ASWF TAC Meeting - October 21, 2020
Voting member attendance
- Daniel Heckenberg - Chairperson, Animal Logic Pty Ltd
- Gordon Bradley, Autodesk
- Pilar Molina Lopez, Blue Sky Studios, Inc.
- Michael O’Gorman, Cisco Systems Inc.
- Henry Vera, DNEG
- Bill Ballew, DreamWorks Animation
- Matt Kuhlenschmidt, Epic Games, Inc.
- Brian Cipriano, Google & OpenCue Representative
- Sean C McDuffee, Intel Corporation
- Larry Gritz, Sony Pictures Imageworks
- Jean-Francois Panisset, VES Technology Committee
- Cory Omand, The Walt Disney Studios
- Kimball Thurston, Weta Digital Limited
- Eric Enderton, NVIDIA
- Sean Looper, Amazon Web Services
- Michael Min, Netflix
- Michael B. Johnson, Apple
- Dave Fellows, Microsoft
- Sean O’Connell, Advanced Micro Devices
- Laurent Ruel, Unity Technologies
- Ken Museth, OpenVDB Representative
- Michael Dolan, OpenColorIO Representative
- Cary Phillips, OpenEXR Representative
- Joshua Minor, OpenTimelineIO Representative
- Chris Kulla, Open Shading Language Representative
Other Attendees
- John Mertic, Linux Foundation
- Emily Olin, Linux Foundation
- JT Nelson, Pasadena Open Source consortium / SoCal Blender group
- David Morin, ASWF
- Carol Payne, Netflix, Diversity+Inclusion WG
- Nick Porcino, Pixar
- Patrick Hodoul, Autodesk / OpenColorIO
- Mitch Prater, Laika
- Alex Meddick, Rising Sun Pictures
- Rachel Rose, ILM, Diversity+Inclusion WG
- Mathieu Mazerolle, Foundry
- Deke Kincaid, DIgital Domain
Apologies
- Michael Min (Netflix)
- Cary Philips (OpenEXR)
Agenda
-
Goals for TAC: Year 2
-
Follow ups
-
Survey: project usage, dependencies: Daniel, Emily
-
Gather project specific questions that have come up repeatedly, affecting our CI project, how we manage dependencies, how we address security issues (OpenCue / OpenTimelineIO, trying to cross the CII badge process).
-
Useful start from work done by OpenCue for a project specific survey, reached out all all project TSCs to formulate questions specific to their project, as well as questions common across the projects (support for historical versions of VFX Platform for instance).
-
Want to start survey in Early November, don’t delay the work already done by OpenCue. Want to follow up on this annually so we can provide an annual update. Also good for the annual ASWF cycle, show / indicate progress in our ecosystem to our members.
-
What could we do as action items based on these answers?
-
Emily: will help with the process of submitting the survey. Will work with John to refactor common questions. Would be good to provide specific examples, such as “what software do you use” under OpenColorIO. Good to have Yes/No or multiple choice answers. Daniel: can always have an “other” option.
-
John has set up a SurveyMonkey account. Would be good to have a branching strategy so people can easily skip sections not relevant to them. Similarly for whether survey takers are OK with sharing diversity info.
-
Larry: would be useful to have basic info such as “developer/user”, “studio / independent / vendor”? Good to identify our “core constituency”. Emily: yes, will have identifying info, but optional to allow anonymous contribution. Larry: would still be useful to have some anonymous answer such as “studio vs independent”.
-
Eric: are we looking for a per-studio answer? Emily: more individuals, “the more the merrier”.
-
Daniel: OpenCue survey may have been written with a more targeted audience? Brian: yes, initially written with a “one response per deployment” target. Can always deduplicate based on company name. Emily: not clear how to regulate this. Eric: concerned that users won’t fill a survey that takes more than 5 minutes. Current state is much more than that, few respondents would know answers to all the questions.
-
Larry: is this just a wishlist? At first glance it looks overwhelming.
-
Daniel: we should do a cost/value evaluation, can use UI design to “hide” some of the size of the survey? Can we offer some kind of (intangible) reward? Emily: projects should rank their questions, should be 5-7 per project at most, and multiple choice is better. Can always provide a “reach back to me for further discussion” entry.
-
Brian: this is already the “trimmed down” version, and has mostly multiple choice questions. Were hoping to get contact info from the survey so they could reach out to respondents.
-
Daniel: need to find right way to present this. Larry: project leaders should look at their questions and answer “would you change the course of development based on answers”, and if the answer is no, maybe that question can be dropped.
-
Daniel: Ken from OpenVDB : question about “why are you not using a project” can be useful as well. Maybe we do need to split into broad based vs more targeted.
-
-
Packaging and deployment (OpenCue, OTIO): Brian
- Started email discussion, no conclusions yet
-
Security best practices
- OpenEXR process summary: Cary (absent)
-
-
Technical Project updates
- Via slack: Contribution of rust bindings for projects
-
Working Groups
-
CI
-
Details of next incarnation of VFX Reference Platform, relationship to CUDA, OptIX, how to access those components that require signing EULAs for downloads.
-
How to facilitate distribution of projects without getting lost in minefield of binary distribution (licensing, IP, export controls). Already in that territory somewhat with Docker images.
-
Consideration of how to use Docker Hub in the future, may need to sign up for a paid account to not have to deal with throttling of free tier (slower CI builds due to delays of pulling ASWF containers from Docker Hub).
-
-
Diversity & Inclusion
-
Rachel: having conversations with universities, how can students learn more about Open Source, learned a lot about how universities understand Open Source. Several professors joining the D&I WG.
-
Ambassadors from projects: identified 4, looking for 2 more, this person will join project discussions with specific perspective of D&I.
-
Will join TSC meetings to present the ambassador program.
-
Blog initiative is making progress, list of topics we want to cover, putting dates against these topics, first publish targeted next month on aswf.io blog. First one is “language in code”, next is “unconscious bias”.
-
Emily: looking for volunteers
-
Rachel: working on a webinar series, recorded, available for college age students. Followed by series of career based webinars.
-
JF: could leverage facility HR teams? Emily: yes, trying to reach out, but webinar wants to present to comp sci grads what are the different career paths, what skills are needed, and which open source communities can help develop these skills. Facility HR teams such as DNeg are already participating.
-
Wave: lots of facilities have internal web pages / blogs, could get ASWF D&I blog posts onto some internal blogs. Could be possible to get our content “shopped out” to a lot of companies. Helps to get people in these large companies to know more about ASWF.
-
Emily: once webinar series is kicked off, there will be a page on the ASWF web site, including a D&I section, but need content first.
-
Wave: there’s an internal “Open Source at Apple” site for instance.
-
Rachel: similar to conversation with SIGGRAPH, how we could partner and share resources / sources.
-
Wave: there are opportunities to reuse work, and establish contacts. Carol: a lot of the work is establishing connections, so this could be really useful, welcome any contacts on who to talk to at various companies.
-
Emily: for D&I committee, scope is so large, trying to focus on students for now, somewhat younger demographic. Will develop material that can be reused later on, but current focus is mostly students.
-
Rachel: ways to reuse the same work. Michael: just hired a new intern, so material that targets interns in companies can be useful. Rachel: working on a package of information that can be delivered to professors. But yes, eventually could target that info with SIGGRAPH, companies (same material, different packaging).
-
Emily: blog content can definitely be repackaged. If you know about existing content that could be reused, please share it.
-
Have been using a public Google Driverather than a repo, also a shared calendar, good “source of truth”.
-
-
Python 3
-
USD
-
Cory: USD’s role alongside the VFX Reference Platform, which part should be useful for who
-
Idea of a “universal bridge”, need concrete ideas, there have been ideas for proposals to the group, such as how NVIDIA Omniverse does extensions.
-
A couple of additional proposals: NVIDIA + Pixar physics scheme, what are the motivations.
-
Still TBD for scheduling, but working on a presentation agenda for coming working group meetings.
-
Next meeting start of November
-
-
Assets
-
Meeting yesterday
-
Eric: Rob Bredow shared experience of how to publish the Moana data set. Details of licenses, looked at some assets already out there, “Hyper Space Madness”, “Sony Zombie Rig”. Some people shared projects under way at their studios.
-
Interested in estimating hosting cost, how big would contributed projects be. “What was the first reason an exec said no when you suggested contributing”?
-
Michael: Next meeting is Election Day in the US, should we move it forward a week or a week after? Eric: OK with any option. Probably don’t want to change the cadence. Can discuss on mailing list and figure out the next meeting.
-
Larry: plea to make sure that everyone is included on the mailing list. Please everyone interested should join the mailing list.
-
John: can provide Confluence space if required for a wiki
-
-
Review
-
Daniel: WG focussed on scope and mission, still in “WG scoping phase”, will likely take a little while to set up the infrastructure.
-
“Our WIP goal statement is “ The Purpose of this working group is to investigate and define areas of relevance and interest for the Academy Software Foundation to participate in the support of a publicly available open source playback and review system targeted at the media and entertainment industry.”We anticipate generating a report and recommendations to the TAC and Board with those opportunities enumerated. We are currently aligning on scope, by trying to define the attributes and expectations of a complete system, then narrowing to what requirements might fall within the scope of a playback device.We are currently iterating on the definitions of those attributes and a collection of user stories to identify the audience for any future recommendation.”
-
-
Rust bindings
-
Larry: started on OpenImageIO discussion, multiple attempts at writing Rust bindings for OIIO. A lot of the work for C++ projects is to write a simplified C binding around the C++ API, then you can add a other languages relatively trivially.
-
Joshua: C bindings, then Java bindings for OTIO was a GSoC project. But concern about testing load, but best if a single official implementation is used. But who will maintain it? If there is a community that is motivated to support those (Rust) bindings, then it’s worth doing if there’s a real user base.
-
Larry: would be happy to adopt this in OIIO, but only if there are people committed to maintain and develop it. Several people volunteered themselves, but revealed they had already started on projects. There are industry people who are very enthusiastic about Rust.
-
Joshua: trying to put these repos in a separate repos that sub-modules the main one, so not limited in releasing OTIO versions because the bindings aren’t up to date for instance (not have to be in lockstep).
-
Daniel: how do we encourage this, do we create a working group? Or a per project basis? JT: project by project, if the developer is enthusiastic about Rust, but not a big movement to Rust in independent space.
-
Larry: intrigued by it, but not written Rust yet due to lack of library support, “chicken and egg” situation.
-
Invite person who proposed it to next CI meeting, help with administrative aspect of crate registration.
-
Kimball: for OpenEXR, redoing core as C to help make bindings to multiple languages.
-
Patrick: looked at OCIO, discussion of having new bindings is really interesting, but has a cost in development, also we need to provide infrastructure, documentation, unit tests. Should be seen as a project, investigate the use case, what are the consequences of doing this. OCIO has another binding (Java) and third party code, but no knowledge or CI support, so not supported in V2 OCIO. If you have a “one shot” contributor, this can be an issue, consumers of the project will expect continuing support of an additional language binding. It increases the effort to add new APIs / functionality to projects if you want to do it correctly (API headers, test cases, documentation…).
-
Daniel: having a compact C API helps with reuse (as OpenEXR is doing). Patrick: most of the projects have C++ API, adding a C API, that’s another interface to support.
-
-
-
Update on candidate projects
-
Next meeting: Nov 4 2020