2024-05-13

May 13, 2024

Host: Carol Payne

Secretary: Carol Payne

Attendees:

Apologies:

• Michael Dolan

• Sean Cooper

• Mei Chu

OCIO TSC Meeting Notes

• ACES 2.0 planning

  • First step - Update Spreadsheet & add stubs for new Built-In Transforms

  • For future: hard to support both ACES 1 and ACES 2 at the same time - it's basically two different repos with all the changes. Will need to talk about what we prefer in a bit. 

  • Remi is trying to catch up with the meetings of the ACES OT group he missed, and update his python version with the changes

  • Then will start on the OCIO code, hopefully will have an update on that for the next TSC meeting

  • Also for next meeting, start talking about what parameters to expose - work to make a spreadsheet with the starting points from Kevin and Mark R and go from there

  • Carol will work to get a few meetings with the ACES folks scheduled to stay in communication

• NanoColor update

  • Working group meetings starting next Monday at 1pm PT, will be weekly. Will be added to the OCIO calendar. If you're interested in this initiative, please try to attend - we don't have time in the TSC meeting to get through all of these discussions. We'll try to provide high-level updates at one of the next few TSC meetings. 

• Adding external references to color spaces (Cody Cuellar)

  • https://github.com/AcademySoftwareFoundation/OpenColorIO/issues/1975

  • Hard to embed metadata to rendered images (specific example is display colorspace) using OCIO currently

  • Thoughts around where to look / decide what ICC profile to use - would need standard places to look for the built-in configs, as there are no LUTs / search path defined, as they're all built-in 

  • OR - define an OCIO way of saying what this file is, and leave it to the application to link that to a certain profile/tag etc 

  • Kevin - we've talked about chromaticities before - but we are talking strictly about displays for this use case?

  • Cody - could be used in non-display as well, but primary case is display for viewing 

  • Carol - I like the idea of OCIO providing what it is and the application going from there

  • Cody/Kevin - yes, but we need a way to be specific and say exactly which one we want as a config author

  • Carol - philosophically, OCIO is a separate color management system from ICC. Not sure we'd want to go down the path of embedding ICCs (built in) to the library - but providing a way to get from OCIO to ICC in a way the user wants seems logical

  • Kevin - either you have a blob describing the colorspace than OCIO knows nothing about, or we go a step further and be more explicit about it is we know this is rec709, so we can be more precise about it

  • Cody - ideally OCIO doesn't need to know much about it, just provide APIs to get at the metadata according to what is defined in the config/built in colorspace

  • Thomas - we do already ship ICC profiles with the configs - but those are not what we want here, they have the view baked in

  • Kevin - we could have a side utility that mostly only deals with the metadata portion of the configs? Keeps version management easier etc

  • Remi - also Mark B's use case around display colorspaces 

  • Carol - please add comments and use cases to the issue, and we'll talk about it again at another TSC meeting. Starting point might be what we agree we're NOT going to do. 

• Standard texture color spaces (ACIP_core-texture-color-spaces)

  • Please take a look at this and provide feedback before May 20, at which point we will have a draft list that will hopefully be close to final. Know this is a tight turnaround, but we're trying to meet some tight timelines for the MaterialX release at the end of June.