04 papers — open research, published for the industry
Show··
R-01Warp at the Geometry BoundaryGPU kernel languages have made the compute half of procedural simulation close to free. We built a compiled Houdini operator that executes NVIDIA Warp kernels directly on geometry, then measured where the time goes. Across every workload we tested, the kernel was never the cost. The boundary between the host application and the compute layer was, by an order of magnitude or more, and it is the part almost nobody instruments.PUB — SEP 2026CAT — vfx pipelineLIC — OPEN SOURCE
R-02AI Slop Is a Craft Problem, Not a Tech ProblemWe are an animation and visual effects studio. We spent years learning to make frames that hold up on the biggest screens in the world, and we use AI to amplify that craft, never to replace the judgement behind it. Our position on AI slop, from the floor of a working studio.PUB — AUG 2026CAT — publicationLIC — OPEN SOURCE
R-03CrucibleEvery procedural effect in a Houdini pipeline passes through an irreversible step: the bake that turns a live asset into geometry a renderer will accept. We argue that step is in the wrong place. Crucible is an open-source Hydra scene index plugin that moves it to render time, cooking HDAs lazily as Karma pulls on the scene graph, with no pre-cook, no LOPs recook and no HAPI session to manage.PUB — MAY 2026CAT — vfx pipelineLIC — OPEN SOURCE
R-04A Splat Should Be a PrimitiveGaussian splatting produces photographic reconstructions faster than modelling can, and then strands them outside the pipeline as a foreign format that composes with nothing. We argue the fix is not a better importer but a change of status: define splats as first-class USD primitives with their own schema, imaging adapter and Hydra delegate, and every composition feature in USD applies to them for free.PUB — MAY 2026CAT — vfx pipelineLIC — OPEN SOURCE
Bring us the unsolved