Notice of Meeting - ACES Gamut Compression Implementation VWG - Meeting #8 - 5/20/2021

Recap:

  • We gave an update to the implementation TAC today, all went well. The only actionable request was for visual documentation, before/afters, etc, showing that where you apply the compression matters.
  • @jzp : the point about how all of this fits together brought up by @CClark was a good one - the working groups do need to talk to each other and make sure interop can be successful.
  • OCIO 2.1 Implementation:
    • @doug_walker : I thought at first that only the hard coded parameter version should have the ACES name on it, but now I’m wondering if that is realistic - people are going to use both.
    • @matthias.scharfenber - if they’re both ACES, people might use them interchangeably and cause confusion. If it has ACES in the name, there should be no doubt of what it is.
    • @LarsB : Agree. We need to be clear on what is ACES and what is not.
    • @Alexander_Forsythe : Parameterized version feels like a power user tool. Fixed version is to avoid confusion, provide a vetted set of defaults.
    • @doug_walker : but the reality is that there’s already the Nuke, DCTL, etc implementations - these things need a name. If we don’t name it, folks will call it ACES anyway.
    • @sdyer - this convention happens with the SSTS. It’s an algorithm with a default set of parameters. If someone modifies it, it’s still ACES.
    • @michaelch : it’s very important that we keep this split. For example, we exposed the parameter for the mid-grey in the ODT, and to be honest that’s not tracked very well. We need to be very clear.
    • @matthias.scharfenber : yes, that’s exactly what what we’re thinking. Once you start tweaking things, it’s not a reference transform anymore, it’s a creative grading operation.
    • @nick : And the transform ID tracks a fixed version via the AMF - how would you indicate a custom transform in an AMF?
    • @Alexander_Forsythe : You can reference a transform ID, in your own namespace, not publicly available, but the production can use it. You can also use a file path reference to a custom CLF or something too.
    • @sdyer : Also, our current transform ID naming convention doesn’t cover all parameters, only some - would need updating.
    • @doug_walker : Should there be three names? ACES Reference, version where you’ve just changed the power curve parameter, and a version where you change whatever you want.
    • @carolalynn : To summarize: seems like we agree that the Reference will have a specific, consistent name. We also agree that this parameterized version needs a name. We don’t agree on whether the parameterized version should have ACES in its name.
    • @matthias.scharfenber suggested something to start from, though all agree it’s too long and not marketable: Immutable Version: ACES RGCT (Reference Gamut Compression Transform) , Creative Version with Parameters: AMPAS VTGCT (Variable Threshold Gamut Compression Transform)
    • @doug_walker : Should get input from the community, without the OCIO PR in the picture, and then make a decision for all implementations.

Recording and Transcript