The Academy Software Foundation held its annual Open Source Days event in August, helping to further open source software development for VFX, animation and digital content creation.
The Academy Software Foundation recently held its annual Open Source Days conference in early August. The event is dedicated to furthering open source software development for the visual effects, animation and digital content creation industries.
The Foundation was established in 2018, as a partnership between the Academy of Motion Picture Arts and Sciences and The Linux Foundation, to create a base for cross-industry collaboration in the development of tools supporting film and animation production. Today the Foundation has over 30 member companies including Adobe, AMD, Autodesk, Dreamworks, Epic Games, Intel, LAIKA, Microsoft, Netflix, NVIDIA, Samsung, Sony Pictures Imageworks, Walt Disney Studios, Weta FX, Warner Bros. and others.
Coming full circle to SIGGRAPH in Vancouver where it was first launched, this year's Open Source Days event showed the Foundation's industry momentum and participation, with over 500 in-person attendees and others tuning in virtually from around the world.
Open Source Days News
News included LAIKA and Skydance Animation joining the Foundation as Premier Members. The Academy Color Encoding System [ACES], the worldwide standard for colour management, became a new hosted project. Also, the new Machine Learning Working Group was launched, getting underway with two projects – Dailies Notes Assistant and Rongotai Model Train Club [RMTC].
All of these developments are exciting, but come at a time when VFX and animation production is experiencing disruption. Therefore, the Foundation brings a welcome spirit of cross-industry collaboration and sharing that will strengthen the business overall.
"The Academy Software Foundation was created to ensure that we have a healthy, vibrant open source community that can maintain and grow the projects that the motion picture industry relies on," said David Morin, Executive Director of the Academy Software Foundation. "Over the past seven years, we have seen an increase in developer participation, cross-industry collaboration and community adoption of Academy Software Foundation projects. ACES is a critical project for our industry, and we are honoured to provide a home where it will continue to thrive and grow."
The Foundation’s New Projects
ACES is an open source framework for colour management and image interchange across the full motion picture production life cycle, from on-set acquisition, visual effects, post-production, mastering and archiving. It has become an effective standard for ensuring a consistent colour experience and maintaining creative vision, used in recent years on films including Captain America: Brave New World, The Wild Robot, Wicked and others.
Originally developed and previously maintained by the Academy of Motion Picture Arts and Sciences for more than a decade, ACES will now be maintained by the Foundation, taking advantage of its open governance model, legal framework and community infrastructure. This move will benefit both creative users and technical implementers, driving collaboration and integration between ACES and other key open source projects such as OpenColorIO, OpenEXR and MaterialX.
The new Machine Learning Working Group was formed to plan, incubate and launch sound, open source projects within the Academy Software Foundation that use ML technology ways that serve artists and productions. The group will also establish a community and communications channels for the ML experts across Foundation projects and members to be able to share information and expertise. Part of its charter is to lead the evolution of a vision, terminology norms and technology related to ML within the film industry. It desires to present a positive, productive, ethical direction for ML-based tools that can help artists and productions.
The Machine Learning Working Group already has two projects underway. Dailies Notes Assistant will simplify the dailies process by transcribing dailies meetings, analyzing the content using a large language model [LLM], and integrating notes directly into Shotgrid for production tracking. The Rongotai Model Train Club [RMTC], which began as an internal project at Wētā FX, will develop a VFX-specific framework for simplifying the production and deployment of ML models and datasets by clearly tracking their provenance and ensuring connection to rights holders.
Foundation Family Expansion
Skydance Animation has joined as the Foundation's most recent Premier Member. The Skydance Animation production pipeline uses several open source projects governed by the Foundation including OpenColorIO, OpenEXR, OpenImageIO, OpenTimelineIO and Rez. Skydance Animation develops and produces feature films and television series across two studios in Los Angeles and Madrid. The studio's releases include the feature films Luck and Spellbound with upcoming releases Pookoo, directed by Nathan Greno, Ray Gunn, directed by Brad Bird, and an untitled Jack and the Beanstalk project directed by Rich Moore.
LAIKA, an award-winning feature film animation studio known for its distinctive artistry and films such as Coraline, ParaNorman and Kubo and the Two Strings, also recently joined as a Premier Member. In 2016, LAIKA received a Scientific and Technology Oscar for pioneering the use of rapid prototyping for character animation in stop-motion film production, enabling artists to leap ahead in character expressiveness, facial animation, motion blur, and effects animation.
LAIKA has been a longtime contributor to the Academy Software Foundation, with engineers active within the Open Shading Language project, the MaterialX project and the OpenUSD Working Group, among others. As an official member of the Foundation, LAIKA can support the health of these critical projects and contribute directly to expanding the Foundation's mission and addressing the needs of smaller studios.
Video of all Open Source Days programming is now available on the Foundation's YouTube channel here. www.aswf.io
