Home › Forums › Help and Support › Difference in blacks – darktable vs. XNView MP on custom DisplayCAL profile
- This topic has 16 replies, 2 voices, and was last updated 10 months, 3 weeks ago by
Vincent.
-
AuthorPosts
-
2025-06-30 at 17:18 #143707
Hi everyone,
After calibrating my screen (all specs below) and getting the color management workflow properly running on my system, I’ve found that the near-blacks aren’t as deep as I’ve thought they would be. They are lighter in an image viewer compared to a RAW photo developer. I was recommended to ask here, please help me find out if it is an issue at all, or where in my workflow am I doing something wrong.
Here is everything I use:- Windows 11 23H2
- darktable 5.2.0
- DisplayCAL 3.8.9.3
- XNViewMP 1.9.2
- Asus ProArt 24″ (PA248QV)
- Calibrite Display SL
A few information for a start…
I have followed this tutorial to calibrate my Asus monitor: https://phototacopodcast.com/photographers-guide-to-screen-calibration-with-displaycal/ . I also made sure my profile is loaded in Windows settings (Color management app), and also in darktable and XNView MP as a display profile, all set to perceptual.But there is a slight problem regarding darkest areas of my photos – compared to what I see in darktable, my photos are not as dark in blacks as I’d have hoped. What darktable reads as exposure clipping, is rendered in XNView MP as something close to rgb(8,9,1), which should be around 4% lightness in HSL, so not quite black. Btw. I sample the screen with a PowerToys color picker.
This problems happens for every color profile I have made so far (I calibrated 3 different monitors) and also for a profile sent by a different person.
I have attached the photo I tested this on as well as a screenshot of the area that I’m roughly sampling from. I also added the RAW + xmp to be loaded into darktable. I also added screenshots of darktable and XNView MP color management settings and the profile itself.
Attachments:
You must be logged in to view attached files.Calibrite Display SL on Amazon
Disclosure: As an Amazon Associate I earn from qualifying purchases.2025-06-30 at 17:19 #143713The RAW and xmp didn’t make it through, here’s a zip file.
Attachments:
You must be logged in to view attached files.2025-07-02 at 1:34 #143726Finite contrast actual display TRC vs profile TRC vs “ideal infinite contrast colorspace profile” + each app blackpoint compensation of out of gamut colors.
Old topic. Photoshop rendering of developed TIFF is likely the one to trust, apps that do not match this are likely to apply simplifications.2025-07-06 at 15:52 #143787Okay, so this is all normal then, right?
2025-07-06 at 22:45 #143788Depends on display ICC TRC vs actual TRC, measure.
Maybe it could be improved (by chosing a more accurate profile, a profile that matches better actual monitor behavior… at some cost like out of gamut color cliping on sme color engine) or maybe it’s all you can get from that display.
Check near black predicted L* (ICC) vs L* measured on a large greyscale testchart, no simulation profile.-
This reply was modified 11 months, 1 week ago by
Vincent.
2025-07-10 at 7:55 #143807Check near black predicted L* (ICC) vs L* measured on a large greyscale testchart, no simulation profile.
I’m fairly new to profiling and such, so I have no idea how to do that… do you have any tutorials available?
2025-07-10 at 9:31 #143808Check near black predicted L* (ICC) vs L* measured on a large greyscale testchart, no simulation profile.
Just do as instructed, run a measurement report. Then look for predicted vs measured in L*
2025-07-12 at 15:28 #1438112025-07-12 at 15:45 #143815Report shows L* in profile is tracking accurately display behavior and contrast is not very low, so Photoshop rendering of an sRGB image should be accurate even with Adobe’s black point compensation in RGB 0-5? of out of gamut blacks. “Simpler” color management engines like some browsers or macOS may clip out of gamut blacks.
I mean, your profile choosing is “accurate type”, hence no black lift because CMM thinks you have a infinite contrast display. Only the mild lift in out of gamut blacks if CMM has “black point compensation”. Whatever Photoshop is showing is the best we can get from a well behaved 1000:1 display with an accurate black profiling.Also and aswering to the original question, whatever Darktable or Xnview show diferent from Photoshop with the type of profile you are using (real TRC including real black) is Darktable or Xniew fault (if there are differences).
