[argyllcms] Re: tiffgamut 1.0.3

  • From: Klaus Karcher <lists@xxxxxxxxxxxxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Thu, 04 Sep 2008 11:42:39 +0200

Graeme Gill wrote:
Klaus Karcher wrote:
first of all: thanks a bomb for adding multi-TIFF and popularity filtering to tiffgamut. I love it by now while I didn't even test it!

I'm interested to know if the color popularity filter is useful
or not in creating image specific gamut mappings.

Me too. And so is my fine art repro client :-)

As far as I can estimate it until now, the differences are subtle and the "processing time penalty" is high, especially when processing sequences of images.

I have nearly finished a Droplet that greatly improves productivity and reduces the gateway hurdle for those unfamiliar with command line tools (it's a Mac application bundle that includes everything one needs). It facilitates batch processing, image- sequence- and source-profile-specific gamut mapping and maintains a device link profile cache. I'm not sure whether I should call it "AngeL" or "LucifeR"? ;-)

I ponder if I should also include a downsampling option (easy to implement with sips) as the performance advantage is considerable (e.g. tiffgamut with 5 roman16 images: approx. 4.5 ... 6 min, downsampled to 50%: 1:40 min incl. downsampling. The differences between downsampling to 50% plus tiffgamut -100 and tiffgamut -f70 without downsampling are nearly invisible. Also the differences in gamut shape and volume are negligible:

Intersecting volume = 202336.4 cubic units
'roman16_scaled.gam' volume = 211135.7 cubic units, intersect = 95.83%
'roman16_f70.gam' volume = 219275.9 cubic units, intersect = 92.27%

But I have to do much more testing before I can arrive at a conclusion.


Other related posts: