- This topic has 6 replies, 2 voices, and was last updated 2 months, 1 week ago by
.
-
Posts
-
Hi,
I have a question regarding black point calibration.
My monitor (Dell UP3017) supports both gain and offset in the OSD.
Before starting the profiling and calibration, I set the white point via RGB gain to 5500K, and the black point via RGB offset (still to 5500K) and I can obtain a good match for both.
After the calibration is complete, I run a verification and when I inspect the correlated color temperature curva as a function of grey level I get values very close to the target of 5500K for grey levels >15-20%, while below that it exponentially grows to over 12000K at 5%.
How can it be so far off given that I had set the black point to a satisfactory level at the beginning?
After the calibration is complete, I run a verification and when I inspect the correlated color temperature curva as a function of grey level I get values very close to the target of 5500K for grey levels >15-20%, while below that it exponentially grows to over 12000K at 5%.
How can it be so far off given that I had set the black point to a satisfactory level at the beginning?
CCT does not matter. What matters is dE to target, for greys it means how close they are to nominal gamma value (and that may include actual TRC vs fake infinite contrast TRC) and a*b* (or equivalent perceptual color distance) to display white.
DisplayCAL HTML report shows these metrics by default and an additional “combined a*b* range” or “grey range” as a metric about how different are grey from each other regarding color tint.
Also if you use delta ICtCp instead of delta E 2000, it will be less forgiving regarding blackpoint brightness values ( delta I, actual black 0.1 nit in a 1000:1 100nit IPS vs profile black 0 nits), and that’s the reason it’s the suggested metric for HDR.
Since you are using 5500K (print work?) this does not impact you, just in case you see the high error in delta I don’t be alarmed.Also remember that if you are working with reduced contrast on purpose, for example 300:1, you may want to store actual black level in profile. This is done in profile tab, uncheck black point compensation on profile.
For typical scenarios on a >1000:1 contrast display you leave black point compensation active on profile, so ICC profile will store that display has infinite contrast. On dE 00 metric and high contrat ratio this idealization does not give you a high L* error but simplifies color management.Thank you for the clarification.
I can confirm that in terms of dE the error shown in DisplayCAL html report is the same as what I achieve during the initial tuning of the display via OSD.
The calibration I make is for print work. I keep the black level “as measured”, relying on the “simulate paper and ink” checkbox in Lightroom to give me a representation of the low contrast of the print.
Would it be more accurate if I lower the contrast on purpose? If I do that, can I still use the “simulate paper and ink” (assuming I disable black point compensation)?
The calibration I make is for print work. I keep the black level “as measured”, relying on the “simulate paper and ink” checkbox in Lightroom to give me a representation of the low contrast of the print.
It’s the straight forward approach but has some requirements to meet: you’ll be simulating paper white color under the same illuminant used to make the printer profile.
Photoshop have separate settings for simulatimng paper and for ink. I do not remember LR right now, I’ll check.
Would it be more accurate if I lower the contrast on purpose?
As long as you store the black level (low contrast black level) on display profile the “canvas” on Photoshop/Lightroom or any other tool will show the same or almost the same color coords than on a high contrast display where you simulate black ink. That’s the point of color management as long you are within device gamut.
Lowering the contrast is used to avoid seeing high contrast UI elements, like tools, menus, dark background at the same time as your softproofed photo.
This way same softproofed image with the same color coordinates looks less washed out/dull in comparision with all the other UI elements.If you turn “light theme” instead of dark on a high contrats display may give you something close to that. Softproofed image will look less washed out “in comparison”, but color coordinates on softproofed image will be the same.
Suroundings have an impact on how we percive colors.
If I do that, can I still use the “simulate paper and ink” (assuming I disable black point compensation)?
(I assume that you mean DisplayCAL’s profile “black point compensation”, not Photoshop or Lightroom’s black point compensation when softproofing)
Ink yes, but if display contrast is very close to paper’s DR, the effect of simulate black ink in may be tiny. 250:1 on glossy paper for example. Your display is almost there.
Paper usually no, because it will simulate paper white under the same illuminant used to make printer profile. That illuminant (D50 typ.) includes UV content so if your paper has strong OBA content, it may look bluer when you simulate it than viewing actual paper under an UV-less lamp.
Photoshop have separate settings for simulatimng paper and for ink. I do not remember LR right now, I’ll check.
It seems LR cannot separate these 2 options… less advanced than Photoshop.
CaptureOne/Affinity take other approach: it always simulates black when softproofing. I think that this a better approach than Lightroom’s.
Anyway, use another editor or live with this “feature”… as long as your papers do not have high oba content, paper won’t look too much blue.
Lowering the contrast is used to avoid seeing high contrast UI elements, like tools, menus, dark background at the same time as your softproofed photo.
This way same softproofed image with the same color coordinates looks less washed out/dull in comparision with all the other UI elements.If you turn “light theme” instead of dark on a high contrats display may give you something close to that. Softproofed image will look less washed out “in comparison”, but color coordinates on softproofed image will be the same.
An example (it’s a thought experiment, you do not need to do it : D ): 2x computers, 2x Dell U3017 same unifority, same white color with gains at RGB 255.
You calibrate & profile:
-set 1 to 5500K daylight 100nit, 0.4nit black ~250:1, single curve matrix profile WITHOUT black point compensation, so it will store actual black level on ICC profile
-set 2 to 5500K daylight 100nit, maximum native contast 800:1 or higher, single curve matrix profile WITH black point compensation, so it will store a false black level with 0nit on ICC profile
Calibrate & profile, install profiles, open a color managed editor on both computers. Softproof an image using some matte (L15* black) or glossy paper (L* 8 black) of your choice. Same zoom on image.Now you cover both screen with a card board with a rectangular opening of the size of the image.
Both displays, covered with cardborad excluding a hole for the images and only that. Both should show the same image with same color, same precieved contrast and simulated black (this assumes no user error in the process).
Now you remove the card board. On display with full contrast if it is using some dark theme, once you see the dark elements in UI, the softproofed image may look dull or washed in comparision, while on low contrast display dark theme UI elements will look closer to image. The brighter the black is on paper profile and the darker is the UI theme, the more strong is this effect by surroundings.
If you use a ligther UI theme on your editor the less pronounced is this effect… but you may find more confortable the dark UI theme, so lowering the contrast is the solution.
