From text to FBX file in less than a minute, without a camera

Viggle's PINOC translates a written prompt into an animated FBX file in less than a minute. What this text-to-motion model changes for shot blocking.

Describing a motion in a single sentence is now enough to get a 3D animation in PINOC, Viggle's motion capture tool. Until now, the application started from a video to extract an animated skeleton. It now accepts a written prompt, processed by an in-house developed text-to-motion model. You type the motion, set the duration and the number of takes, then export. According to the company, the journey from prompt to FBX file takes less than a minute, without any reference video.

The model performs best on high-energy and expressive action: sports, combat, dance, performance. Two settings frame the rendering. An image uploaded as the first frame starts the motion from a precise pose, and the same principle applies to the last frame to end exactly where you want. Before exporting, you can upload a character image, from which PINOC reconstructs a 3D model using Gaussian splatting, to preview the motion on it in the viewport.

The output comes in the form of an FBX or GLB file, on a 65-bone Mixamo rig ready for Blender, Maya, Unreal Engine, Unity, Cinema 4D, or Houdini. Body physics remain accounted for, including ground contacts, weight transfers, and range of motion. As for use cases, Viggle cites shot blocking, gameplay prototyping, and populating a scene with NPCs. Currently free, access will remain so for a while, according to Viggle.