Academy Software Foundation Technical Advisory Council (TAC) Meeting - May 14, 2025

Join the meeting at https://zoom-lfx.platform.linuxfoundation.org/meeting/97880950229?password=81d2940e-c055-43b9-9b5a-6cd7d7090feb

Voting Representative Attendees

Premier Member Representatives

Project Representatives

Carol Payne - Diversity & Inclusion Working Group Representative, OpenColorIO Representative
Cary Phillips - OpenEXR Representative
Chris Kulla - Open Shading Language Representative
Diego Tavares Da Silva - OpenCue Representative
Jonathan Stone - MaterialX Representative
Ken Museth - OpenVDB Representative

Industry Representatives

Jean-Francois Panisset - Visual Effects Society

Non-Voting Attendees

Non-Voting Project and Working Group Representatives

LF Staff

David Morin - Academy Software Foundation
Emily Olin - Academy Software Foundation
John Mertic - The Linux Foundation
Michelle Roth - The Linux Foundation
Yarille Ortiz - The Linux Foundation

Other Attendees

Lee Kerley - Apple
JT Nelson - Pasadena Open Source consortium / SoCal Blender group
Tommy Burnetten - ILM
Cottalango Leon - Sony Imageworks
Dhruv Goil - Apple
James Spadafora - ILM
Lorna Dumba - Framestore
Mark Peters - OTOY
Cory Omand - Pixar
Colin Chargy
Eric Powers - Dreamworks
Jim Geduldick
Jim Helman - MovieLabs
Juan Buhler - Sony Imageworks
Lydia Zheng
Olga Avramenko - SPI

Antitrust Policy Notice

Linux Foundation meetings involve participation by industry competitors, and it is the intention of the Linux Foundation to conduct all of its activities in accordance with applicable antitrust and competition laws. It is therefore extremely important that attendees adhere to meeting agendas, and be aware of, and not participate in, any activities that are prohibited under applicable US state, federal or foreign antitrust and competition laws.

Examples of types of actions that are prohibited at Linux Foundation meetings and in connection with Linux Foundation activities are described in the Linux Foundation Antitrust Policy available at linuxfoundation.org/antitrust-policy. If you have questions about these matters, please contact your company counsel, or if you are a member of the Linux Foundation, feel free to contact Andrew Updegrove of the firm of Gesmer Updegrove LLP, which provides legal counsel to the Linux Foundation.

Agenda

General Updates
- Dev Days - May 15, 2025 #966
- OpenQMC #434
- Rename project lifecycle stage ‘Adopted’ to ‘Graduated’ #999
- DigiPro CFP closes 5/16 #1003
- Open Source Days 2025 #1005
- Add question for resources to project proposal #1024
- Propose to remove vote for annual reviews #1016
- Launching Machine Learning WG #1046
- SMPTE 2025 Annual Technical Conference - CFP closes 5/31 #1049
New Project/Working Group Proposal: New Project Proposal - Dailies Transcription / Notes Taking Tool #1040
Annual Review: ASWF Language Interop Project #704

