way too much shadow noise at low ISO.
