I know this is a 5D3 thread, but has anyone tried similar tests with a different camera like a 5D2? i.e. is this a problem with LR processing 5D3 RAW files or a LR problem in general?
trueimage wrote:
I know this is a 5D3 thread, but has anyone tried similar tests with a different camera like a 5D2? i.e. is this a problem with LR processing 5D3 RAW files or a LR problem in general?
I'm going to try this tonight as I've just installed DPP and also have LR 4.1 RC2 and C1 6.3.
trueimage wrote:
I know this is a 5D3 thread, but has anyone tried similar tests with a different camera like a 5D2? i.e. is this a problem with LR processing 5D3 RAW files or a LR problem in general?
I don't use LR but find DPP 5D2 shadow rendering better than ACR in PS. Aperture is close behind DPP and thus I use it most of the time (much better interface and control features). I only bust out DPP for an especially problematic file.
AGeoJO wrote:
Keep in mind that LR 4.1 is a Release Candidate and not the final version. So, I fully expect that they, Adobe that is, is still doing a low of tweaking. We will see....
I was wondering what program Fred used back then when he tested/compared the shadow between the 5D Mark III and D800? Fred, will you chime in, please?
Here is what I wrote on the review:
"For all tests, I used the same exposure for both cameras under the same light conditions with the same Zeiss lens. I wanted to make sure that I captured RAW images in exactly the same settings. Afterwards, I used Lightroom 4.1 (with the latest Adobe profile) and kept color noise reduction and sharpness at 25 percent which are essentially the default values."
skibum5 wrote:
i still see banding in his original DPP version and the shadow are looks like mush like it's made out of blobs of 4-6 pixels at once or something
Then you may want to convert another RAW file on your own using DPP and LR, then compare. The question here is DPP vs LR, not Canon vs Nikon.
Fred would you be able to perform the same comparison using this new RAW workflow via the DPP/CS5.5 method on those original 5D3 vs. D800 DR images you posted and post some 100% crops? This would be vastly more helpful and objective!
thw2 wrote:
Then you may want to convert another RAW file on your own using DPP and LR, then compare. The question here is DPP vs LR, not Canon vs Nikon.
umm where did i mention Nikon
i was noting difference between DPP and LR, his DPP shadows looked like they were made out of large NR blobs
in fact YOU were the first person in the thread to mention Nikon (and the ONLY one other than myself in direct reply to you)
skibum5 wrote:
i was noting difference between DPP and LR, his DPP shadows looked like they were made out of large NR blobs
OK. I too find the DPP shadows devoid of details as compared to the bright areas. It's almost like selective NR except that very slight banding and chroma noise are present. But I can't find original-size LR converted image for comparison.
It's unlikely that ACR/LR is "causing" the banding. The only periodic pattern I recall ACR generating was the maze issue with 7D images, which I believe was the result of the 7D's unbalanced G2 channel. If there is lower visible banding in DPP's conversion it's likely because it's employing banding-removal NR. The question is whether it's doing it in a camera-specific intelligent way that utilizes the invisible masked pixels or if it's using more generic techniques like with Dfine 2. Either way there have been previous threads on dpreview showing very impressive banding removal w/Define 2 on 5DM3 files - the banding was completely eliminated with no noticeable artifacts left behind.
It's not hard to analyze 5D3 shadows using undebayered data (dcraw, RawTherapee, etc.) Such an analysis would not depend on the quirks or abilities of any particular RAW converter.
rsg_1 wrote:
When I use Aperture and push the shadows, I don't get anywhere near the banding what folks were getting with LR. The files seem much cleaner than the 5D2.
I shoot a 5D2 and I never got "anywhere near the banding [that }folks were getting with LR," either! ;-)
Dan, who is at this very moment putting on his fireproof suit and helmet...
^^ If you are using Adobe's 2012 process in LR/ACR, the fill light slider should not be available. Please check your profile in Camera Calibration tab. The comparison is very interesting, but the new 2012 process should perform better.
