Lars Tore Gustavsen wrote:
with a lut based profile. I can see all the black numbers in a not color aware viewer, but with my calibration loaded. If I view the same image with a perceptual rendering intent and a lut based profile it looks almost fine. A matrix profile also handles this well but the overall accuracy elsewhere are much porer.
It's not clear to me from your description what effect a calibration curve is having. Normally I'd expect a calibrated display to have good graduations from black. If not, then the display is behaving in ways that can't be compensated for, or calibration isn't working as expected.
So how should I deal with only relative colometric rendering? I tried to increase the quality with colprof to qh but to may surprise the situation is even worse. ltg@buntu:~$ seq 1 1 5| awk '{print $1,0,0}'| xicclu -fb -ir lut-qm.icc 1.000000 0.000000 0.000000 [Lab] -> Lut -> 0.003046 0.004410 0.003134 [RGB] 2.000000 0.000000 0.000000 [Lab] -> Lut -> 0.006078 0.008762 0.006242 [RGB] 3.000000 0.000000 0.000000 [Lab] -> Lut -> 0.009080 0.013078 0.009325 [RGB] 4.000000 0.000000 0.000000 [Lab] -> Lut -> 0.012060 0.017356 0.012382 [RGB] 5.000000 0.000000 0.000000 [Lab] -> Lut -> 0.015007 0.021578 0.015395 [RGB] ltg@buntu:~$ seq 1 1 5| awk '{print $1,0,0}'| xicclu -fb -ir lut-qh.icc 1.000000 0.000000 0.000000 [Lab] -> Lut -> 0.000000 0.000000 0.000000 [RGB] 2.000000 0.000000 0.000000 [Lab] -> Lut -> 0.000000 0.000000 0.000000 [RGB] 3.000000 0.000000 0.000000 [Lab] -> Lut -> 0.000000 0.000000 0.000000 [RGB] 4.000000 0.000000 0.000000 [Lab] -> Lut -> 0.007913 0.009346 0.006997 [RGB] 5.000000 0.000000 0.000000 [Lab] -> Lut -> 0.016811 0.019825 0.014771 [RGB]
Typically a display has a non-zero black, so it is not unexpected that if you ask for blacker than its black, the B2A table clips to the device gamut minimum of zero (or thereabouts). It is also not unexpected that this transition from within the gamut to being clipped is a (relatively) abrupt transition that is better represented by the hight resolution grid.
I wonder if there is other Argyll related tricks to handles this, or should I start asking all graphical software developers about implementing perceptual rendering intents?
The latter would be a good change in my opinion, since it provides a more sophisticated alternative to "black point compensation". Graeme Gill.