Viggle releases JST-2, a 3D generation model for scenes and characters
What happened
Viggle released JST-2, a 3D generation model for scenes, characters, and full-body motion. The model generates a scene in roughly one second. It is free to try via the Viggle website, with an API available for teams building it into their own tools. The release also includes MCP support, which means JST-2 can be called as a tool from AI agent workflows and code editors that support the Model Context Protocol.
Why it matters for designers
Motion designers and 3D artists who have been waiting for generation speed to catch up with iteration speed now have a concrete benchmark to test against. One-second scene generation changes the economics of early-stage concepting: instead of spending time on rough sketches before committing to a direction, a designer can generate and discard options at a pace closer to ideation.
The MCP integration is notable because it moves 3D generation out of standalone tools and into the agent workflows some design teams are already building. A pipeline that today relies on hand-off between a language model and a separate 3D tool could instead call JST-2 as one step among others without leaving the environment.
For teams evaluating AI in their motion and 3D workflows, the combination of free access, an API, and MCP support makes JST-2 worth a direct test rather than a wait-and-see.