0-800 contains 801 unique values (ie is continuous)
801-1424 contains 320 unique values (skips 1 out of every 2)
1424-1427 contains one unique value (skips 2 of 3)
1428-2023 contains 149 unique values (skips 3 of 4)
2024-2029 contains one unique value (skips 5 of 6)
2030-4093 contains 258 unique values (skips of 8)
p.2 #2 · Nikon D800/e users - a little help please ?
All theoretical debate about "lossless" compression aside, I just don't understand why people worry about needing to compress files in the first place?
p.2 #3 · Nikon D800/e users - a little help please ?
MaxBerlin wrote:
More than anything (even shutter blur) the Sony 'lossless compression' is causing artifacts and bad tonal transitions. It's not too visible in the 1st iteration but when you go to post (or zero post) and then JPEG it's glaring. I haven't been able to create a color JPEG without the issue.
If you have a good viewing setup (large screen, high resolution) have you seen such problems with tonal gradients in the 800(e) ?
Thanks in advance.
Max, it takes some amount of courage to potentially gore the new sacred ox in these and other forums. I appreciate you sharing your findings. I'm a D800e and 5D3 owner and am very interested in the A7R so I've been following user reports closely.
Regarding my D800e files, I haven't noticed anything troublesome in uncompressed NEFs converted in ACR (to Adobe rgb). But I'll go back and take a better look - I usually don't pixel peep beyond 100%.
As for your findings regarding the A7R files, I thought you'd like to know you're not alone. Here's a quote from a D800 owner who has returned his A7R due to the compression issue:
" The file compression applied to the raw files is lossy. Artifacts are clearly visible at 100 percent view as sort of an egg shell stipple on surfaces that should be smooth. While this would be unlikely to show in a print up to native size (and actually might help smooth transitions while maintaining a sense of surface), heavily working a file in Lr and Ps will make it worse and visible in prints. Tonal banding is also likely in more extreme cases since these are essentially 8 bit files."
0-800 contains 801 unique values (ie is continuous)
801-1424 contains 320 unique values (skips 1 out of every 2)
1424-1427 contains one unique value (skips 2 of 3)
1428-2023 contains 149 unique values (skips 3 of 4)
2024-2029 contains one unique value (skips 5 of 6)
2030-4093 contains 258 unique values (skips of 8)
Digilloyd also discusses the issue
Lossless compressed is less than 40mb/s on the D800(E) and never shows your issues. Uncompressed can indeed be 70mb, but I don't know any photographer that uses uncompressed on the D800 (out of 6 with a D800(E)).
Have you tried printing the issues yet? (not taking a screenshot, but printing the whole file or a crop)
What will you be doing when your problems stay and you don't see a solution?
p.2 #5 · Nikon D800/e users - a little help please ?
akclimber wrote:
Max, it takes some amount of courage to potentially gore the new sacred ox in these and other forums. I appreciate you sharing your findings. I'm a D800e and 5D3 owner and am very interested in the A7R so I've been following user reports closely.
Regarding my D800e files, I haven't noticed anything troublesome in uncompressed NEFs converted in ACR (to Adobe rgb). But I'll go back and take a better look - I usually don't pixel peep beyond 100%.
As for your findings regarding the A7R files, I thought you'd like to know you're not alone. Here's a quote from a D800 owner who has returned his A7R due to the compression issue:
" The file compression applied to the raw files is lossy. Artifacts are clearly visible at 100 percent view as sort of an egg shell stipple on surfaces that should be smooth. While this would be unlikely to show in a print up to native size (and actually might help smooth transitions while maintaining a sense of surface), heavily working a file in Lr and Ps will make it worse and visible in prints. Tonal banding is also likely in more extreme cases since these are essentially 8 bit files."
THANK YOU ! I was called a troll, a Canon tout, an idiot, made to believe I was 'seeing things' that weren't there and wasn't seeing things that were there.
