Hunyuan World
Image-to-3D world generation that turns single images into explorable panoramic scenes.
Studio is in early access; availability resolves per project after sign-in.
- Endpoints
- 2
- Eligible
- 2
- Input schemas imported
- 2 / 2
- Delivery
- fal.ai
- Weights
- Not stated
- Developer
- Not stated in source
Overview
Hunyuan World 1.0 transforms single images into explorable 3D panoramas through direct image-to-world conversion, skipping mesh generation. Output is an interactive world file presenting a panoramic scene rather than a static mesh. Inputs are a single image plus semantic labels with foreground and class annotations. It targets game developers and virtual-environment creators needing spatial context.
Capabilities
- Single-image to explorable 3D panorama
- Direct conversion without mesh reconstruction
- Interactive world-file output
- Annotation-guided scene expansion
Best for
- Game environment prototyping
- Virtual tour generation
- Spatial scene expansion
Use cases
- Navigable scene mockups from one image
- Virtual tours from stills
- Spatial context for game levels
Endpoints
2 eligible. Cataloged is not the same as executable: Studio resolves which endpoints a project can run.
| Endpoint | Task | Catalog status | Inputs |
|---|---|---|---|
fal-ai/hunyuan_world | 3d generation | Eligible | 2 fields · requires image_url, prompt |
fal-ai/hunyuan_world/image-to-world | image generation | Eligible | 5 fields · requires image_url, labels_fg1, labels_fg2 |
Questions
Mesh or panorama output?
An interactive panoramic world file, not a static mesh.
What input does it need?
A single image plus semantic foreground and class labels.