ASWF TAC Meeting - December 1, 2021

Video Conference Link

Voting member attendance

Other Attendees

John Mertic, Linux Foundation
Shannon Deoul, RazPR
Rachel Rose, ILM / D&I WG
JT Nelson, Pasadena Open Source consortium / SoCal Blender group
Scott Wilson, Rust WG
Matthew Low, DreamWorks Animation
Ashley Whetter, Python 3 WG
Sergio Rojas, Arena World
Allen Rose, Madison Square Garden Entertainment
Jim Helman, MovieLabs
Nick Porcino, Pixar

Apologies

Joshua Minor, OpenTimelineIO Representative

Agenda

Google Summer of Code participation discussion (cont)
- Rachel: would make sense to apply to GSoC, also pursue a ASWF specific program for additional internships that don’t fit in GSoC requirements / other projects we want to get done.
  - Anyone responsible for a project, in your TSC meetings, discuss possible projects would be useful. Interested not only on pure coding, but also documentation / other projects, will want to have a good list of those early next year.
  - What does it look like to have our own program, what would we want out of this, 3 months, everyone at the same time?
- Scott: for projects that have already had discussions, who to talk to to send proposals? Rachel: Carol and I will take care of applications to GSoC, as well as spearhead process of pulling together our own program, possibly leverage LF infrastructure. Please send to Carol and Rachel
- Anything not coding would have to be under our own, since that doesn’t fit under GSoC
- Board has not been approached yet for funds, that had been approved last year, but we didn’t do it last year, also had strong focus on summer learning program. So will be important to put together a proposal for funds approved by the board.
- Larry: for GSoC, have we decided who will take responsibility for filling application? Rachel: offered for Carol and Rachel. Larry: agree with that. Can provide records of what was submitted the last time we did this if that’s useful / not have to start from scratch. Rachel: that would be great. Larry: all projects must get project ideas well ahead of time so it can coalesced into a single page. Eric; should we declare a deadline? Rachel: yes, GSoC application is due end of January, so January 15th would be a good deadline for proposals.
- Rachel: anything we can levery from other LF-backed foundations? Best way to move forward is once we have list of projects, we can identify the ones that don’t fit in GSoC, and can point to overall size of internship program.
- Eric: we have a TAC meeting on Jan 12th, should we make deadline the 11th? Rachel: yes, that sounds like a good idea.
Open Source Forum 2022
- Shannon: rough agenda we’re thinking about. Foundation will send a “save the date” later this month, planning March 10th at Academy Museum, board, tac, 2-4 spots per member company, will also have remote attendance (week of VES Awards).
- 11-12 ASWF Governing Board Meeting
- 13:00-13:15 State of the Union
- 13:15-15:15 TAC project updates
- 15:15- … Going Forward
- Contingency plan in case it can’t happen in person? Shannon: LF will require proof of vaccination + masks at all times
TAC chair
- Kimball will be on the Governing Board next year, is there someone interested in chairing the TAC, we should have a rotation, but there’s also benefit to continuity. We don’t have term limits? John: no official term limits, but we revisit once a year. We can keep a period open until next meeting, can send “notice of interest” / nomination to John. Role can be demanding, one thing could consider could be a “co chair”, one person who takes on the meetings, the other interfacing with the board. Kimball: yes, that could be an option. So let’s say that if someone wants to pitch their hat (or someone else’s hat) in the ring… Larry: I’d like to nominate Kimball! Rachel: it has been very helpful to have Carol & I to chair the D&I group at the same time, so a co-chair setup for the TAC could definitely be helpful. Kimball: please send names to John, and we can discuss next time.
Project Updates
- Asset WG: Eric. Proposal from AWS for a fully rigged Maya character assets. Not USD at this time. Have a video explaining what the controls do in the rig, also talked about documentation. Enthusiastically voted to accept into WiP section of the repo, can help to develop to the point it can be public. Discussion continues with LF for a website, will use Garett with a friendly front end. Relative merits of just having large tarballs (not have to deal with git), vs pointers to individual files, but some of these are quite large, so will need git-lfs. Current plan is to have in git/git-lfs form, will make it easier to access / propose modifications. If these assets turn out to be quite large, need to keep an eye on download costs, make sure it’s not being done by robots.
