2026-06-22 TSC Meeting notes

Host: Doug Walker

Secretary: Doug Walker

Attendees: 

• Doug Walker (TSC Chief Architect) - Autodesk

• Kevin Wheatley (TSC) - Framestore

• Remi Achard (TSC) - DNEG

• Mark Titchener (TSC) - Foundry

• Zach Lewis (TSC)

• Michael Dolan (TSC) - Epic

• Cuneyt Ozdas (TSC) - Autodesk

• Scott Dyer - AMPAS

Apologies:

• Carol Payne

• Sean Cooper

OCIO TSC Meeting Notes

• New PR to update CIF Texture Asset Color Space Rec. to v1.1

  • https://github.com/AcademySoftwareFoundation/ColorInterop/pull/16

  • Adds g24_rec709_scene for compatibility with OpenUSD and MaterialX

    • Kevin/Zach: No objection to add it as long as it is clear it is for backwards compatibility and not recommended for new assets. Remi: Would like more detail on its use in USD.

  • Adds bypass, per Zach’s proposal

    • https://github.com/AcademySoftwareFoundation/ColorInterop/issues/15

    • Zach: Since this adds some complexity, I’d like to make sure we’re all on the same page before adding it.

    • Remi: Is sample code for how a file format reader should handle this available? Zach: Nothing has been shared at this time.

    • Sounds like this needs further discussion, we will pick it up again at the next meeting.

• PR for the CIF OpenEXR Recommendation

  • https://github.com/AcademySoftwareFoundation/ColorInterop/pull/13

  • Doug: If anyone has any remaining discussion topics, now would be the time. All known issues have been addressed and this will be merged in the coming weeks.

• PR for the CIF Interop ID Recommendation

  • https://github.com/AcademySoftwareFoundation/ColorInterop/pull/11

  • Doug: This was substantially revised based on the Color Interop Forum meeting last week and is awaiting review. All known issues have been addressed.

  • Doug reviewed the changes since the previous draft, especially the proposed OCIO implementation.

    • Kevin: Would the name searches match roles? Doug: Yes, since they use Config::getColorSpace.

    • Kevin: When a name is sanitized, is it checked to ensure it’s legal in the config? Doug: Yes. Kevin: Should ociocheck prevent duplicates after sanitization? Doug: I think ociocheck should validate the name attribute of the config, but would not need to prevent collisions after sanitization.

    • Kevin: Is alias order significant now? Doug: Yes, the first acceptable one would be used as the local name, when generating an ID. But ideally config authors would set the interop_id, in which case alias order would not matter.

    • Kevin: Would we need a cache for the situation where an app may be looking up IDs for thousands of texture assets? Doug: It’s a valid concern, to be determined if it’s needed. One workaround would be for the app to set the interop_id in its in-memory instance of the config, based on the first search.

  • Action for group: Please provide feedback on the PR and proposed OCIO API additions

  • Action for Doug: Prepare a PR with implementations for the OCIO functions

• Siggraph planning

  • ACES / CIF / OCIO have a three hour block during Monday’s Birds of a Feather track

  • Any ideas for presentations? Know of anyone doing cool OCIO stuff we could invite?

• Next week's working group meeting proposed topics:

  • HDR display handling

  • Single-string color transform syntax

• Color temperature controls

  • Doug: I’ve been looking at the color temperature control in USDlux and the ways it's implemented in the OpenUSD code. There is a slight pink cast at 6500 due to the fact that D65 is not aligned with the Planckian locus.

  • Doug presented a diagram illustrating various possible approaches to improving this and asked for feedback.

  • Kevin: The IES TM-30 option does not look good due to the slope discontinuity at D50.

  • Remi: I’m not sure the schema specifies an algorithm, so it’s probably up to how each renderer will implement it. Doug: Exactly, I was looking at the hdEmbree and hdPrman implementations in the OpenUSD code that leverage the newer GfColor/NanoColor along with an earlier function named UsdLuxBlackbodyTemperatureAsRgb. All of them have the pink issue. Would be nice to have an improved recommended algorithm.

  • Remi: The other aspect that is a source of variability is how the luminance normalization is done. Doug: Agreed.