2025-12-22 TSC Meeting notes

December 8, 2025

Host: Doug Walker

Secretary: Doug Walker

Attendees: 

• Doug Walker (TSC Chief Architect) - Autodesk

• Kevin Wheatley (TSC) - Framestore

• Zach Lewis (TSC)

• Scott Dyer (AMPAS)

Apologies:

• Carol Payne

OCIO TSC Meeting Notes

• OCIO 2.5.1 Release

  • Doug: As communicated earlier, we will proceed with a 2.5.1 release the first week of January. PRs that need approvals are the following:

    • PR #2224: Avoid clamping on ICC profile linear and pure-gamma TRCs

    • PR #2226: Fix issue with texture index binding for Vulkan

  • Kevin: There is the array indexing problem for ACES 2 output transforms reported on the forum. I may have some time to take a look. Doug: That would be a great addition for 2.5.1, if there is a PR by our next TSC meeting.

  • Zach: How should we handle CICP in ICC profiles? Doug: Good question, won’t make it for 2.5.1, but we should discuss in the new year.

• Finalizing CIF Display Color Space Recommendation

  • Doug: Would like to finalize this at the January CIF meeting. Please review the document and suggest any changes. The only recent addition (in orange text) are the new linear display-referred spaces. Group: These are good additions.

• OpenImageIO color space handling

  • The group discussed the ongoing conversations in OIIO about color space handling.

  • There was agreement on Zach’s proposal to only write CICP for display-referred color spaces. Kevin: In full agreement. There are multiple reasons to take this approach:

    • CICP is, in a practical sense, a display-referred spec

    • Scene-referred color spaces cannot be viewed directly, need more information than just CICP to know how to display it as intended

    • At a library level, only want to do what is strictly correct and that we all agree on, namely that CICP should be written for display-referred spaces

  • One could reasonably argue (as Brecht has) that having CICP in scene-referred files is useful for basic interop in certain workflows. But the group felt that that sort of decision would be more appropriate for a specific app to take rather than having the library impose it on all apps.

  • There was discussion of Postel’s Law ("Be conservative in what you do, be liberal in what you accept from others") and Hyrum’s Law ("With a sufficient number of users of an API, it does not matter what you promise in the contract: all observable behaviors of your system will be depended on by somebody") and how these are relevant to this decision.

  • The group agreed that OIIO should allow someone to write CICP in scene-referred files, but there should be some extra step required, such as retagging with a display-referred color space or manually setting CICP values. It should not happen automatically.