- OpenCue: Brian. Not many updates, missed a few planned TSC meetings due to a mix a factors (holidays / vacations / illness), have a TSC planned next week. Most priorities are the same as before: CY2022 full support including Python 3.9, new planned user submission system, nothing concrete for now.
- OpenEXR Cary: steady trickle of issues / minor bugs / doc fixes. Likely to stay quiet through the end of the year. Accumulating list of initiatives that will be spun up in the new year, for instance fixing up so versioning / namespace practices. Kimball: have official production renders using OpenEXR 3.1, changes made for the core stuff is making 30% performance improvement to shading.
- OSL Chris: missed two last TSCs due to schedule issues, no new updates.
- MaterialX Jonathan: had good presentations from SideFX and Solid Angle showing latest integrations into Solaris and Arnold. Both projects look very promising and teams are very engaged with USD and MaterialX. Have had new proposals from SideFX for nodes such as Worley noise, had been specified but not implemented, looks promising. Also contributing ramp nodes that were missing, looking to integrate in 1.3.9. Upgraded version of PyBind to support Python 3.9 in GHA test suite. Discussions about refining the scope of MaterialX, removing overlap with USD so you can fully support MaterialX with the material functionality, will try to integrate in 1.3.9 as well, and discuss at next USD / MaterialX WG. SideFX put together chess board asset, pure USD / MaterialX which they use in Karma tutorial videos, made freely available. Could be a contribution to Asset WG, a pure USD / MaterialX production quality asset, with subsurface scattering. Started discussion to see if they would be interested in contributing asset. Would be good to have that part of test suites. SideFX Karm Rendering Course Announcement
- OCIO / Michael. Not a ton new, put a date on next round of releases on Dec 14, updates that have been in the works. Second PR from Apple with Metal test suite. ACES Config part: almost done with CG config, a fully self contained ACES config based around CG work, should be very useful.
- OpenTimelineIO update (Josh pasting this in) (Nick)
  - Autodesk’s latest release of RV includes support for OTIO: https://community.shotgridsoftware.com/t/rv-2022-0-0-release-is-available/13949
  - Foundry is showing interest in OTIO support & attending upcoming TSC meetings
  - Lots of work done on Python wheels, automatically deployed and tested for all platforms and supported Python variants. Happy to share how to automate Python wheels..
- USD WG (Nick). USD Web getting close to their approach, should be able to start rolling out patches to compile USD to WASM. USD MaterialX group hasn’t started yet. USD Camera: created GitHub repo to gather ideas / sets of attributes that have been in Slack channels until now. Lots of new data being gathered by Michael Min, new Slack channel #wg-usdcamera. Larry: USD currently doesn’t built against OpenEXR 3 aka VFX Platform 2022, would be good if we could figure out what’s happening and put them in touch with OpenEXR team. Nick: talking specifically about building USD with imaging and getting OpenEXR vs OIIO? Larry: no, it’s more about internal usage of Imath, so just needs touch up with #include paths.
- Rust Scott: looking at priorities for January
- CI (JF) Docker images updated to OpenVDB 9, feature complete for VFX Platform 2022
- OpenVDB: Ken: releasing 8.2 with TBB 2021 and OpenEXR 3, so features which don’t change the ABI. Working a project that’s pitched to TSC, new command line to expose all the tool functionality in a single command line tool. Read a mesh file, convert it, level set processing, ray tracing… Duplicate a lot of the workflow you can find in Houdini, but from the command line. Will share repo next week and TSC will discuss on directions. May integrate with Ax, working on that at NVIDIA.
- Raw2aces: Kimball: had a couple of TSC meetings, going through legal documents, getting assignments to ASWF sorted out. No specific code transfers yet.
Python 3 Working Group Request for Closure
