Histogram difference from RAW before export and final JPEG

[Matthew_918]

Hello to everyone!
This is my first post here in pixls.us. Unfortunately I did not find the exact answer to my question, I hope this discussion is not a duplicate of an already existing one and feel free to point me to other discussion or video tutorials.

My first question is on the profiles I should use:

• display profile: system display profile or sRGB (web-safe)? My monitor is not a professional one, so no AdobeRGB or wider gamut.
• histogram profile: work, export, soft proof, system display?

• The working profile in the input color profile module is set as here below, which I think should be the default setting:

So far I have been working with system display profile as display profile, which I think was the default setting. I have never calibrated my monitor, but the profile is there. I am just trying to understand what should I use. sRGB is giving more vibrant colours and contrast.
In case I need to print instead I have downloaded the dedicated ICC profile and I set it to softproof profile and I work with the soft proof activated. This requires a slightly different edit if my final goal is printing the photo or sharing it online, but I guess this is normal.
I am always exporting in sRGB (safe-web) no matter the final application. Is this correct?
I have recently printed from Saal Digital and it is suggested to apply the ICC profile of the paper together with a list of settings that I did not find in DarkTable.

ICC Profile

• Colour space: RGB
• Preserve RGB Numbers: Deactivate
• Rendering Intent: Relative colorimetric
• Black Point Compensation: Activate
• Simulate Paper Colour: Deactivate
• Download ICC Profile

I found the Black Point Compensation in GIMP however. How should my workflow be in this case? What if the Paper Colour has to be simulated?

After noticing some differences in colour rendition and contrast between the photo I edit in DarkTable and the final exported JPEG in Windows Photo viewer, I decided to import the edited JPEG back into DarkTable. In this way, if I change the display profile or the histogram profile, both RAW and JPEG undergo the same process and the comparison should be better.
Having the picture in DarkTable, I have started comparing the histogram, since it make more sense than relying on the perception of my eyes.

Here I have on purpose pushed the photo out of gamma, because I though that the “out of gamma” areas were responsible of the different percetion in colours and histogram.

So I also decided to make the picture very flat, for not clipping any white, black, saturation. The result is still that there are some differences in the histogram.

Is this because the RAW file has a higher number of bit (14bit) compared to a JPEG one (8bit)?

I have also moved my test to Nikon Studio NX and I have also noticed some differences in the histogram, after doing the exact same process, therefore I must be something related to the nature of RAW vs. JPEG.

In the end, even if my monitor might not be the most accurate one, I would like the workflow to be consistent: i.e. what I see in DarkTable should also result in the exported file.

If someone could give some more technical info on the profiles (first part of my post) and on the differences I have noticed in JPEG vs. RAW (second part), that would be much appreciated.

Thanks.

[priort (Todd Prior)]

Your display profile should be system if Windows OS is set correctly… I never rely on the OS so I place the icc file for my monitor in a folder in DT config directory. You make a sub folder called color and then two other below that called in and out. Display profiles go in the out folder. I do calibrate my monitor so that is where mine goes…Then you will be able to see them as options in the display profile dropdown. Camera profiles go into the in files for example if you use a colorchecker and custom icc profile or profiles for that …

In your case it would be the icc file that comes with your monitor… WIndow might be using its version of sRGB so you should check… There can be subtle differences to using profiles…if your monitor is close you might get away with using sRGB but the black point might be handled differently so I would stick to the monitor icc for your profile… Histogram I always leave set at rec2020 to monitor gamut deviation in the pipeline… I trust the color management to handle mapping it correctly for output by the output profile… I dont’ use the DT srgb profile as I like to use rendering intents and it doesnt support that so I use one from color.org.

As for viewers I make sure to use a color managed one…I don’t trust the windows photoviewer. It is notorious for some misrepresentation via some auto enhancing that you can’t disable…etc etc… I use xnview. It is excellent and is also great for getting a look at all the exif data. You can specify the same display profile explicitly for it as well so that DT and it are using the same profile to display your exported images.

Others in WIndows have also suggested Faststone Image viewer… being sure again to configure color management.

The histogram profile will be the color space used for the colorpicker values so you might find yourself changing it if that is something you are tinkering with…

[Matthew_918]

Thanks a lot for your help! I want to share the results of my tests in case they could be useful to someone else.
The conclusion is to use the system display profile as you have already suggested, but I needed some proofs for a better understanding.

I have discovered (maybe the most already knew) that the mismatch was due to the fact that original photo had some “out of gamma” areas and during the conversion to JPEG sRGB of course some manipulation happens in order to handle this.

For this comparison I have captured the full screen, copied the screenshots in another software, aligned the two screenshots and cut part of the one on top, so that the images for comparison are fully aligned.

In the first part of the test I proved that the exported JPEG is aligned with the edited RAW file when the image is fully within the sRGB space. Here both the blue of the sky and the grey-brown of the wall are not very saturated and there are no deep shadows/bright lights.

I moved than to a more complex image, with very deep shadows. Here there is difference, not so pronounced but visible. You can spot a sort of “frame effect”, there is like a “JPEG” rectangule, especially the top and right border are visible.

I edited the picture in order to remove every possible clipped/out of gamma pixel and now the “frame effect” between RAW and JPEG is completely disappeared.

I still got some minor differences in the histogram but are not visible on the photo now.

What about the development of the RAW file for printing? DarkTable misses the Black Point Compensation and the Simulate Paper Colour? RawTherappe has the Black Point Compensation and commercial codes (as LightRoom) include also the Simulate Paper Colour. How can I overcome this lack and eventually take into account the Colour space CMYK, which is required by some material? Should I integrate my workflow with other software?

In case I need to print instead I have downloaded the dedicated ICC profile and I set it to softproof profile and I work with the soft proof activated. This requires a slightly different edit if my final goal is printing the photo or sharing it online, but I guess this is normal.
I am always exporting in sRGB (safe-web) no matter the final application. Is this correct?

[priort (Todd Prior)]

You can’t print from the WIndows version so you will need to export to Gimp or other application. In that case you will want to export in a way that makes the use of that application the most flexible.

In DT as to your first comments about the histogram its likely expected and as you say maybe more so if you are pushing gamut… For the raw you are coming in to it in the camera colorspace and making it to the large gamut working space and then finally after editing you make the jpg using the settings in the output profile to map your jpg. If you take that jpg back in to DT and through the wide gamut working profile then you are trying to unwind that and its not completely reversible so it will be a bit different… You can see this is you move the working profile between srgb and the default rec2020 when you reload the jpg… using the waveform you can see some small changes… likely setting your histogram profile to srgb for the raw edit and comparing that to the loaded jpg with that same setting and the working profile also at srgb should be a fairly close match as there isn’t that extra colorspace conversion… but I think its pretty minimal in most cases and making sure that you have a color managed viewer that is set properly to view your exports is more important. I see you use windows photo viewer… I’m still not at the point I trust it and microsoft is always tweaking with color management…