UCSB wrote:
^^ If you are using Adobe's 2012 process in LR/ACR, the fill light slider should not be available. Please check your profile in Camera Calibration tab. The comparison is very interesting, but the new 2012 process should perform better.
yup, what I meant was the latest release not the process-sorry.
I use LR 3.6. I have downloaded high DR 5D3 RAW files from internet that didn't show any or very little banding in the shadows when pushed in PP. I have high DR 5D2 files that don't show it either after pushing. But I also have seen pushed RAW files from both cameras that do show banding. The 5D2 banding is on two axis, the 5D3 banding only on one axis. So, the 5D3 is already 'better' than the 5D2 in this regard. Wether a RAW file will show banding or not depends on wether a certain threshold is exceeded. I suspect it has to do with the entire DR that needs to be captured. The greater the DR, the deeper the shadows (no detail), the greater the chance for shadow banding.
I wouldn't be surprised if Canon found a way to eliminate the banding in ther own software. If anybody would know how to correct it in software, it would be Canon. To bad DPP is not suitable for a professional workflow that contains several hundreds of shots per assignment. But maybe it can be used for some isolated cases that will show banding in other software.
Still not convinced that DPP does the trick though. I see banding in Speedmasters DPP conversion too.
Daan B wrote:
I use LR 3.6. I have downloaded high DR 5D3 RAW files from internet that didn't show any or very little banding in the shadows when pushed in PP. I have high DR 5D2 files that don't show it either after pushing. But I also have seen pushed RAW files from both cameras that do show banding. The 5D2 banding is on two axis, the 5D3 banding only on one axis. So, the 5D3 is already 'better' than the 5D2 in this regard. Wether a RAW file will show banding or not depends on wether a certain threshold is exceeded. I suspect it has to do with the entire DR that needs to be captured. The greater the DR, the deeper the shadows (no detail), the greater the chance for shadow banding.
I wouldn't be surprised if Canon found a way to eliminate the banding in ther own software. If anybody would know how to correct it in software, it would be Canon. To bad DPP is not suitable for a professional workflow that contains several hundreds of shots per assignment. But maybe it can be used for some isolated cases that will show banding in other software.
Still not convinced that DPP does the trick though. I see banding in Speedmasters DPP conversion too....Show more →
FPN is always there, it doesn't matter where or who the RAW file comes from. It is an artifact of image sensor. Of course it only shows in shadows where shot noise falls below read noise level (HDR or severely underexposed images)
Canon do not perform extra NR in their software to artificially remove FPN. That will destroy low frequency detail and turn the photo into a mush. In fact Canon DPP has the ability for a pristine linear conversion with absolutely no NR applied. Many RAW convertors still apply some NR, even when set to 0. I don't know about ACR though.
DPP does not remove FPN, it cannot be removed by the RAW converter, so there is no such thing as "the trick". However the lack of color noise and coarse grain (for high ISO) in DPP in comparison with ACR makes it possible to remove the banding with advanced NR techniques after conversion.
Also, several hundred photos per assignment is really nothing, I go trough say 2000-3000 images a day using DPP when I am on assignment, sometimes more, not a problem at all.
speedmaster20d wrote:
FPN is always there, it doesn't matter where or who the RAW file comes from. It is an artifact of image sensor
That is my believe as well.
Canon does not perform extra NR in their software to artificially remove FPN. That will destroy low frequency detail and turn the photo into a mush.
This wasn't my statement/comment. I just repeated what has been said in this thread and at the same time expressing my critic on it.
I never said DPP removes FPN, it cannot be removed by the RAW converter, so there is no such thing as "the trick".
I never said you did.
However the lack of color noise and coarse grain in DPP in comparison with ACR makes it possible to remove the banding with advanced NR after conversion.
I still see banding in your DPP conversion.
Also, several hundred photos per assignment is nothing, I go trough say 2000-3000 images a day using DPP when I am on assignment.
Time is money. My workflow would be consideably longer if I used DPP instead of LR. I can't make that compromise. Good to hear it works for you though
