Le 10/10/2018 à 12:38, Randy Thomae a écrit : > Hi, > Hi ! > First, thanks for darktable! > > I wasn't sure how/where to ask this, is this mailing list the > appropriate place? I signed up for redmine, but couldn't find the link > to post a bug. > https://www.darktable.org/contact/ suggested sending bugs to this > mailing list. > > I may have a bug or more likely is that I'm doing something stupid. > Either way I need help. > > Camera: Sony A6000 > Lab: BayPhoto. They provide a single icc soft-proof profile: > https://support.bayphoto.com/customer/en/portal/articles/2144592-bay-photo-lab-s-icc-profile > They want files(MetalPrint) submitted in AdobeRGB. > "toggle "Gamut Warning" to make sure the colors in your file are > within our printing gamut." > > Monitor is BenQ SW2700PT, calibrated with i1display pro. darktable-cms > says its setup correctly, although I have 2 other monitors that are > not color managed currently. > darktable-cms output attached at bottom of this email. > > core settings: > OpenCL support and "very fast GPU"(Nvidia GTX1060SC 6GB purchased for > darktable.) > always use LittleCMS2 > do high quality resampling during export > demosaicing for zoomed out darkroom mode - full > pixel interpolator - lanczos3 > > File Settings: > input color profile(icp) - defaults (profile-std; gamut clip - off) > output color profile(ocp) - intent - perceptual profile AdobeRGB > (discussion here seems to indicate these dont do anything with > AdobeRGB matrix) > softproof/gamut settings (right click): display intent: perceptual; > softproof profile: BayViewing_1E103108; display profile: system > display profile > > I am editing for print. When I first set up the above settings, I > found that I was out of gamut (OOG) over most of the image I had > edited originally. I started over with the original image and edited > from the beginning checking gamut, and tweaking every module > iteratively to stay in gamut. Yes, its been tedious and I have about > 12 hours invested in this so far on this single image. I have been > studying Harry Durgin's YouTube videos for 6 months and I have 48 > operations in my compressed history stack. > First, don't freak out too much on the gamut. Out-of-gamut area will be remapped accordingly to the intent you set. You just need to ensure critical colors are not OOG, that is, for example, the brand color if you are shooting products. And of course, if low-saturated colors comes OOG, that means something is wrong. But having some OOG areas is normal and harmless : after all, you camera has a much larger gamut than any printer.
Also, gamut check is done in monitor-space RGB (tbh, I don't understand all the silly bits of that) : change the display color profile, and see how the gamut alert changes. If your monitor has a hardware LUT, and you re-apply that LUT on the OS side, you are correcting twice the colors with the same profile, so that could be an explanation. > UNEXPECTED BEHAVIOR: > I began to notice that a small but significant contiguous area that > showed in gamut when viewed at 100%, showed out of gamut when zoomed > out to fit the screen. The area in question is inside a large sun > star. So high saturation, high luminance. There are blown raw sensor > highs as well, so I'm using highlight reconstruction and color > reconstruction in this area. > That's normal : the gamut preview is computed on the interpolated preview at the current zoom level. Depending on the rounding errors and interpolation artifacts, you could see differences between zoom levels. Nothing to worry about if the differences stay reasonable. > SECONDARY UNEXPECTED BEHAVIOR: > I exported the image as a 16bit tiff with the following settings: > TIFF, 16 bit, uncompressed; max size 0,0, profile:AdobeRGB, > perceptual, none > I have seen Harry Durgin do this many times in his edits to speed up > processing. When I opened the new image, large parts of the image are > out of gamut, about 30% of the image. > icp: profile:AdobeRGB, gamut clipping:off(also tried AdobeRGB - no > difference) > ocp: perceptual, AdobeRGB > oftproof/gamut settings (right click): display intent: perceptual; > softproof profile: BayViewing_1E103108; display profile: system > display profile > If you really are a gamut nerd, use relative colorimetric as the intent so the luminance doesn't get shifted in OOG areas. Usually, luminance accuracy matters the most. > I also