Notes

Dev Days - May 15, 2025
- Starts tomorrow, lots of chatter on Slack. Larry: how many people are signed up? We’re now at 60 something (signups).
- John: some people will also just sign up in Slack
- Larry: good luck everybody, be prepared to receive the contributions!
- Emily: we’re at 70
Open Source Days 2025
- Emily: 2 more days for CFP for Open Source Days. Projects which want to do a BoF, pick a time in the spreadsheet, same for Virtual Town Hall.
- Want to be ahead of the game and publish schedules early so we can get people to attend.
DigiPro CFP closes 5/16
- John: if people have submitted for this, let Emily know, what to keep a tally.
- Emily: same for SIGGRAPH, OpenVDB running a course, OSL lab. If there are activities related to projects outside our own events, let us know so we can include this in our communication efforts for SIGGRAPH and DigiPro.
SMPTE 2025 Annual Technical Conference - CFP closes 5/3
- John: until end of month for CFP, let Emily know if you plan to submit. Emily: we would like to put together a foundation level submission, if you are going / attending, let us know.
Launching Machine Learning WG
- Vote passed!
- Larry: look at the TAC issues page for the original proposal / submission. Now that it’s there, we’ve already set up the #wg-ml Slack channel to see what’s going on. Soon will launch a poll for the meeting time.
- If you have ML activities at your company and would like to participate, let people on your teams know so they can join the Slack. Project presentation you will see later will give a flavor of the directions we hope to go in.
- WG isn’t meant for people to pontificate on ML, mission is to figure out what we can productively do together, and propose projects to spin off to meet those goals. We will concentrate on successful Open Source projects. Also a room for ML experts to talk.
- JF: suggest to post the #wg-ml to #announce? Larry: want to be careful to not turn it into a general “chat about AI”
PR #1024 against TAC repo
- Add question for resources to project proposal
- John merges it
New Project/Working Group Proposal: New Project Proposal - Dailies Transcription / Notes Taking Tool #1040
- Presentation Slides
- Tommy Burnette : ILM, have been there for 28 years, Director of Core Pipeline, was in Singapore and London, back in US. First involved with Open Source in mid 90s contributing to Python pre 1.0. Oversaw Alembic contribution to Alembic.
- Background
  - Rob Bredow summarized what has been happening, ILM was about to begin a hackathon where all engineers working in small teams on ML related projects. Surge of activity in this space, giving confidence in what can be achieved.
- ML in Content Creation Workflows
  - The problem
    - Many shared problems ML techniques could be applicable to
    - Nobody knows what to build or how to talk about it
    - Need a safe space to explore
  - The solution
    - ML working group
    - …
- Dailies Transcription / Note Taking
  - Timely, high quality notes from dailies and reviews are crucial…
- Goals
  - Automate Transcription and note-taking
  - Modular design
    - Be able to update to new model
  - Shotgrid integration
    - Not everyone uses it, but enough to that we can aim for that at first
    - Provide context for analysis, send output back to it
  - Flexibility and on-premise deployment
    - Can use on prem vs cloud model
- Features - MVP
  - Transcription
    - Advanced speed to text
    - Speaker identification
  - LLM-based analysis
    - LLM integration
    - Note generation
    - Action item extraction
    - Can get pretty far with a good prompt
  - Shotgrid integration
    - Shot/Asset recognition
    - Note creation/updating
    - Note linking
    - Timestamp navigation
- Additional Features
  - In scope
    - Manual editing , note-taking, and tagging
    - Search and filtering results
  - Out of scope
    - Real-time ID direct from audio/video
    - Sentiment analysis
    - Anything but Shotgrid
      - Structure code to allow for other system, avoid Shotgrid-specific
- Challenges and Opportunities
  - Starting frmo scratch - together
  - What if they don’t want this?
    - Talked to production workflow people to introduce idea, first reaction was “no”. Senior coordinators take these notes, they understand the context, they don’t want to proofread AI generated notes.
    - WG wants to think through where challenges come from. Want something that benefits people.
  - Starting to leverage Gemini transcription in Google meetings
  - Even just transcription of a busy dailies conversation would be hard to object to?
  - Natural Language query of production tracking info is exciting, “give me top 3 assets mentioned in notes”, having full transcriptions of notes that query has access to could be useful. What does the result of a query look like?
- Moving forward together
- Discussion
  - Jim G: have done some of this on some supervision projects. Have you looked this not only for in-studio note taking, but also data coming in from set? Reccies, on set capture. A lot of data we have to deal with. Tommy: not exactly, we’ve explored what we have entities in Shotgrid, describing schemas of the database to the LLM to generate natural language queries. If it’s in Shotgrid it can be done. Another approach beyond that, an agent talking to Shotgrid could then talk to asset tracking database. So could stitch together multiple data sources. Don’t have PoC for that yet. Jim G: asking LLMs, starting off with CSVs / text files, what is done at ILM will be different than Sony than a production side supe. What are the definable fields that should be in a template to work / interop across industry. We’ve needed this for a long time. Maybe you are at a main vendor but have sub vendors. So a gret project. Tommy: possibilities are infinite!
  - James S: in rally / hackaton we figured out you can extend frameworks, if the framework can accept various tools from various studios, can extend LLM to access that. As long as the prompt explains the tool and how to use it, it can get info from there.
  - Tommy: when describing the agent, it’s almost like docstrings in your code, but the agent is reading them. The better the description the better it can figure out what to do. ILM can provide their description, Sony provide theirs… but under a common framework.
  - Cottalongo: every production string. has input/output, typically CSV. Asked for some sample data, impressive what coordinators do, the interaction is very fast, will be interesting technical problem to solve. From SPI side people are open to this because they are understaffed, reviews happen all the time, so looking for assistance / curious. They tried Gemini transcript but stopped using it since it was useless at that point. Custom LLM solution seems appropriate, third party LLM wont have as much context. Tommy: tuning the model that’s listening is important. Larry: potential tie-ins to ORI projects, media review tools themselves, audio tool is coming from a meeting looking at synced media, so that’s another source of info, what shot is being looked at when someone says something.
  - Tommy: even pronouns, “her dress is too green”
  - Carol: proposal would be a sandbox project, a great example of a “greenfield” project, we don’t exactly know what it is yet, ORI came in this way, hopefully can get momentum behind it. Larry: there isn’t an existing codebase, ILM / Imageworks are throwing their hat into the ring and pledging resources, but if other companies are doing work, this is not just ILM / SPI, this is wide open to get the best ideas and implementation collaboration. This is why we want to start this off as a “real” ASWF project.
  - Tommy: a fast moving space, conversations will be different in other studios. Perfect thing to start in a sandbox.
  - Dhruv (chat): Will this be completely greenfield in the ASWF or will there be something from ILM/SPI to bootstrap it? Larry: We’ve both done some prototyping, but no, it’s greenfield.
- John: will take to a LFX vote
- Larry: if people need to be part of the discussion, #wg-ml chat already exists, point them there for now. There will be dedicated channel
Annual Review: ASWF Language Interop Project
- Scott: proposal to move to Archived state. Running out of steam for project, and other maintainer also unable to participate, the WG is in limbo. So want to archive it for now and put it to the side.
- Carol: this is very fair, and thank you for the work you’ve done so far. Language Interop is important, but may need to be handled in the projects themselves. May be the best way to move forward. Scott: Interop WG could still provide some documentation for a project who wants to integrate project X, here’s what to pay attention to, what works, what doesn’t. But haven’t had any energy to get the documentation started.
- John: moving to Archive isn’t a death knell, just puts in a frozen state, but it can be brought back later.
- Larry: any project can do additional work for language bindings individually, overall WG may not be needed unless there’s a required cross project initiative. This isn’t a rejection of Rust / Swift… just a resource issue.
- John: this will go to LFX vote
- John: thank you Scott for all your work on this
Carol: if there are any questions about Dev Day, we have some time now.
- Olga: please if you know anyone who still wants to participate, encourage them, even if they don’t get it all done on Thursday
- Carol: got 4-5 people in OCIO grabbing issues today
John:
- OpenCue and CI WG on May 28th
- June 11th: OpenImageIO and OpenFX
- May have some flexibility, but SIGGRAPH coming up
- New leadership for raw2aces, getting them into the mix
- Also going through backlog
- JF: plug the ASWF PyPI org
- John: will also want to inventory any other account credentials