Re: [Openexr-devel] Introduction and CTL/ICC question

[Drew Hess]

Kai-Uwe Behrmann <address@hidden> writes:

> As I understood the CMS will work hardware accelerated on the GPU instead 
> CPU. Up to what size does CLUTs deliver reasonable performance there?


I'll leave it to others to answer your questions on the ICC spec
(except to say that there is a lot of work going on right now to
figure out how to make ICC profiles work in a "digital cinema" world),
but this one I can answer.

It depends on what you mean by "reasonable performance," but on an
NV30-class GPU, you can get ~20fps playback on HD-resolution 16-bit FP
RGBA frames using 3 full 1x65536 LUTs; that is, one LUT for each color
channel (R, G, B) that can transform any 16-bit half value to any
other 16-bit half value.

With more modern GPUs, you can get well in excess of 30fps.

Also, interestingly enough, if you use a more sophisticated color
transformation scheme, such as a 3D lattice encoded as a 3D texture,
you can get even better performance, but this requires NV40 or better
hardware, as earlier versions of hardware don't support 16-bit FP 3D
textures.  (Although you can do some dirty tricks to make this work
with 8/10-bit fixed-point 3D textures, too, if you're willing to live
with some minor differences between it and the FP 3D texture lookups.)


d

pgpDIxHzQlqSN.pgp
Description: PGP signature