Work in progress, posted as it happens. Newest first.
#3
Talk to her and a cinematic conversation begins — the MetaHuman "Informant" NPC
Approach and talk, and the camera cuts to a 50mm close-up with cinematic letterboxing. Lip sync and facial animation are generated from the voice audio with MetaHuman Animator, and subtitle timing is split automatically from the waveform.
Daifuku's Grooming Salon — 1.2 million hair strands in real time
Our fluffy pet Daifuku takes a one-minute bath, from shower to blow-dry. About 1.2 million strands are rendered with UE5.8 Groom, with wet clumping, foam and blow-dried fur driven by physics simulation and Deformer Graph.
Motion that never breaks — text-to-3D animation with NVIDIA Kimodo
We run NVIDIA's text-to-motion model Kimodo locally and built it into our pipeline. Passing the last 5 frames of each clip as full-body keyframe constraints lets generated motions chain seamlessly. Exported as FBX for Unity, Unreal Engine and Godot.