[argyllcms] Re: perceptual black too light

Gerhard Fuernkranz wrote:

FYI, below is an example for the mentioned problem.
The example is CMYK. I also tried a RGB profile - and here I did not see the problem (B2A0 maps black to RGB=0) 

Unfortunately I don't have your dataset, so I can't really reproduce
this. If I try and duplicate this using a dataset I have (cromalin proof),
I don't see the problem:

profile -v -qm -r1 -c3 -d0 -l250 -kh -S sRGB.icm test

iccdump -v3 -t bkpt test.icm
XYZArray:Channels = 3
  No. elements = 1= 9
    0:  0.015549, 0.014771, 0.015457    [Lab 12.461413, 3.643509, -4.050391]

xicclu -ip -fb -pl test.icm
0 0 0
0.000000 0.000000 0.000000 [Lab] -> Lut -> 0.360593 0.000000 0.999295 0.995774 
[CMYK]

xicclu -ia -ff -pl test.icm
(Note it should be -ia to be comparable with the bkpt, which is an absolute 
value)

360593 0.000000 0.999295 0.995774
360593 0.000000 0.999295 0.995774 [CMYK] -> Lut -> 13.176778 0.069794 -0.790074 
[Lab]

Which seems about right given that we've asked for the darkest neutral point,
while the black point is the darkest, not necessarily neutral point.
The difference between the two will depend on the how white the white point
is on the device, and also how black the source profile black point is (I 
think).

xicclu -ir -fif -pl -l250 -kh test.icm
15.000000 0.000000 0.000000 [Lab] -> Lut -> 0.322992 0.075402 0.856541 0.993811 [CMYK] Lim 2.248745
14.000000 0.000000 0.000000 [Lab] -> Lut -> 0.306565 0.000000 1.000000 0.997427 [CMYK] Lim 2.303992 (clip)
13.000000 0.000000 0.000000 [Lab] -> Lut -> 0.308843 0.000000 1.000000 1.000000 [CMYK] Lim 2.308843 (clip)

Note for absolute:

xicclu -ia -fif -pl -l250 -kh test.icm
15.000000 0.000000 0.000000 [Lab] -> Lut -> 0.287097 0.084209 0.687339 0.986873 [CMYK] Lim 2.045519
14.000000 0.000000 0.000000 [Lab] -> Lut -> 0.324056 0.053279 0.889522 0.992856 [CMYK] Lim 2.259713
13.000000 0.000000 0.000000 [Lab] -> Lut -> 0.301464 0.000000 1.000000 0.996532 [CMYK] Lim 2.297996 (clip)
12.000000 0.000000 0.000000 [Lab] -> Lut -> 0.297574 0.000000 1.000000 1.000000 [CMYK] Lim 2.297574 (clip)

All pretty consistent, so I'm not sure if I've fixed something, or the problem is only triggered by certain devices. Perhaps you can mail me the .ti3 file you use ?

Graeme Gill.

Other related posts: