2021-08-17 at 21:44 #31287
So I got a Spyder3, calibrated things, etc… Aaaand yes, the white points now look the same, and uh, that’s about it for almost everything.
When I used GIMP to open it with specific color management profiles applied – why yes, they match now.
Why is color correction separated into two things? As far as I can understand, there is /some/ color modifications being applied directly now. I’ll call this Layer 1… but it’s not actually correcting the colors… And then some applications support a 2nd-layer of correction, that actually fully corrects it?
Terminology I’m using:
Layer 1 :
dispwin -c<– This effect. Calls it “Calibration”?
Layer 2: Whatever magic that some apps use and make the colors actually correct. “Color managment”?
So, Suppose it’s so that the 2nd layer allows applications to correct things better: – This allows you apply it before converting it to RGB to allow maximum range… But then the layer 1.. that is working on RGB… Well that will compress some of the range and extend the other areas, and is limited. (ex: 255 values per channel per pixel… To correct 222 to 223, well now either 223 is also 223, or everything above it also goes up by one, but then 254 and 255 are both 255 now)
So, okay, sure, 2nd layer allows some more of the range maybe… but not to the full extent as if it was controlling it all… And then all the normal apps looks worse. And everyone’s attempts at matching two displays to the same color via calibration… doesn’t work.
As far as I can tell, these are all either lookup tables or curves… so.. why.. not just merge them..? Can I do that? Where/How? Why wasn’t this done?
Who/Why made them separate?2021-08-18 at 5:30 #31288
This is technical discipline essentials. Color management is not dummy friendly, but it covers many cases in professional production. It comes out of printing and inherits some of its traits. And it has flexibility to morph current ICC profile state.
Here are established terms:
- your Layer 1 is called calibration for displays and linearization for printing. The second term is more clear. This stage is targeted to correct wrong contrast growth, but it also allows to change display white point (WP).
- your Layer 2 is called profiling or characterization, meaning measuring beahviour of non-linear system. But sometimes term “profiling” is used for whole ICC profile building process, including linearization. This stage determines color primaries (actual R/G/B corners of your screen gamut, their intensity and tint) and applies mixed colors correction.
Calibration is separated for the next reason: all real color reproduction systems are not stable. After some aging you need to correct color again. But in most cases color primaries are still the same, so you only need to correct contrast and white point. Good exaple is Xerox photoconductors (drums) burn: if you’ve printed 1000 Coca Cola posters, yellow and magenta drums has some wear, while black is almost the same and blue is untouched. Color toner and paper are the same. Note that custom display profiles includes calibration curves (they’re loaded in videocard, this is called vcgt – videocard gamma table, but beware: term “gamma function” has strict meaning in color management and it isn’t gamut). Factory profiles and standard profiles (like sRGB or Fogra 39 for press) don’t include calibration curves as they’re targeted to linearized devices.
The most sensitive thing in color management for novices is color saturation limits. “Display calibration” does not mean right contrast and tint only, color is also limited to image profile. By example, modern iMac screen (close to DCI-P3 standard display profile) can show very intensive red, but sRGB image red extreme point is much less saturated. At the opposite side, iMac screen can’t show deep cyan color of Fogra 39, so extreme cyan tints have to be clipped. Non-color managed applications don’t understand these limits, they don’t show color clipping, but they show wrong color.
P.S. Try to correct WP using display menu. If you can do it, you can avoid calibration (use native display gamma setting in DisplayCAL) for XYZLUT type profiles and get smoother gradients, but use ~1500 patches to get good quality. You can also use display menu correction together with calibration, in this case set native white point after you’ll achieve 6500K by hand tuning.
2021-08-18 at 13:54 #31293
- This reply was modified 4 weeks, 1 day ago by Алексей Коробов.
I’ll add another example.
Stage 2 is separated because I want to open an sRGB image, an AdobeRGB image and a ProphotoRGB image at the same time in the same application, change tabs between one image to another and show colors properly on each one (at least the colorspace covered by display).
Stage 1 & 2 get fused in applications where ther is only one working colorspace, like for example video editiong in Rec709, stage 1 and 2 fused in a LUT3D that transforms from Rec709 to my display colorpsace.