Warnings when using an icc profile to pre-condition a higher quality targen command
Hi I have been experimenting with ArgyllCMS using an X-rite DTP-70 and ColorPort for reading charts. Converting the targen ti1 file use as an input into ColorPort is a bit fiddly, but, especially for larger target numbers makes reading targets a breeze. Definitely an acceptable tradeoff. With this approach I have produced some quite adequate profiles. My next step is to use targen with a "pre-conditioning" icc profile as a "-c" input. However, in doing so, the targen processing produces a huge number of warnings about "vertex location solving failed ..." and "Unable to locate vertex at node comb..." (see details below). My question is do these warnings have an effect on the quality of the resulting "ti1" file? I don't get these warnings if I use "-c sRGB.icm" as a switch. I am hoping to understand what elements increase the gamut volume and this is my current line of experimentation. Grateful for comments. Also, happy to document my method of converting a ".ti1" file for input into ColorPort if anyone is interested. Best wishes Rob Wignell ========= Roberts-MBP-2:Argyll rob$ targen -v2 -d2 -G -f 1728 -c PrinterPre.icc PrinterA iRGB test chart White patches = 4 Black patches = 4 Full spread patches = 1728 Degree of adaptation: 1.000 Adaptation weights: Device = 0.350, Perceptual = 0.650, Curvature = 1.000 Perceptual cache resolution = 13 Seeding cache...done Acceleration grid res = 13 There are 2 unique fixed points to add (8 total fixed points) There are 1720 far spread points to add Added 23/1722targen: Warning - vertex location solving failed after 41 tries targen: Warning - Unable to locate vertex at node comb 26 21 14 1 Added 45/1722targen: Warning - vertex location solving failed after 41 tries targen: Warning - Unable to locate vertex at node comb 45 25 23 3 . . . targen: Warning - vertex location solving failed after 41 tries targen: Warning - Unable to locate vertex at node comb 1190 796 433 34 It 20: Maxmv = 0.001434, MinPoint = 6.946, Min = 4.463, Avg. = 6.835, Max = 8.124, 0.5 secs. After optimization: MinPoint = 6.946, Min = 4.463, Avg. = 6.835, Max = 8.124 Total number of patches = 1728 Execution time = 9.998820 seconds
Rob Wignell wrote: Hi,
My next step is to use targen with a "pre-conditioning" icc profile as a "-c" input. However, in doing so, the targen processing produces a huge number of warnings about "vertex location solving failed ..." and "Unable to locate vertex at node comb..." (see details below).
typically this is targen struggling with non-linearity in the space. Doing adaptive Voronoi type tessellation in a non-linear/non-cartesian space is more of a heuristic than an exact mathematical algorithm, and if the space is sufficiently badly behaved, you get complaints like the above.
My question is do these warnings have an effect on the quality of the resulting "ti1" file?
They are a hint that there could be problems. You should be careful in evaluating whether using such a preconditioning profile is beneficial or not. Another thing you can try is to increase the smoothing when making the preconditioning profile (colprof -r parameter).
I am hoping to understand what elements increase the gamut volume and this is my current line of experimentation.
I doubt much to do with the profiling process will increase gamut volume, bar the TAC. You need to change operation of the printer and/or driver itself to change gamut volume - i.e. changing ink channel maximums, type of ink, screening patterns etc. Cheers, Graeme Gill.
participants (2)
-
Graeme Gill
-
Rob Wignell