The GetDPI Photography Forum

Great to see you here. Join our insightful photographic forum today and start tapping into a huge wealth of photographic knowledge. Completing our simple registration process will allow you to gain access to exclusive content, add your own topics and posts, share your work and connect with other members through your own private inbox! And don’t forget to say hi!

Compressed VS Uncompressed DNGs - A quick test

MaxKißler

New member
Hi guys,

I finally had time to do some testing and would like to share my results with you. It is about something that has always kept me wondering since first testing an M9: Is there a difference between shooting compressed or uncompressed raw files with the Leica M9? And if yes, how significiant is it? After all file size is twice as large and this is something for me to consider.
The first thing I did, when I received my camera in June of this year was to set it to uncompressed capture mode as I'm always aiming for the highest IQ possible. Logic says (at least mine does) that the words "lossless" and "compression" don't add up too well.

So what I tried to achieve in this test was to determine how severe an adjustment had to be, in order to show a negative effect on image quality and whether it shows earlier in the compressed raw file. Please note: These adjustments are beyond all reason and a sane person would probably never adjust that much. I personally try to do as little PP as necessary but it's a test after all, so please put on your color muting protective goggles! :cool:
What you could say is that I basically tried to destroy the files through postprocessing...


The first image of each set was shot in uncompressed mode and the second one was in compressed mode. The first set was overexposed just enough for LR to show a highlight clipping warning.
This is what the files looked like right out of the camera:



And this is what they looked liked after these adjustments to the right. I boosted vibrance and saturation to try to cause banding in the sky...



Here are some 100% crops side by side:








Next I did the exact same thing with an increadibly underexposed file.
Before:



And after this processing-torture:



100% view:




Even though I did this test with an M9 the same might apply to the new M aswell.


So in the end what does this tell us? Well, to me it is very confusing that there doesn't seem to be the slightest difference between compressed and uncompressed raw files.
What am I missing here? I would at least have expected some banding issues in the sky, more noise, artifacts whatever. Or is it just my screen that isn't capable of displaying the horrible differences?

On the bright side it shows what a great camera the M9 still is. :rolleyes:



What do you think?

Regards
 
V

Vivek

Guest
Photographers are weird. Over at the Sony section, many are demanding uncompressed RAWs and how the compressed RAWs "crunch" their images at the pixel level and yet here you are enjoying compressed RAW. :)
 

fotografz

Well-known member
I guess it depends on whether the compression is lossy or not, even if they say is it lossless.

I can't see the difference with my S2, which is lossless compression DNGs, but that camera's bit rate is higher than most. The M9 is also lossless DNG files.

I wonder if it is different with other RAW file formats when PPed in Adobe software?

(I'm out of my geek-depth here, so take it all with a grain of salt). :ROTFL:

- Marc
 

k-hawinkler

Well-known member
I was under the impression the M9 was not lossless but the newer M is.
You are correct.
The M9 has indeed lossy compression.
The M240 has lossless compression.

The algorithm used in the M9 throws practically speaking away bits at the noise level. I think I have seen only one carefully crafted example where one could see a difference when pixel peeping.

Of course one can take an M9 uncompressed file and compress it with Adobe software in lossless fashion.
 

MaxKißler

New member
Lossless or not: It's good enough for my taste. So I'll probably change my settings to compressed mode now. Usually I don't over- or underexpose by more than a stop so it won't make much of a difference I guess.

Btw, I get consistent results when using higher ISO speeds and even when I push underexposed high-ISO files I don't see any difference in the amount or character of noise.
 

Shashin

Well-known member
You are thinking your eye is really good at seeing luminance difference. The reason 8-bit image files are standard is that for the eye to perceive a stepless gradient from black to white, the gradient needs approximately 200 levels of gray. So when you have an image starting with so much more, you really need to push it to see it break down.

The other problem is I can actually have far fewer than 200 levels of gray to reproduce a "photographic" image. The more detail and the shorter the gradient, the less I need.

The idea of "the more the better" might be true, it might not be perceived either. And if you cannot see the difference in a photograph, is it really different?

On an associated note, many photographers default to ProPhotoRGB because it is a large gamut. What they don't understand is you don't actually get more discrete colors because of the gamut size--what you get is coarser color gradients and a whole bunch of empty registers (in the case of ProPhoto). Going to ProPhotoRGB actually reduces the number of discrete colors in a file which is why it needs to be 16-bit in order to simply not band. If you don't believe me, take an AdobeRGB image and convert it to ProPhotoRGB and watch the area of the histogram compress. Bigger is not always better.

Now, about this fixation about the number of pixels...
 