I didn't know it would be so controversial. I thought that Sony users together could band together and help Sony provide a lossless algorithm firmware update EVEN if it meant slower frame rates.
It's not a sports action camera by any means and the shutter delay and other issues means we're not looking for high FPS.
Just high IQ with manual lenses through a high magnification EVF. Is that too much to ask for ?
p.2 #6 · Nikon D800/e users - a little help please ?
MaxBerlin wrote:
Just high IQ with manual lenses through a high magnification EVF. Is that too much to ask for ?
That's some of what I'm hoping for (and maybe more useful bracketing and slightly faster fps) and was hoping the A7R would be it. Being fortunate enough to own a D800e, I can wait for an A9R if need be. I certainly hope Sony pays attention to user's review of the A7R. It has so much potential and appeal and could be (will be? should be?) a game changer.
Maybe Sony will issue a firmware "fix" for the compression issue, who knows? But I think we can assume that if the issue wasn't ever raised, Sony most likely wouldn't feel the need to fix a "non-problem" - so good on ya for sticking to your guns!
p.2 #8 · Nikon D800/e users - a little help please ?
Just to add to the debate: the 800e help screens clearly state that there are two different types of compression.
Lossless Compression is specifically stated to be a reversible algorithm. That necessitates there is no data loss; you can take the compressed file, and build the original uncompressed file. You can't do that if data has been lost.
The other type of compression offered is visually lossless, and is specifically stated to be non-reversible. That is, you can't get back to the original decompressed file, because data has been discarded.
This latter compression results in a greater reduction in file size.
0-800 contains 801 unique values (ie is continuous)
801-1424 contains 320 unique values (skips 1 out of every 2)
1424-1427 contains one unique value (skips 2 of 3)
1428-2023 contains 149 unique values (skips 3 of 4)
2024-2029 contains one unique value (skips 5 of 6)
2030-4093 contains 258 unique values (skips of 8)
This is a fairly straightforward bit depth reduction for lossy compression, and it's commonly used.
Since a white balance adjustment is an offset applied to the channels before gain (brightness) or gamma, it's pretty obvious that this can cause problems in very cold or very warm, or very green/magenta light. What happens is the samples are restored with large gaps as you go up the linear range. When a large negative offset is applied the gaps are moved further down the scale. The large negative offset requires a reduction of the whitepoint (clipping the highs of the other channels), resulting in a reduction of dynamic range (scale range) and a higher gain applied to fill out the scale. Then the gamma is added on top of that.
So, you get banding.
For extreme light like moonlight, candle light, etc, shoot lossless - whether compressed or not. Bit depth matters less.
p.2 #10 · Nikon D800/e users - a little help please ?
So what's the truth? I'm really confused on this matter. Some people are saying that these are real RAW files and some are saying that this is a lossy RAW where some information have been discarded to achieve a smaller RAW file.
p.2 #11 · Nikon D800/e users - a little help please ?
lsquare wrote:
So what's the truth? I'm really confused on this matter. Some people are saying that these are real RAW files and some are saying that this is a lossy RAW where some information have been discarded to achieve a smaller RAW file.
It's like selecting normal lossles compression on a Nikon, but at lower megapixels on top of that.
I would never use lossy compressed raw's (even prefer jpeg's above that).
p.2 #12 · Nikon D800/e users - a little help please ?
playerofwar wrote:
It's like selecting normal lossles compression on a Nikon, but at lower megapixels on top of that.
I would never use lossy compressed raw's (even prefer jpeg's above that).
The problem is that Sony doesn't specify whether the RAW option is lossless or lossy. There is only one option and that is RAW. If the rumours (I'm call them rumours because I haven't seen any conclusive evidence) from Dpreview are indeed true, then Sony blew it by not giving us a true lossless RAW option.
I appreciate you sharing your findings. I'm a D800e and 5D3 owner and am very interested in the A7R so I've been following user reports closely.