As Jonathan says, it would be nice if all process the files the same way, that is applying input levels 25 instead of +4 exposure in lightroom.
The levels adjustment does effectively equal +3,35 stops exposure correction.
log(255/25)/log2 = 3.35. Take this with a big grain of salt, I am not 100% sure if the levels operation is linear.
Guys, I've examined the ISO 100 files from Will Patterson and yauyi.
If they really were processed with +2 in DPP and then +4 stops, it looks like those two copies of the 5D2 have the pattern noise under control, with an overall noise level very close to the 1Ds3.
This is based on histogram calculations where I have tried to compensate for different processing.
It would be very, very helpful if all contributors followed the suggested procedure, and also if those who state that they don't have this problem would do the lens cap test.
Even if after this procedure you can't see the pattern noise on your monitor, we need the files to be able to make a valid comparison.
alundeb wrote:
Guys, I've examined the ISO 100 files from Will Patterson and yauyi.
If they really were processed with +2 in DPP and then +4 stops, it looks like those two copies of the 5D2 have the pattern noise under control, with an overall noise level very close to the 1Ds3.
This is based on histogram calculations where I have tried to compensate for different processing.
It would be very, very helpful if all contributors followed the suggested procedure, and also if those who state that they don't have this problem would do the lens cap test.
Even if after this procedure you can't see the pattern noise on your monitor, we need the files to be able to make a valid comparison.
Applying +4 with the exposure layer in PS gives a very different effect than pushing levels in PS. In order to make the pattern as visible as possible you need to apply levels because this will enhance the contrast, which the exposure layer will not do to the same extent.
If Will and Yauyi can redo their test by applying a level push in PS of 25 that would indeed be very helpful.
That 50 iso result is very exciting news. I didn't test 50 because I have always been lead to believe that 50 was fake and inferior to 100. I am going to do a 50 test with fingers crossed. Looks like it might solve the problem for when I just want maximum DR without shadow noise patterns. Great discovery.
For the sake of comparison it is important that everybody processes in the same way. I originally suggested some settings which I will elaborate on here, although they do rely on people using DPP and photoshop...
1/60th second shutter (this matters since we are looking at shadow noise - 1/100th won't do)
smallest aperture
lens cap on (which lens doesn't matter, unless it's very radioactive)
cover viewfinder eyepiece
white balance daylight
all NR and tone effects turned off in camera
DPP brightness +2, faithful style, contrast and sharpening to lowest possible setting, ALO off, NR off, lens correction off
(image will still look completely black at this stage)
photoshop levels with input of highlight set to 25 and no other adjustments
upload at 900x600
yauyi wrote:
Out of curiousity, I tried the same test on iso 50, man.....this is much cleaner than both iso 100 and iso 200! I was expecting it to be much worst than iso 100...WTF
Perhaps the down shifting algorithm applies a floor and sets low values to 0. That would get a lot more 0 values.
I totally suck at photoshop and don't know how to get around using it, but when i get home from work tonight I'll see if I can figure out how to do the levels input 25 thingy. So correct me if i'm wrong, basically I convert the RAW file to JPG from DPP with + 2 stops first, then import this JPG (+2EV) file to CS3 or CS4 to do the level 25 type deal (i assume this can be found under the tools menu or something) , instead of pushing 4 stops exposure in Lightroom like i did, right?
yauyi wrote:
I totally suck at photoshop and don't know how to get around using it, but when i get home from work tonight I'll see if I can figure out how to do the levels input 25 thingy. So correct me if i'm wrong, basically I convert the RAW file to JPG from DPP with + 2 stops first, then import this JPG (+2EV) file to CS3 or CS4 to do the level 25 type deal (i assume this can be found under the tools menu or something) , instead of pushing 4 stops exposure in Lightroom like i did, right?
Almost... Open the RAW file in DPP, adjust brightness/exposure by +2, set to faithfull, every parameter (sharpening, contrast, etc) to zero, NR off, open directly from DPP into PS (under tools: transfer to PS), adjust levels +25 in PS.
p.12 #11 · Are you seeing 5D2 Criss-Cross Pattern?
I have an ftp site that I will put my own raw files into also anyone else that wants to create a raw file using the body cap method please feel free to upload it just ad your user name in the file name.
also add the iso in the name as I am not sure it shows up in the raw metadata.
p.12 #13 · Are you seeing 5D2 Criss-Cross Pattern?
stiksandstones wrote:
wow, taking pics with lens caps on....the intardnet has been broken.
Sorry - this is not your usual service. To get the intardnet again you need to plug back into the big socket next to the one for your fistboard. Wipe the drool off your fists before attempting this. If you want to come on the internet again you need to get nurse to supervise you. I won't report you this time, but don't do it again - the second lobotomy's the killer.
p.12 #15 · Are you seeing 5D2 Criss-Cross Pattern?
