[argyllcms] Re: A possible problem with Colorprof and XYZ LUT profiles
- From: Marti Maria <marti.maria.s@xxxxxxxxx>
- To: argyllcms@xxxxxxxxxxxxx
- Date: Thu, 16 Dec 2010 16:42:14 +0100
Hi,
Just FYI, I have spotted the bug source. It is not on the Argyll
profile, nor in the
lcms2 absolute colorimetric code: it is in the source profile used by
the application
to model HDTV. The profile, a V4 one, had ill-formed CHAD. Such low
media white
point made D65 destination profile look like D93.
Best regards
Marti Maria
On Thu, Dec 16, 2010 at 8:15 AM, Kai-Uwe Behrmann <ku.b@xxxxxx> wrote:
>> Date: Wed, 15 Dec 2010 17:37:46 +0100
>> From: =?ISO-8859-1?B?SuFub3MsIFTzdGggRi4=?= <janos666@xxxxxxxxxx>
>>
>> I had a conversation with Marti Maria about a possible lcms2 bug (possibly
>> broken absolute colorimetric rendering intent).
>> He asked me to send my profile. Here is the package:
>> http://www.tar.hu/janos666/abstest.rar
>> It includes my spectral measurement data (and a spec2cie converted version
>> from the ti3 file), my colorprof settings and the resulted ICM file.
>
> The profile is indeed ~D64. As well I tested the conversion with lcms2
> actual git and its behaves like D64 with a absolute colorimetric intent.
> I found no visible whitepoint adaption for D65[e.g. sRGB]->D64[your_profile]
> conversions, as expected.
>
>> He inspected my problem and his judge was that lcms2 works fine because my
>> display has D93 white point. (And the very brownish result is correct for
>> my
>> display.)
>
> For D93 that assumtion would be correct. But with the profiles from the rar
> archive, I do not see a D93. So it might be a renderer issue.
>
>> But it's impossible, because this is is a factory calibrated display. It
>> can't be so far from D65.
>> According to the spec2cie converted measurement data, my white point is:
>> X=93.943
>> Y=100.00
>> Z=105.17
>>
>> So, why does it look like that my WP is D93?
>
> Can you make sure to use lcms from actual git in your renderer? tificc would
> be helpful to track things down.
>
>> I have no idea, except that there is a bug in ArgyllCMS / colorprof. (I am
>> using v1.3.2)
>> As you can see, I used my original ti3 file, the spec2cie result stands
>> for
>> debugging purposes only. And I didn't use any tricky or strange build
>> settings, I think.
>
> kind regards
> Kai-Uwe Behrmann
> --
> developing for colour management www.behrmann.name + www.oyranos.org
>
>
>
--
Marti
Other related posts:
