Scene generation
Turn a text prompt into a 3D scene
Describe a place, generate its panorama, and download a Gaussian-splat scene. This guide covers the text-to-3D workflow in JST-2 Studio, an independent application using Viggle’s API.
Inspect real JST-2 examplesDescribe a place rather than a finished shot
A useful scene prompt gives the model a spatial brief. Name the setting, describe where the main objects sit, and choose a consistent lighting condition. A room with a clear floor and a few distinct objects is easier to assess than a crowded scene with several competing styles.
Editorial prompt suggestion, not a tested output: “A small pottery workshop with a central wooden table, shelves of clay bowls along the left wall, a clear walkway and soft daylight from tall windows.” Treat the first result as an interpretation, not an exact layout specification.
Generate, review, then keep the useful result
Sign in to the studio, enter your prompt and choose low or high quality. Check your credit balance before submitting. Generation is a task with a completion state; a request being accepted does not mean its files are ready to download.
Review the returned preview and panorama before deciding whether to keep the scene. If the composition misses your intent, change one part of the prompt at a time. This is an editorial workflow suggestion, not a promise that repeated prompts reproduce the same geometry.
A panorama and a spatial scene serve different jobs
The panorama is an image for inspecting the generated surroundings. The SPZ file contains the Gaussian-splat scene used by compatible 3D renderers. A panorama alone does not demonstrate how the scene behaves when the camera moves away from its original viewpoint.
Inspect the spatial output in a compatible viewer before using it in a project. Look for broken silhouettes, floating detail and surfaces that change unexpectedly between viewpoints. SPZ is not a promise of a clean polygon mesh or a manufacturing-ready model.
Keep the studio workflow separate from the research demo
Viggle’s broader release includes world expansion and reconstruction from real observations. Those capabilities do not add image inputs or expansion controls to this text-scene form. Use the official developer documentation for workflows beyond the controls shown here.
The studio charges 6 site credits per scene request and does not award credits just for registering. Viggle’s model inference measurements describe research conditions; they are neither this site’s retail price nor an end-to-end generation-time guarantee.
Questions & answers
Does text-to-3D generate a video?
This scene workflow returns scene assets rather than a finished video. A preview or panorama is an image; the SPZ is a spatial representation. Rendering a camera animation is a separate workflow in a compatible application.
Can I upload my own panorama here?
The studio scene form described here accepts a text prompt. Viggle documents a separate direct-API workflow using a supplied panorama. That does not imply the same input is available in this studio.
Can I download an OBJ, FBX or STL scene?
The scene output documented here is SPZ, alongside image outputs when returned. OBJ, FBX and STL are not guaranteed exports. Check your destination software’s splat support before generating assets for it.
Check the original sources.
Reviewed . Model claims are attributed to Viggle; practical suggestions are editorial guidance.