Thanks a lot John. I would recommend that anyone uploading to the ftp put her username in the filename as well as the iso, because ftp often has no warning for overwrite.
p.12 #16 · Are you seeing 5D2 Criss-Cross Pattern?
brainiac wrote:
I wonder whether at 50 iso we lose a stop at the highlight end.
Yes, since it's pulled ISO 100, you'll lose about a stop DR but you should get better shadow detail, at least that was the case with pulling on film.
I've always been amazed about the negative comments about ISO 50 especially regards noise. It is in no way noisier than ISO 100 and I use it where DR is not pushed to the max and I need slower shutter speeds.
p.12 #19 · Are you seeing 5D2 Criss-Cross Pattern?
brainiac wrote:
My problem with the 5D2, and the reason that I am interested in this limit is that Canon stated that the 5D2 would be as good or better than the 1Ds3. I duly traded down to a pair of 5D2's and sold my 1Ds3 which I had used for a year. I subsequently found that I could not achieve the same dynamic range from the 5D2 because the shadows were a mess, with quite a large scale criss-cross pattern stopping me from eking out maximum dynamic range. I am now trying to find out whether this experience is general, with a view to making an assessment about how likely Canon is to fix this, and which way to go in the future....Show more →
you traded a $7000 camera for a pair of $2600 cameras and expected better performance? Somehow I find it difficult to believe canon stated their new $2600 camera was better than their last model $7000 camera. They might have said something like it handles higher iso better, but really, I'm not sure I would have made the trade you made. My 5dMKii only does that when I'm trying to bring something out that wasn't there, but yah, if I overpush I get that... I got it worse on my older cameras. I would expect that result any time someone were trying to bring back the dead...
p.12 #20 · Are you seeing 5D2 Criss-Cross Pattern?
jamey_cc wrote:
you traded a $7000 camera for a pair of $2600 cameras and expected better performance? Somehow I find it difficult to believe canon stated their new $2600 camera was better than their last model $7000 camera. They might have said something like it handles higher iso better, but really, I'm not sure I would have made the trade you made. My 5dMKii only does that when I'm trying to bring something out that wasn't there, but yah, if I overpush I get that... I got it worse on my older cameras. I would expect that result any time someone were trying to bring back the dead... ...Show more →
I have dealt with this query a few pages back, but since we're here now, this is what Canon said when it announced the 5D, a statement which has not been retracted:
"Although the individual pixel dimensions of the EOS 5D Mark II camera are the same as the 21.1-megapixel CMOS sensor used in the EOS-1Ds Mark III digital SLR, the new sensor incorporates an improved output amplifier and a more advanced color filter that improves light transmission while retaining excellent color reproduction. By applying the same kind of advancements in sensor design and image processing technology as the recently introduced EOS 50D camera, but at higher resolution and with larger pixels, the EOS 5D Mark II achieves the highest level of image quality of any EOS Digital SLR released to date."
Before the low iso image quality issue was raised, many people were saying how great it was to have 1Ds3 quality for a fraction of the price, and how the price of the 1Ds3 was justified only by AF and build and weather sealing and pro features etc.. All of that is irrelevant because Canon has stated that 5D2 IQ is better, and it has provided no warning to those who might want best performance at 100 or 200 iso.
If it is so obvious that a budget camera will be inferior to a 1 series, then how do you explain the relative performance of the 1Ds2 and 5D? Or the 1D2N and 5D? Or the D3 and D700? The 5D IQ was a match for the 1Ds2 in every way except resolution. Your presumption has often been in error in the recent past, and there was no reason to suspect that Canon would deliberately handicap the 5D2, or claim it was better than the 1Ds3 if it wasn't. So Canon has previously allowed the 5D to challenge its top pro body in IQ, and it stated boldly that this 5D was even better than 1Ds3. At the very least Canon should have explained that the 5D2's low iso performance was inferior to the 1Ds3, regardless of how the high iso advantages were sold.
It is not unreasonable for a pro to set store by the announcements and claims of the company whose cameras she uses. It may seem strange, but I do expect Canon itself to tell something like the truth when making public announcements about technical aspects of its own cameras, which many of us use to make a living. These are not really amateur cameras. Some amateurs might use them, but without lens they cost $2700 and $8000. The 5D2 is aimed as much at wedding photographers as it is at any non-professional, and therefore Canon has a duty to tell the truth about the camera's performance. Surely Canon knew perfectly well before this camera came to market that low is performance would not be as good as the 1Ds3. As a result, the announcement amounts to an effort to mislead, in my opinion, and that's why I'm a bit miffed about all this.
This particular shortcoming isn't easy to spot without extensive use of the camera, and regardless of that, I do make plans based on the quality, features, and prices which are announced, because I consider upgrading my gear to be part of my job. I can't hand over 3 megapixel images to customers who have 12 megapixel cameras in their own pockets.
I was expecting it to be much worst than iso 100...WTF
