>>> My question is, how can I "instruct" lcms to use the B2Ax tables instead?
>
> This is a feature I miss in LCMS too - as well as a way to tell the 
> transform not to use any LUTs at all.

Would be nice, but unfortunately it goes against the spec,

please see ICC.1:2004-10:

"8.10 Priority of tag usage
There are several methods of colour rendering that can function within a
single CMM. If data for more than one method are included in the same
profile, the following selection algorithm shall be used by the software
implementation.

For input, display, output, or colour space profile types, the priority
of the tag usage for each rendering intent shall be:
1.BToA0Tag, BToA1Tag, BToA2Tag, AToB0Tag, AToB1Tag, or AToB2Tag
designated for the rendering intent
2.BToA0Tag or AToB0Tag
3.TRCs (redTRCTag, greenTRCTag, blueTRCTag, or grayTRCTag) and
colorants (redMatrixColumnTag, greenMatrixColumnTag,
blueMatrixColumnTag) The available valid tag with the lowest
priority number defines the transform."

That means that any CMM bypassing this priority is
misunderstanding the profile and using it in a way it was
never designed to work. I have seen some CMM claiming to do
color management and then only using shaper and matrix on
CLUT-based profiles. This is clearly wrong and in the best
case it may result in mixed intents. Note that a profile
creator may choose to use the matrix-shaper for other
functionality, for example a profile having matrix-shaper
and AtoB0, when properly handled will  use the AToB tag
for device to PCS and the inverse of matrix for PCS to
device. That does not mean a CMM could safely use the
matrix for device to PCS.

Anyway the topic is interesting enough.

Regards
Marti


> So more control flags would be nice. (At the moment I'm creating a 
> profile copy and deleting the LUT tags....)
>

------------------------------------------------------------------------------

_______________________________________________
Lcms-user mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/lcms-user

Reply via email to