Asset Repository WG - 2 Feb 2021

Attendees

[x] Michael Johnson (Apple) + WG Chair

[x] Joshua Minor (Pixar + OpenTimelineIO)

[ ] John Mertic (Linux Foundation)

[x] Nick Porcino (OTIO, OpenEXR / Pixar)

[x] Eric Enderton (NVIDIA, DigiPro)

[x] Sean McDuffee (Intel)

[x] Eric Bourque (ADSK)

[ ] Erik Hansen

[x] JT Nelson (Pasadena Open Source consortium / SoCal Blender group)

[x] Lee Kerley (SPI)

[x] Phil Sawicki (Autodesk)

[x] Roman Zulak (Imageworks)

[x] Orde Stevanoski (Imageworks)

[ ] Cary Phillips (ILM, OpenEXR)

[X] Sebastian Herholz (Intel)

[x] David Morin

[ ] Sam Richards (Disney Imagineering)

[x] Will Telford (Autodesk)

[x] Sebastian Herholz (Intel)

[ ] Carson Brownlee (Intel)

Apologies:

Agenda Items

Update on Blender discussion
Request from TAC to bring forward official proposal of this WG

Update on Blender discussion

Wave + Josh in discussions.
Francesco (COO / producer on next short): Already have 65 revisions, plus data from older projects. Willingness, details TBA. Exciting! How can we assist or reciprocate? Revisions + WIP would unlock cool use cases.
Mainly pursuing: 2D - Blender. 3D - Moana.

Moana

USD version coming soon. Then what?
Host? If it’s 10s of GB; maybe not if it’s still 100s of GB.
Mine: use to generate multi-AOV EXR assets. Need to create fly-throughs etc.

Camera calibration data - good to store?

Start with one asset type, to pipe-clean?

Start by gathering ASWF projects’ needs. Such as Blender assets for OTIO. Possible VDBs or OpenCue jobs, from Moana.
USD WG subcommittee: Gathering lens info from camera vendors.
Size, shmize. We have big cloud companies on our board. Just need to find the right model. No, size matters (especially while we’re all working at home).
Let’s acknowledge some contributions will be large. We may have different process for small and large. We can at least index.
ASWF doesn’t need 65 cuts of the Blender film. But we may want 5, or 10, pick interesting set.

New project requirement, to come with assets? Alternative to Asset Repo?

centralized vs decentralized. TAC decision?
both: project could have examples. asset repo could have challenging light-transport examples. and assets that don’t map to an OSS project.
Or instead of project required to have assets, maybe require them to state what assets they want. “EXRs with this many AOVs, half-float, and YUV.” “Contemporary camera model meta-data in EXRs.”
OTIO converts things - donor doesn’t have to.

This WG is working towards recommendations to TAC+Board on Assets.

here’s what we recommend
here’s how we could go about it; but details not required at this stage
[David M] start with prototype, for one asset type

Draft Proposal (live edit)

Goals

Goals are the distinct outcomes that is to be anticipated from the working group, serving as a method for validating activities.

[Specify industry segment - see ASWF pitch]
Curated exemplars of the data sets used in our industry and by our software projects

Test sets for OSS projects, for vendor software, and for research
At realistic, up-to-date production scale and complexity

entry level/educational assets as well complex production grade assets? [BIg project, more Khan Academy / Pixar-in-a-Box / etc]
Reaching out to studios for film production assets, as well as non-studio film producers like Blender Foundation.

Non-Goals

Non-goals are things that the working group are intentionally choosing not to do, the things not expected to change due to the working groups efforts, or the subject matter areas that the working groups doesn’t want to address at this time.

not a backlot of reusable assets for production
not a dumping ground (i.e. we will curate the collection, not allow just anyone to put something there)

Deliverables

Make it easy:

Licenses: One, or a very small set.
Hosting: TBD if this proves practical / valuable for all sizes of outlets.

Clear criteria for acceptance, and clear curating mechanism.

Formats. Who does conforming. How do we version it, label for versions of specs.
Scope - “Film and media industry” or similar