Text-prompted motion generation model that animates rigged 3D assets with arbitrary skeleton topologies, no retraining.
Updated EU status checked Last shown
Sources: the vendor's own pages (linked above) and creator videos at the linked timestamps. EU status is information, not legal advice; see how we check.
UniMate is a research model from Princeton (Linzhan Mou, Adam Finkelstein, Szymon Rusinkiewicz and co-authors) that generates articulated motion for rigged 3D assets from a text prompt. A single unified model handles skeletons of differing topology, including bipedal, quadrupedal, avian, marine, insectoid, serpentine and articulated rigid objects, conditioning directly on each target skeleton without per-skeleton retraining. It is presented at SIGGRAPH Asia 2026.
The project page shows the same prompt applied to different rigs, different prompts on one rig, and diverse samples for a fixed rig and prompt. Applications include motion editing (fixing a subset of joints while resampling others), in-betweening between start and end poses, and motion expansion from sequential prompts.
The authors also introduce UniML3D, a dataset of motion sequences with skeleton annotations and text prompts, built through a pipeline of clip filtering, LLM-assisted joint-name standardization, multimodal captioning with manual review, and canonicalization. A YouTube creator reported that code and training code are open source, but no repository or license link appeared in the provided material. The site states that its 3D models and animations are copyrighted and not licensed for download.
Watch the moment: the exact second top creators show it. Videos load only when you press play, via youtube-nocookie.com.
Animates rigged 3D models with arbitrary skeleton types (non-human characters) from a text instruction, without retraining. Code and training code are open source.
Closest neighbours by meaning, not just by category.
Model
Research framework that conditions one video world model on force, camera, hand pose and robot pose actions.
Model · NVIDIA
NVIDIA's real-time generative motion model for game characters and humanoid robots, with smart primitives for authoring.
Research
Research method that rigs animatable Gaussian characters with an inner kinematic skeleton and outer free-form bones.
Model
Unified motion-language model for text-to-motion, motion captioning, completion and editing, from HKUST and Tencent Games.
Model · Nvidia
Nvidia platform for generating character animations from text prompts, used as a source of reference motion.
Model · NVIDIA
NVIDIA autoregressive diffusion model for real-time, text- and constraint-controlled 3D human motion generation.
