January 25, 2021

Host: Michael Dolan

Attendees:

• [X] Mark Boorer (TSC) - Industrial Light & Magic

• [X] Mei Chu (TSC) - Sony Pictures Imageworks

• [X] Sean Cooper (TSC ACES TAC Rep) - ARRI

• [X] Michael Dolan (TSC Chair) - Epic Games

• [X] Patrick Hodoul (TSC) - Autodesk

• [ ] John Mertic - Academy Software Foundation / Linux Foundation

• [X] Carol Payne (TSC) - Netflix

• [X] Mark Titchener (TSC) - Foundry

• [X] Carl Rand (TSC) - Weta Digital

• [X] Doug Walker (TSC Chief Architect) - Autodesk

• [X] Kevin Wheatley (TSC) - Framestore

• [X] Bernard Lefebvre - Autodesk

• [X] J Schulte - Industrial Light & Magic

OCIO TSC Meeting Notes

• Move apps to OpenColorIO-Apps repo? Or a separate CMake project?

  • J: Yes, break them up.

  • Kevin: Suggest splitting CMake instead of using separate repo.

  • Michael: Separate repo would create version dependency between them.

  • Patrick: Some apps make sense to keep with core, but strange for some to be out of repo.

  • Sean: Good to distinguish between OCIO apps and utilities. Utils include ocioperf, ociocheck, etc. Apps like ociodisplay, ocioconvert, etc.

  • Patrick: ociodisplay more of an integration example.

  • Sean: ociodisplay and ocioconvert are useful for debugging CPU/GPU renderer and finding cause of artifacts.

  • Michael: If we split within repo, do we have two builds or just an apps option, like we have now but defulting to OFF.

  • Mark B: CMake doesn't do well with finding internal dependencies. Won't see the installed version of lib. Will try to rebuild and link against it. Benefit of separate repo is no cross-dependency between app and lib. Could be in one repo, and change dependencies of apps to depend on new installed target. Needs to be done in two steps.

  • Patrick: Becomes an additional process for development.

  • Mark B: Could write Cmake module to find lib if it exists locally.

  • Carol: Or back to what Michael was working on, swapping in OpenEXR.

  • Mark B: Since half moving out of OpenEXR, can move to use Imath to get half instead. Can then remove OpenEXR depenency from OCIO core.

  • Sean: Why does ocioperf depend on OIIO. Can it just generate noise image? Worth looking into. Makes more sense for consistency. Can create all data types.

  • Patrick: Yes, goal of tool to measure color transformation, not image IO. Could generate in memory image.

  • Doug: Performance can change with image IO because of cache usage.

  • Sean: Clean to have utils be internal projects that don't depend on third-party libs, and apps to depend on third-party. Might invite devs of other apps too.

  • Patrick: ocioperf generating live image based on critereon could be good work to do sooner.

  • TODO: Michael will update issue with recent discussion points. Continue offline discussion to agree on an approach proposal.

• ASWF 2020 CCLA 2.1 transition:

  • Michael: Propose to switch to this CLA, but with a more gradual transition. Start getting them signed over next two months, and then flip switch once most contributors are covered.

  • All sounds good.

  • Doug: Brings us closer to original ASWF premise. Simplify contributions.

  • Patrick: No problem, but suggest no changes for first four weeks of OCIO v2.

• Release discussion:

  • Michael: Thanks to Doug for adding remaining ACES builtin transforms over the weekend! Aweomse work. Ok with calling OCIO v2 done?

  • Doug: Think everything is ready. Delayed as long as we can. In CY2021 now. It's time to move forward.

  • Multiple TSC members and others in community have been testing and using it prior to release. Agree it's ready to ship. Exciting!

  • Huge thanks for Autodesk, Doug, Patrick, Bernard, TSC, and community. Awesome accomplishment that will have great impact on our industries!

• 2021 leadership thread:

  • Michael: Reminder to TSC members to check out the maillist thread on TSC leadership for 2021.

• GSoC 2021:

  • Carol: Talking with Larry about internship with D&I. GSoC has half the hours and stipend this year compared to last year. Email on tac maillist. GSoC deadline coming up soon. Are there smaller chunks of work to propose for OCIO? Details around ASWF sponsored interships not finalized yet, but got budget approved. Asked for three full time students. Sourcing diverse candidates. Don't have existing network, so more challenging to get going. Since not every project got GSoC students last summer, can take that into account for this year. For GSoC, same mentor time requirement, but possibly less student time. Could do docs perhaps too.

  • Michael: Additional GPU render path (i.e. CUDA)?

  • Patrick: Good but big. Maybe too much. A lot of work.

  • Carol: GSoC student can still work full time, but might be compressed timeline. For D&I, full summer, mimicing previous GSoC full time schedule. For D&I internship, going to use LF platform for communication and progress tracking, etc. Have more control of process. Inclusion focused with more emphasis on introductions to the industry and how we use the project in jobs, etc. Mentorship can be shared by multiple people.

  • TSC agrees it would be great to have D&I and/or GSoC intern, based on student interest and available mentor(s). We could go for one in particular, but since the outcome of applications is hard to predict, considering both could be an option too, so that we could get at least one intern. Group will think about potential projects and make a decision on approach next week. In either scenario it's a win for the project and student.

  • Doug: How many willing to be mentors? Will need to review applicants.

  • Patrick: Reviewing proposals is good test for mentoring too.

  • Carol: All ASWF projects are specialized. Hoping to source people from upcoming D&I webinars. Hosting career and software developer based webinars. Tie back into indistry with how things work. Why is OSS important, etc.

  • Patrick: Could also think about one or two lines about project to raise interest.

• UX Working Group:

  • Michael: At last docs working group discussed transitioning to UX working group, which will help with v2 adoption and docs and examples. Starts this Friday, led by Mark T and myself.

  • Mark B: Will TSC meeting frequency change? There are a lot of open source meetings to attend.

  • Michael: Yes, we did talk previously about reducing cadence once v2 was released. Once or twice a month?

  • Doug: Every other week ideal. Important time to make sure rollout is succesfull.

  • Michael: Every other week starting in March?

  • TSC agrees on cadence change starting in March.

  • Mark B: Will Autodesk involvement change with v2 rollout?

  • Doug: Patrick and I will stay involved. More features to do for 2.x release. Pace of work will slow, but will continue contributions. Been encouraging to see AD step forward and contribute. Hope others do too. Makes a difference to get people focused on it. Part of AWSF about getting members to contribute. Been critical to have this group's participation. We've enjoyed the experience a lot.

  • Michael: ASWF premier members do have an FTE expectation. Think there was intent to make that more of a focus a couple years in, so perhaps will more of a topic this year.

• Items for next TSC meeting agenda:

  • GSoC ideas for next week. Willing mentors.