discovered by accident that if I change the softproof profile > on the second image(rt click menu) from Bayviewing to AdboeRGB, the > area shown out of gamut is just a little bit larger than the original > image was with the Bayviewing soft proof profile (which does not make > sense to me). > I would really check what color profile is installed on the monitor chip. Try using Adobe RGB as your display profile (on both OS and dt sides). I have a BenQ photo screen from a previous generation than yours, the LUT is flashed into the screen ROM after the calibration, so I use Adobe RGB as a reference space in the OS. That's all I can say from only a description of the problem. But I know for a fact that some photochemical prints have a gamut almost as large as Adobe RGB, so it depends on the technology they use. Although metal prints are usually not that great, if they are direct inkjet on alu Dibond. > I have read advice to "let it clip", this conflicts with the > recommendations from my printing lab. I have gone to great pains to > put color management in place, and I would like to be able to match my > prints to my monitor. I am ordering large, expensive, metal prints. I > would like to eliminate the out of gamut areas before printing. The > problem is that the out of gamut warning seems to be a moving target. Your lab is probably just trying to avoid later issues with customers by being overly conservative and retain the right to say : "we told you so, but you didn't listen" if you ever come back unsatisfied of the color rendition. Most labs fix that before printing without even telling you. Again, whether OOG is a problem or not depends on what you are shooting and what percentage of the picture is out. If they are that high-end, they will allow you to print smaller proofs at a lower cost to ensure everything is right before the big order. At least, my lab does it (Whitewall, deutsche Qualität über alles – Audi set aside 😂). > > QUESTIONS: > 1. Am I doing something stupid? > 2. Why would out of gamut indicator change between fit to screen and > 100% zoom? Which one is more likely to be correct? > 3. For exported image: > 3A. Why is it any different than the original image? > 3B. Given that I exported the image as AdobeRGB while in gamut with a > softproof profile that is presumably smaller than AdobeRGB, how is the > exported file radically more out of gamut with respect to both the > softproofing profile and even AdobeRGB than the original file? > 4. How should I generate files for my lab given the exported file > seems different than what I exported? > 5. Any suggestions for a workaround? > > Again, thanks for all that you do. I know its a tremendous amount of > work to support an application like this and I really appreciate your > efforts. > > -Randy > Good luck ! Aurélien. > randy@uBAQ2:~$ uname -r > 4.15.0-36-generic > > darktable-cmstest > darktable-cmstest version 2.4.4 > this executable was built with colord support enabled > darktable itself was built with colord support enabled > > primary CRTC is at CRTC 0 > CRTC for screen 0 CRTC 3 has no mode or no output, skipping > > DP-0the X atom and colord returned the same profile > X atom:_ICC_PROFILE (468044 bytes) > description: SW2700 1_D65_L100_G22_Rel_2018-10-06T22 > colord:"/home/randy/.local/share/icc/SW2700 > 1_D65_L100_G22_Rel_2018-10-06T22.05.26Z.icm" > description: SW2700 1_D65_L100_G22_Rel_2018-10-06T22 > > DVI-D-0the X atom and colord returned different profiles > X atom:_ICC_PROFILE_1 (0 bytes) > description: (none) > colord:"/home/randy/.local/share/icc/edid-eaeaee92ac0c149edc33ec1cae279935.icc" > description: ASUS VS228 > > HDMI-0the X atom and colord returned different profiles > X atom:_ICC_PROFILE_2 (0 bytes) > description: (none) > colord:"/home/randy/.local/share/icc/edid-67a5b7369293c3cb9128ad0fe14e29a6.icc" > description: SMB2270HD > > Better check your system setup > - some monitors reported different profiles > You may experience inconsistent color rendition between color managed > applications > > > ___________________________________________________________________________ > darktable developer mailing list to unsubscribe send a mail to > darktable-dev+unsubscr...@lists.darktable.org ___________________________________________________________________________ darktable developer mailing list to unsubscribe send a mail to darktable-dev+unsubscr...@lists.darktable.org
