/
2025-12-08 TSC Meeting notes

2025-12-08 TSC Meeting notes

December 8, 2025

Host: Doug Walker

Secretary: Doug Walker

Attendees: 

  • Carol Payne (TSC Chair)

  • Doug Walker (TSC Chief Architect) - Autodesk

  • Kevin Wheatley (TSC) - Framestore

  • Thomas Mansencal (TSC) - Epic Games

  • Cuneyt Ozdas (TSC) - Autodesk

  • Remi Achard (TSC) - DNEG

  • Sean Cooper (TSC) - Skydance Animation

  • Zach Lewis (TSC)

  • Scott Dyer (AMPAS)

Apologies:

  • Michael Dolan, Mark Titchener

OCIO TSC Meeting Notes

  • Gaussian Splats Color Spaces:

    • Doug: USD and Khronos are developing schemas for gaussian splats and plan to be able to identify the color space used in the file. The splats generated by the open source tools are currently difficult to insert into a typical rendering pipeline because they are trained in a video color space and the blending must be done in a video color space for it to look as expected.

    • Doug: In one of the threads there was mention of wanting to use a scene-referred version of PQ. Group: That does not sound like a good idea, would be confusing.

    • Kevin: There is a recent fxguide podcast that describes a project where we used splats.

    • Doug: On the USD side, they will be using the USD ColorSpaceAPI in the schema. I proposed that Khronos also consider using the same Color Interop Forum color spaces, so there is commonality. Thomas: That would be great.

  • Interop IDs:

    • Doug: I’m worried that we will start seeing people make up their own interop IDs, just by looking at the pattern of the existing ones (transfer-func_primaries_image-state) and not reading the interop ID spec and therefore not realizing a namespace is needed. This is probably unavoidable, but something we should try to discourage. At least ociocheck will warn if this occurs.

    • Thomas: We should have a place for people to post and validate their IDs. Doug: In the current draft spec it says we will set up a Wiki for people to register namespaces and IDs.

    • Sean: We could set up a GitHub issue template so that people could register them that way.

    • Zach: Would be helpful to steer people towards Bradford chromatic adaptation when creating new IDs for better consistency with ACES and ICC.

  • Contribution offer to improve API web theme:

    • Doug: A contributor has kindly offered to implement issue #1872 and upgrade the API docs Sphinx theme. It looks like they used the same theme used by OIIO in a similar effort for another project. Are people ok with using that theme, should he proceed?

    • Carol, Sean, Kevin: Yes, sounds good!

  • OpenImageIO recent color space PRs:

    • Zach updated the group on the recent OpenImageIO PRs from Brecht and the discussion in the color workspace thread.

    • Zach: Latest thinking about handling CICP is as display-referred. We would only write it for display-referred versions of a color space. Carol: Makes sense, the CICP have mostly been used in display-referred formats such as video.

    • Kevin pointed out that the recent revisions to H.273 have added language that brings in mention of the EOTF to the existing OETF formulas but it is still quite unclear as well as inconsistent among the various transfer function options. In addition, some are mirrored and some are not.

    • The group agreed that it would be best to handle CICPs as being display-referred but OIIO should provide a way of overriding this if needed.

    • Doug: OIIO could potentially have a global preference setting that would allow changing the interpretation.

    • The group discussed the various options including using file naming conventions and ICC profiles.

    • Zach: What about using ICC profiles to tag scene-referred color spaces? Doug: I worry about that because ICC is primarily used for display-referred color spaces and uses a display-referred profile connection space. Tagging from other apps would likely be in a display-referred context.

    • The group discussed the relative priority of metadata from the file header vs. file naming. Currently OCIO File Rules are only based on the file naming.

    • Sean, Carol: It would be good for the Color Interop Forum to make recommendations along those lines that OIIO, and other tools, could then adopt. Group: Agreed. Next step could be further work on Sean’s Mermaid diagrams from the OpenEXR draft recommendation.

    • The group discussed maximal vs. minimal approaches regarding what metadata to set. Kevin: Best advice is to be able to read anything but on writing only set what is considered best-practice.