XR AI Spotlight The Role of AI in 3D Animation
Feb 11, 2026
Alexander Grishanin, CTO of Cascadeur and veteran technical lead in game and animation tools. He explains Cascadeur’s AI vs machine learning, how their transformer-based in-betweening works, and why it’s considered generative. Learn about mixing neural generation with post-processing, local execution and privacy, recommended generate-then-refine workflows, and the roadmap for layers and facial animation.
AI Snips
Chapters
Transcript
Episode notes
AI Continues A Long Lineage Of Motion Tools
- Alexander places Cascader's AI alongside decades of procedural and physics-driven animation rather than as an isolated invention.
- Prior systems like mocap, walk generators, and physics-muscle simulations all generated motion in different ways.
Fix AI With Physics-Based Cleanup
- Combine physics tools with AI output to fix physics errors like foot placement and improve realism.
- Use physics-based cleanup after generation because AI alone struggles with accurate physical interactions.
Use Sparse Keys Then Unbake For Control
- Start with sparse key poses, enable AI in-betweening, then add intermediate keyframes to refine motion.
- When keys get dense, use the unbake tool to lock AI interpolation into editable standard keyframes.