D&A

Well-known member
Shashin, couldn't you sum all you said by saying "Less is More"? :lecture:

All kidding aside, I'm appreciative this test was performed as I often wondered about the M9 lossy compression. If these tests indicate anything, it's that the compression algorithms used are extremely fine. I've seen other examples of high quality lossy vs. loseless compression in other cameras and there often is a world of difference.

Dave (D&A)
 

MaxKißler

New member
...

On an associated note, many photographers default to ProPhotoRGB because it is a large gamut. What they don't understand is you don't actually get more discrete colors because of the gamut size--what you get is coarser color gradients and a whole bunch of empty registers (in the case of ProPhoto). Going to ProPhotoRGB actually reduces the number of discrete colors in a file which is why it needs to be 16-bit in order to simply not band. If you don't believe me, take an AdobeRGB image and convert it to ProPhotoRGB and watch the area of the histogram compress. Bigger is not always better.
...
Maybe I misread what you were saying but I don't understand why anyone would do this?

I personally use ProPhotoRGB because I know (or am quite certain) that my cameras deliver files that have a large enough color gamut for ProPhotoRGB to make sense. Though I admit, it doesn't make much of a difference to using Adobe RGB. Before printing, the files get translated to the according color spaces anyway which happen to be a lot smaller than any of the former ones.

Sadly, none of this is making me take better images...
 

MaxKißler

New member
BTW, I exported the first two images as 16bit tiffs and processed a little more aka "made completely senseless adjustments" that should have produced a difference between the compressed and uncompressed image, viewed them side by side on one of my school's Eizo screens and didn't see a difference. As I've previously said, from now on compressed mode it is....
 

Shashin

Well-known member
Max, if you use an unnecessarily large gamut like ProPhotoRGB, you can simply be throwing away data. I never simply default to that unless it is clear AdobeRGB is not large enough. I have yet to find AdobeRGB too small for what I shoot. Also, if you are letting your CM system simply rescale one gamut into another, then you are also transforming the image. Another reason I like AbobeRGB as it is closer to the print space.
 

k-hawinkler

Well-known member
Well, forward looking to better software and better displays doesn't it make sense to use ProPhotoRGB and 16 bits?
 

MaxKißler

New member
Max, if you use an unnecessarily large gamut like ProPhotoRGB, you can simply be throwing away data. I never simply default to that unless it is clear AdobeRGB is not large enough. I have yet to find AdobeRGB too small for what I shoot. Also, if you are letting your CM system simply rescale one gamut into another, then you are also transforming the image. Another reason I like AbobeRGB as it is closer to the print space.
Good point Will.


Though I feel using a perceptive rendering intent is usually getting me very close to what I want / previously saw.
 
Last edited:

Shashin

Well-known member
Max, just as your post is really alluding to, it is about the results. There are so many variables in a photograph, to to mention the big variable of the photographer's taste, that is really hard to make hard and fast rules. I have stopped sweating the small stuff long ago. It is good to know because sometimes it does make a difference.
 

Shashin

Well-known member
Well, forward looking to better software and better displays doesn't it make sense to use ProPhotoRGB and 16 bits?
16-bit is always worth using. Why lose data by going to 8-bit?

ProPhotoRGB might be worth using depending on the gamut of your image. AdobeRGB will be better and retain more detail if the image falls into that gamut. A particular color space does not make the image "better" as long as nothing goes out of gamut. You can maximize your data if you can closely match your image to a particular gamut.

Also, as far as choices for bit depth and color space, it has a lot to do with the ability to process the image rather than the image "looking better." You will not see the difference between a perfect 8-bit or 16-bit image.
 

k-hawinkler

Well-known member
Thanks. Processing the image as good as possible is my main concern.
I am also finding going back to old RAW files and processing anew gives me better results.
Partially through improved software and partially because I learned a thing or two. :D
 

ced

Member
For me the images on the right are softer(less crispness in details) & mushy in comparison, but then I am no expert.
I also use the M9 and as I don't shoot thousands and rapid shoot etc. the file size is not an issue, you can dedicate extra disks for storage at today's prices.
 

MaxKißler

New member
For me the images on the right are softer(less crispness in details) & mushy in comparison, but then I am no expert.
I also use the M9 and as I don't shoot thousands and rapid shoot etc. the file size is not an issue, you can dedicate extra disks for storage at today's prices.
Maybe you either see it because you want to see it or because the windows snipping tool is compromising image quality. What if I didn't write which is which or if I now told you I switched it on purpose?

Perhabs I should have mentioned that it was very windy so there might be some motion blur in it.
 
Top