2025-07-12 at 16:27 #143818Ok, thank you for your help.
I mean, your profile choosing is “accurate type”, hence no black lift because CMM thinks you have a infinite contrast display. Only the mild lift in out of gamut blacks if CMM has “black point compensation”.
Is there a setting in DisplayCAL during the profile setup that I could change to better adapt it to my display? I understand what you said as that the profile renders blacks slightly differently due to an incorrect assumption about my display, is that right?
I pretty much rely on FOSS at least freeware applications like Darktable or XNView MP. There’s also DigiKam for a profiled image viewer, but I remember it had similar results (I could try again).
What image viewer would you recommend me to use? Is there anything out there that I could try, either free or at least has a trial?
Also, I’ve attached a tone response curve for my profile, based on what I’ve seen elsewhere it should be okay then.
Attachments:
You must be logged in to view attached files.2025-07-12 at 17:52 #143820Ok, thank you for your help.
I mean, your profile choosing is “accurate type”, hence no black lift because CMM thinks you have a infinite contrast display. Only the mild lift in out of gamut blacks if CMM has “black point compensation”.
Is there a setting in DisplayCAL during the profile setup that I could change to better adapt it to my display? I understand what you said as that the profile renders blacks slightly differently due to an incorrect assumption about my display, is that right?
No, just the opposite. All in gamut colors are rendered in the most accurate way (exclusing rounding errors)
As in any non infinite contrast display, you cannot render RGB0 or 1 in sRGB, it0s out of gamut. But photoshop black point compensation is doing the best it can to preserve grey distances in “out of gamut black” without lifting it too much because your ICC profile stores actual display behavior without idealizations.
There is no other better setting for rendering the brightness near black tonesI pretty much rely on FOSS at least freeware applications like Darktable or XNView MP. There’s also DigiKam for a profiled image viewer, but I remember it had similar results (I could try again).
What image viewer would you recommend me to use? Is there anything out there that I could try, either free or at least has a trial?
Also, I’ve attached a tone response curve for my profile, based on what I’ve seen elsewhere it should be okay then.
As said before, Photoshop’s BPC should be your reference. If Darktable is doing the same visually… it is what it is regarding out of gamut blacks.
-
This reply was modified 11 months, 1 week ago by
Vincent.
2025-07-22 at 22:42 #143888Hm, I’ve taken a look at the image viewers’ settings and they have an option for black point compensation, but it doesn’t seem to do anything, I wonder why…
2025-07-22 at 22:59 #143889I have just learned that selecting “Absolute Colorimetric” in the image viewer settings with BPC “on” make the blacks noticeably darker… Perhaps this is the correct setting?
Perceptual, Relative Colorimetric and Saturation have brighter blacks. Are there any caveats to using Absolute Colorimetric? (like the way it renders the rest of the tones and colors)2025-07-23 at 0:09 #143890Unlikely to be true abs colorimetric, otherwise 255 white in a ProPhotoRGB/eciRGBv2 tiff rendered on a D65 display (profiled) will look yellow. Such “custom” abs colorimetric in those viewers are likely to crush/clip out of gamut colors near black (unwanted on most scenarios), or at least preserve “less” tonal separation between out of gamut greys, since it is the only way to show darker near black colors.
As said before, with your current XYZLUT display profile with no BPC (DisplayCAL), Adobe Photoshop rendering (Adobe BPC ON) should be your “reference” about how trhings should look “the best they can” within your display contrast window (best balance between near black “brightnes lift” and tonal separation of out of display gamut colors).
-
This reply was modified 10 months, 3 weeks ago by
Vincent.
2025-07-23 at 10:14 #143892Adobe Photoshop rendering (Adobe BPC ON) should be your “reference” about how trhings should look “the best they can” within your display contrast window (best balance between near black “brightnes lift” and tonal separation of out of display gamut colors).
Yes, but what about image *viewers*? There has to be some dedicated viewer that can display images correctly besides an editor app that costs whopping 27€ / month. Using such a heavyweight software is not practical for image viewing anyways, say you’re going through an album of photos – that’d be crazy impractical. My use case of viewing images simply wouldn’t justify the price even if I got the bundle with Lightroom for editing as well.
-
This reply was modified 10 months, 3 weeks ago by
Vente.
-
AuthorPosts