2022-02-01

Owned by Michael Dolan
Last updated: Feb 01, 2022
2 min read

February 1, 2022

Host: Michael Dolan

Attendees: 

Carol Payne (TSC) - Netflix
Doug Walker (TSC Chief Architect) - Autodesk
Kevin Wheatley (TSC) - Framestore
Michael Dolan (TSC Chair) - Epic Games
Thomas Mansencal - Weta Digital

OCIO Config Working Group Meeting Notes

  • Studio config:

    • Doug: Code I sent to Canon should simplify the problem a lot. Generates CLF with Python. They need to review for metadata and LUT accuracy. We could post some non-vendor-specific example code like this somewhere for other vendors, to show how to generate it.

    • Carol: Will give it a couple weeks and can check in. Would be great to include Python snippet in materials we're sending to other camera vendors.

    • Doug: Been having similar conversation about what is the raw material for these transforms. Directory of CLFs and Python to add those to config. Do we include Python script to generate those CLFs, or have CLFs be source of truth? 

    • Michael: No objection to not including Python source if it's just replicating the same data.

    • Carol: Agree. Having CLF be source of truth sounds good. 

    • Doug: In Canon's case, it's a LUT, so having the code is good for visibility.

    • Michael: Are colour-science matrices hard coded? Or generated from chromaticities?

    • Thomas: Depends on the situation. Some have different implementation. Published matrices are all over the place. Think it's good to have code, so you know where things are coming from. If there's a problem, or a transform changes, etc.  Think it's good to have what generates it somewhere.

    • Proposal: Include CLF source code where it gives insight to the implementation or source of data, but not where it's redundant.

    • Carol: Draft of mail for camera vendors that already have ACES IDTs, that we haven't talked with yet: https://docs.google.com/document/d/1cg8nqUVIUeY-E2c7yvR-3I_EuWFF2FsRm7xiNVHs3KI/edit

      • Will hold on doc for a day or two before sending.

    • Michael: Benefit of having contribution from vendor, is it traces back math to them, so people know its correct.

    • Carol: Going to follow up with Blackmagic team, in case they have any questions.

  • CG config:

    • TODO: Michael will submit PR with Doug's CLF source code and prep release after merge. We can do draft release to review config.

  • CLF Transform ID:

    • Doug: Comparing aces-dev transform ID version string and our transform ID version string. Might need to clarify our versions for vendors.

    • Carol: Scott is working on some changes to this on the ACES side. There's a planned update to that. Gamut compression transform is published under the new spec. Not super different. Makes the numbers clearer.

    • Doug: Would be helpful in our doc to clarify difference between the two versions, since we're asking contributors to fill in this string.

    • Doug: ACES transform ID has a direction in it, where ours does not have a direction. Realized when populating metadata, was making two CLFs and putting same CLF ID in both.

    • Michael: Output descriptor is always ACES2065-1 currently, so only one direction.

    • Doug: Would be clearer to have the direction, in case we want to implement output transforms later too, that we want user contributions for.

    • Michael: Can we use ACES instead of ACES2065-1 in name for direction?

    • Doug: Or maybe note input or output transform?

    • TODO: Michael will clarify version numbers in CLF transform doc

    • TODO: Update IDT transform IDs in studio config to include direction