- Ashley: last time we spoke about it, talked about writing up a report, and are we OK to close it. WG has a target and should either be extended or closed down.
- Document has been written: (https://docs.google.com/document/d/1RmvZPwwhLu0KmCXbn4XgS4uzVuFCT2K21eQdQfnWYLE/edit?usp=sharing)
- Goals of the WG were to share and develop best practices between studios and DCC vendors, and coordinate some transition efforts between studios and DCC vendors, and share common experiences to reduce duplication of effort.
- Achieved most of those through deliverables: a resources page that links to external resources to help teach how to move to Python 3. Also a basic transition guide as a starting template for bigger studios to write their own guides to have a more VFX-centric guide.
- SideFX made a page listing some of the things they ran across when transitioning their C API from Python 2 to Python 3
- A deliverable we didn’t manage to produce is a list of risk factors for the VFX industry. Risk factors that led to slow transition were studios / DCC vendors in a deadlock, python 2 support still existing means studios aren’t incentivized to transition and vice versa. Has eased over time. Tried to get statements from vendors as to when dropping Python 2 support, Autodesk provided such a statement. Was looking for a statement from The Foundry but haven’t gotten one yet. Other risk factor is still not all third party software is Python 3 compliant, some packages still not Python 3 compatible.
- WG may not have as much change as it would have hoped to. Also status updates from studios showed that transition work hasn’t been priorititized, at time of last meeting, WDAS and DNEG were the large studios that were making sustained efforts.
- Larry: can you share the Autodesk position? Ashley: have had it for a while. Support for Python 2 in version of 3DS Max after 2021, for Maya and MB, same, so Python 2 gets dropped in what comes after Maya 2022. Available in the WG repo on GitHub.
- Kimball: do you see things the WG should continue working on? Ashley: if the pace of the transitions picks up within the studios (hopefully Maya dropping Python 2 support), then that might be worth picking up the WG again. In the DNEG transitions, haven’t come across problems that are specific to VFX, resources that exist out there cover 98% of what was needed, apart from what is internal to the studio.
- Kimball: WETA has a number of workflows that are Python 3 only, including Houdini and Maya. Standard tools work, finding about ascii / utf-8 issues. Took snapshot of work from WG to access internally. Thank you for the great work!
- Gordon: WG started with VFX Platform starting Python 3 transition, we didn’t have support in packages we depend on. Goal was to have sure to have a plan for transition. What do we think success looks like / do we think all the barriers are down and studios have all the tools they need for the transition? Or are there other definitions for success. Ashley: ideally studios need to be in a place where they can “do the right thing” and start using Python 3 and get rid of Python 2 everywhere, but that’s not necessarily the TAC’s goal. Not clear how to quantify that. Gordon: any other activities that could enable this transition, or got to the end of identified work. Ashley: yes, got to the end of the identified work, at this point studios need to prioritize it. Larry: it wasn’t until we had the DCCs in house to allocate time for the work, with plans to use them before priorities could be adjusted. Kimball: starting to encounter a “forcing function” such as third party open source packages that are only Python 3. Don’t know of additional efforts that would be needed to encourage people to convert. Ashley: breaking point will be when Python 3.10 makes into VFX Platform since it removes so much Python 2 compatibility, and it will be impossible to maintain 2.7 and 3.10 compatibility (3.9 still has some compatibility).
- Kimball: we can celebrate a great job and close off the WG. We can always reopen the process later if needed. Gordon: it’s easier to start WGs than to wrap them up, so it’s good to agree that we’ve gotten to an end-point achievement.
Larry: have we decided the next meeting is happening, since that’s the time of the year when people are less available (Dec 15th). John: will want to keep the 15th, there’s a project proposal likely to come in. But the 29th might be a skip.
Pipeline working group proposal from 17-Nov: Invite Milan to present at next TAC meeting