[argyllcms] Re: Dell U2711 and i1Display Pro - ArgyllCMS vs i1Profiler

  • From: Graeme Gill <graeme@xxxxxxxxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Sun, 25 Nov 2012 12:08:37 +1100

Tom Schumm wrote:
> Now here's a new wrinkle... Both i1Profiler and ArgyllCMS seem to be 
> producing 
> pretty accurate LUT profiles, but the one from i1Profiler looks a lot 
> smoother 
> in Adobe Camera Raw and Photoshop, whereas the ArgyllCMS one has some pretty 
> ugly banding. Is there anything that can be done to get a smoother result? 
> Use 
> a larger testchart? Smaller? Add some other magic ingredient?


It's very hard to know what to suggest, since your observations
are for situations with unknown and (for me difficult to reproduce)
workflows. Which tables are being used in which exact profiles ? A2B ?
B2A ? What intent ? As input or output profile ?

To pin down anything like this it needs to be reproducable
(and I don't have access to Adobe Camera Raw or a modern
 version of Photoshop). So an example workflow of an image
colorspace conversion using cctiff is is something concrete
enough to investigate.

So, assuming for the moment that this is in relation to
the B2A table of the monitor profile, to reproduce such a problem
I would need:

        The .ti3 file used to make the display profile.
        The colprof options used to make the display profile.
        The source sample image used in revealing the banding issue
        The source colorspace profile used to interpret the sample image
        The cctiff options used to transform from the source space
        to display space using selecting the intent that reveals the problem.

I'm not aware of any particular banding issues - the code goes to a lot
of trouble in trying to produce profiles with the fewest possible artefacts
due to limited cLUT table resolution, but that doesn't mean that it always
works as intended, but to understand why, I need a concrete example.

Graeme Gill.

Other related posts: