Viggle's physics-native world foundation model that generates 3D scenes, rig-free characters and motion from text or images.
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.
JST-2 (Joint Space-and-Time, gen 2.0) is a world foundation model from Viggle. It represents the world as persistent atoms in 3D space, each with an xyz position and learned attribute and dynamics vectors, encoded as "Physics-Native Tokens". An encoder, generator and decoder work on these tokens. The decoder outputs images or 3D Gaussian splats.
According to Viggle, JST-2 generates a 3D scene in about one second at an inference cost under $0.01, using a feed-forward architecture with no per-scene optimization. It accepts up to 64 input views as spatial context, so a world can be expanded one view at a time. It can also reconstruct 3D environments from sparse real-world images. Characters can be generated from text or a single image with learned dynamics, so they are ready to animate without separate rigging and skinning. The model also generates full-body 3D motion from text or video, which can be applied to JST-2 characters or standard rigs.
Viggle reports blind human evaluations against Meshy 7.1 and Tripo H3.1 for character generation and animation readiness, and against Cartwheel S for text-to-motion. These are vendor-published results. Text-prompt-guided scene expansion and a higher-detail splat version are listed as upcoming. The model is proprietary and is accessed through Viggle's platform. A related PINOC product for 3D characters and motion has a free tier and paid plans, billed separately from Viggle AI.
Watch the moment: the exact second top creators show it. Videos load only when you press play, via youtube-nocookie.com.
World model that turns text or images into explorable 3D environments with characters and animations, kept consistent via 3D matter representation. It gets a higher win rate than Meshy and Tripo. Not open source but free to try on their platform.
Closest neighbours by meaning, not just by category.
Model
Real-time interactive world model with compressed memory and hybrid action and semantic control, built on Wan 2.2.
Model
Research code that streams video generation chunk by chunk from a rough 3D proxy, built on Code World Model.
Model · Runway
Runway's real-time interactive world model generating 720p/24fps video with audio, steered by text actions and camera input.
Library · World Labs
World Labs tooling that turns a single image into an editable 3D environment with meshes, splats, physics and lighting.
Model
14B autoregressive world model that generates interactive, camera-controlled video worlds in 3 denoising steps.
Model
World model that turns a few context images into an editable, physics-simulated 3D world rendered by a video model.
