Description
Describe the purpose of the group in no more than 4-5 sentences
ML refers to tools that use statistical techniques or by-example training to implement features and achieve results that would be difficult or impossible using traditional programming of explicit step-by-step algorithms.
Every project and member company is considering how this family of techniques can be used in a film production pipeline, in a tech and media environment full of hype and uncertainty, and many of our member companies and projects are wary and unsure of how to adapt to and use the technology, or even how to talk about it.
We want to create a working group where the ML experts among our projects and members can meet, share expertise, better understand what ML-based tools are needed in our industry, and where helpful, spin off and guide ML-based open source projects aimed at addressing the specific needs of our film pipelines.
Goals of the working group
- Plan, incubate, and spin out concrete open source projects within the ASWF that use ML technology in artist- and production-serving ways.
- Establish a community and communications channels for the ML experts among our projects and members to be able to share information and expertise, in a similar way to how ASWF currently has projects and working groups that bring together top experts in color science, rendering, and other areas.
- Lead evolution of a vision, terminology, norms, and technology related to ML within the film industry that offer a positive, productive, ethical direction for ML-based tools that can help artists and productions, to counter the constant external noise of both hype and anxiety surrounding "AI."
When it comes to the primary goal of turning ideas/needs into open source software projects, we wish to prioritize ML related projects that have broad utility among the members and constituents and that multiple members would otherwise need to develop separately and redundantly, and therefore are not going to be strategically differentiating for any of them. By working on some of these things together, we aim to reduce needless work and preserve more of each organization's limited development resources for the tasks that serve their unique or differentiating needs.
Non-goals of the working group
- We avoid both doomerism and boosterism, philosophizing, or navel gazing about "AI" in the general sense.
- We avoid getting involved with issues or politics outside of the film industry. There are other places to do that.
- We are not interested in building tools that take creative control or satisfying work away from artists.
In short, we are narrowly focused on the practical and implementation considerations of specific, concrete ML tools that can empower artists and make our films better and more efficient, no differently than how we would approach any other specific technology area in our space, such as shaders, volumetric data formats, or color management.
Deliverables
It is hard to enumerate specific deliverables for a topic that is as rapidly evolving as ML in production. The most important ongoing deliverable is simply the forum and communications channels by which the ML-focused engineers can forge a path forward that serves artists.
We expect the WG, as it identifies areas that would benefit from open source software development activity, to spin out sandbox or incubating projects within the ASWF, to be developed jointly by members and the community.
We would like to have at least one project in active development and ready to be announced by SIGGRAPH 2025 (and, if at all possible, have an MVP-adjacent preview at that time). Additional projects, perhaps with longer timelines, should be identified and launched by the end of CY 2025. We expect the WG in the long term to continue to identify areas in which the members can pursue open source development of additional projects. There is no specific timeline or goal number of projects; it is the goal of the WG to assess what needs to be done.
Metadata
Metadata
Assignees
Type
Projects
Status