2025-03-03 TSC Meeting Notes

Color Interop Forum

• Provided a brief review of last week’s meeting on OpenEXR

• Please see the #color-interop-foum ASWF Slack channel for more or watch the recording.

SMPTE CLF

• The SMPTE drafting group has published the draft of the Common LUT Format spec for public review and comment.

• https://github.com/SMPTE/st2136-1/

• It's on the roadmap for OCIO 2.5.

ACES 2 library progress

• Cuneyt’s fixes to the CPU and GPU paths have been merged into the aces2_optimization branch. The unit tests are passing. This is now considered a release candidate and people are requested to test it. The test config on the Wiki has been updated: https://github.com/AcademySoftwareFoundation/OpenColorIO/wiki/ACES-2.0-optimization

• Doug reached out to Eric to test the HLSL output.

• Doug posted the results of his testing of the Inverse → Forward round-trip on the cube-faces test targets on the #aces2-wg channel.

• Compared the following:

  • OCIO 2.4.1 CPU

  • OCIO release candidate CPU

  • OCIO release candidate GPU

  • CTL

• The differences between all of these is generally within about 1% on this test but there are occasional larger differences (e.g. 7%). Some examples were shared in Slack. They seem to be roughly what one would expect given the adjustments to the algorithm, but it would be appreciated if people could help review.

• There is a 1% noise level between all the comparisons in places where one channel approaches 0. Doug asked if this could be due to the fixed number of iterations in the inversion. Kevin said this is only an issue above a certain point related to the cusp. Given that the noise is showing up both above and below that, it is probably something else.

ACES 2 Display P3 HDR CTL

• Doug asked what the intended normalization of 1.0 for the 1000 nit Display P3 transform.

• Mark said he has done preliminary testing in Nuke and it is giving at least roughly the intended output.

• Kevin will take an action item to review the CTL.

SDR views for HDR displays

• Doug asked about which SDR view to add to certain HDR displays. The Rec.709 SDR view is currently being added to the HDR Display P3 and ST-2084 P3 displays, is that the best one to use? Perhaps the P3 SDR would be more appropriate?

• The group decided to continue with the Rec.709 SDR views. The feeling seems to be that it is more helpful to see the smallest anticipated gamut, rather than the native gamut of the P3 displays.

ACES 2 Config Generator

• Thomas has made several revisions based on feedback received in the GitHub issue: https://github.com/AcademySoftwareFoundation/OpenColorIO-Config-ACES/pull/130

• The active_view list is still not quite in the correct order. Thomas took an action item to adjust that.

• People are requested to please test these and provide feedback.

• We are not sure if this will be included as a built-in config or only as a separate download.

OCIO 2.4.2 release

• In addition to the ACES 2 implementation, several additional PRs are planned for inclusion:

  • PR #2112: Fix for OCIOZ absolute paths

  • PR #2120: Increase to recommended Imath version

  • PR #2121: Fix for OSL matrix handling

• Doug: If anyone wants other PRs included, please let us know.

• Remi: I may do a fix to add a ConfigIOProxy call to the Python binding.

• We will use the new OCIO Project Meeting next Monday to review the status and then plan to release OCIO 2.4.2 on Tuesday the 11th.

Adding built-in display color spaces that don’t clamp negatives

• Doug: This could be useful in workflows that need to generate SMPTE range video that includes extended range below black. Any opinions on adding these for use in future configs?

• Zach: I agree this would be helpful.

• Zach: How is this handled for ICC profiles?

• Remi: The parametric curve types are clamped, per the spec. More work would be needed to make those extended range.

  • https://github.com/AcademySoftwareFoundation/OpenColorIO/issues/1915

  • https://www.color.org/whitepapers/ICC_White_Paper35-Use_of_the_parametricCurveType.pdf

Daylight Savings Time

Daylight Savings Time in North America starts next weekend. As usual, other countries gradually switch over at a later date. Please double check meeting times over the next month.