[argyllcms] Re: Crushed shadow details

  • From: xun wang <xun911@xxxxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Fri, 9 Sep 2011 22:22:09 -0400

Thanks that works!

I did some some further testing and comparison, and noticed that with
ArgyllCMS I still get less shadow details than the profile created using
X-rite's OEM software(ICC v2). Using photoshop's curve tool, the
lowest discernible shadow details using x-rite profile is 2. With ArgyllCMS
profile it's 4. I don't see any noticeable color casts in the shadows by
either profile.
I also get less gamut volume with the ArgyllCMS: 1,114,180 vs 1,087.920

Below is the DispcalGUI screenshot for the settings I used, please let me
know if there're anything I haven't optimized. Because I would really love
to use ArgyllCMS for my main calibration tool for its powerful features and
flexibilty. But for now the x-rite OEM software seems to give superior
result yet taking a fraction of the time.

DispcalGUI setting screen shot: http://tinyurl.com/3vw7o9y

I also ran "Very Profile", I'm not entirely sure but it also seems to agree
that the x-rite profile is oeverall slightly more accurate?
x-rite profile verification: http://tinyurl.com/44eptzd
argyllcms profile verification: http://tinyurl.com/4yqu4km

Any help and inputs are greatly appreciated!


On Fri, Sep 9, 2011 at 2:54 PM, Florian Höch <lists+argyllcms@xxxxxxxxx>wrote:

> Am 09.09.2011 20:28, schrieb xun wang:
>  Is there anyway I could create a "single gamma + matrix" file for
>> Colormunki using the existing saved calculation data without running
>> through the "Calibreate & profile" process again? it's painfully slow
>> when I use "adaptive hires" + Black Drift Compensation for the Colormunki.
> Yes, you can use the existing ColorMunki data to create a new profile from
> it. (using the commandline tools, give the existing .ti3 as last argument to
> colprof. Or if you use dispcalGUI, set the desired profile type, then in the
> "Options" menu -> "Create profile from measurement data...", then select the
> .ti3 file)
> --
> Florian Höch

Other related posts: