Data
TRACER: Toolset for Realtime Animation, Collaboration & Extended Reality (FABW)
TRACER is a software agnostic communication infrastructure and toolset for plugging open-source tools into a production pipeline, establishing interoperability between open source and proprietary tools and targeting real-time collaboration and XR productions. GitHub; Website
VPET - Virtual Production Editing Tools (FABW)
VPET is a collaborative tool that allows realtime scene editing via an intuitive interface. It takes a holistic approach on light, asset and animation editing, for virtual production environments. Its development is open-source, with growing support for diverse film and VFX pipelines. GitHub; Website
ALVR - Air Light VR (UPF)
This fork of ALVR v20.6.0 introduces several enhancements to ALVR for monitoring and optimizing VR streaming performance.
Music performers dataset (BBC)
This dataset comprises a total of 36 (circa) 60 second clips of music artists performing live, in front of an LED wall, with dynamic background graphics and stage lighting. Four artists performed (Badliana, KDYN, TWST and New Wounds) 3 songs each live and were filmed by 3 locked off (fixed position and settings) cameras: a centrally placed camera, and two either side, all pointing towards center stage. Each video clip set provides an extract approximately 1 min long of the 3 locked off camera views of each music artist, for each of one their songs. Audio is also provided.
To enquire about access, please contact info-maxr [at] bbc.co.uk
Anxious Arrivals (formerly The Day and The Earth) (CREW)
XR performance for multiple users with XR scenography. Website
Survivor AnimHost Motion Capture Dataset (FABW)
FABW is developing AI-assisted software tools and pipelines designed to streamline and enhance workflows for 3D animated film productions. These tools aim to empower filmmakers and animators to create bipedal character animations in real-time with the aid of the latest advances in machine learning.
AnimHost, an integral component of the TRACER ecosystem, addresses these challenges by connecting animation generators to Digital Content Creation (DCC) applications and other clients. The generated animations can be exported, further processed and refined in standard DCC applications such as the open-source software Blender. This integration allows animation artists to remain within their familiar creation environments while utilising machine learning capabilities and maintaining ultimate creative control over their content.
To test, evaluate and visualise this workflow in a realistic production environment, we are currently producing an animated short film titled “Survivor”. The training data for the AnimHost animation generator was developed through a dedicated motion capture session at the Filmakademie Baden-Württemberg, ensuring fair use and traceability of ownership. This dataset has been released under the Creative Commons licence by Filmakademie Baden-Württemberg. Link to asset
Digital Locations - Container City Stuttgart Dataset (FABW)
The DIGITAL LOCATIONS demonstrator exemplifies the advanced virtualisation of shooting locations for film productions. It showcases the successful application of digital replicas of real-world locations for previsualisation, shot planning under physical constraints and full Virtual Production. Additionally, the concept of DIGITAL LOCATIONS enables the virtual and collaborative exploration of buildings, heritage sites and natural landmarks, which can significantly reduce costs and, more importantly, support the Green Deal by minimising the need for travel.
The demonstrator, currently under development, utilises core MAX-R technologies developed by Filmakademie Baden-Württemberg VPET, TRACER, DataHub) and UPF (wgpuEngine). This project was created in close collaboration with members of the MAX-R Creative User Group: Film Commission Region Stuttgart and Third Picture GBR.
Updates on the publicly available demonstrator are expected by early 2025.
Construct: An online developer tool and persistent online world which can host up to 10,000 real time users (IMP)
Construct allows up to 10,000 users to inhabit a persistent world. Users do not need to be authorised to join, they simply need a Google or X enabled account for initial verification. Construct is built on Improbables SOTA technology allowing fully rendered avatars at scale, all able to hear, see and interact with one another. This technology has been used by MAX-R consortium members as well as other projects to deliver high scale events. Users are also able to reach out to enable developer access allowing user generated content to be hosted in their own world. Link to asset