- John Mertic (Linux Foundation)
- Cary Phillips (ILM, OpenEXR, Sci-tech council)
- Darin Grant (Animal Logic, ASWF Treasurer)
- David Morin (Epic / ASWF Exec Director)
- Eric Enderton (NVIDIA, DigiPro)
- Joshua Minor (OTIO / Pixar)
- Nick Porcino (OTIO, OpenEXR / Pixar)
- Sean McDuffee (Intel)
- Will Telford (ADSK)
- Eric Bourque (ADSK)
- Orde Stevanoski (Imageworks)
- Lee Kerley (Imageworks)
- Daniel Heckenberg (Animal Logic)
- Bruno Sargeant (Unity Technologies Inc - M&E PM)
- JT Nelson (Pasadena Open Source consortium / SoCal Blender group)
- Roman Zulak (Imageworks)
- Michael B Johnson (Apple)
- Rob Bredow (ILM, ASWF Chair)
- Welcome/Intros
- Discuss purpose of this working group and background
- Select WG leader(s)
- Align on proposal ( template at https://github.com/AcademySoftwareFoundation/tac/blob/master/process/wg_readme_template.md )
- Next steps and meeting cadence.
Intros - some concerns:
- OpenEXR images repo = an example of best practices
- Testing / conformance
- mentoring / canonical references
- Individual assets
- Scene structure (USD?)
- portability of assets
- repurposing for multiple markets / applications
- film production quality e.g. Moana Island
- research; help make research applicable for real world problems
- no one at studio says no - but no one is authorized / motivated to say yes
Slides from Governing Board 09/22/20
- Asset repo voted number one idea for ASWF “what else”
- Community is open to assets for testing only as well as for production use
- 3D, 2D, and ML were the top three asset repository types requested
- License examples:
- Smithsonian Open Access (view only but CCO)
- Data.world (data science data sets)
- ASWF logos
- Research oriented: https://developer.nvidia.com/orca ;
https://casual-effects.com/data/ .
- These datasets are static (no animation, rigs etc)
- http://videomatting.com/
- Licenses: Unmodified licenses are easier. As few as possible … ideally 2.
- Quality levels: Could have badging / tagging (Testing Only, Stress Test, Production Asset, etc). Production cruftiness; clean and portable; reference asset
- Taxonomy badges (Material, mesh, image, etc)
Goals and Non Goals
- Scope: Small set of exemplars. Quality over quantity.
- One asset per ASWF project? A good place to start, but leaves out many other purposes.
- Describe the MVP Minimum Viable Product.
- Could pick a Blender film as a starting point, and ask each project, does this work for you? Draw out specific lacking features, scale, etc.
- Are there enough AOVs to make them interesting OpenEXR files?
- Modernize OpenEXR Image lib. Have cameras, but what is most needed? Best to have specific demand.
- Stress cases - abusing the format - can come from production or from project.
- Level of goals: rather than “these exact outcomes”, look for value we want to deliver: “serve needs of our projects” or etc.
- To turn “maybes” to “yeses” for studio participation: Establish selfish goals. Such as help this ASWF project you use.
- These could be on-boarding / mentoring / diversity projects.
- Students could be a great help. Under direction of experienced studio person.
- ASWF uniquely positioned to raise ideas in research community: identify important new use cases, etc.
- Remove licensing hurdle, hosting hurdle. Make it viable for a studio to donate.
- Identify or develop assets. But also maintain over time.
- A snapshot in time is easier than maintaining over time. Keep it easy.
- PRMan tests vs USD backlot: Freezing assets for testing over time versus updating formats to reflect current best-practice.
- ABC and PBR shaders → Python script → USD files. Wave changes the script periodically; source material constant, USD format evolves.
Next Steps and Meeting Cadence
- Michael B Johnson steps forward to chair - Thank you!
- Meetings @ 2 weeks, same time slot.
- Homework: Look around your company for potential assets, let’s pool the list, see what’s rare or missing, etc.
- Working Group mission right now: Create proposal(s), what should ASWF be doing about assets. We can boil the ocean later. :-)