TRELLIS.2 Text to 3D
Text to 3D · Text prompt or uploaded reference image
About TRELLIS.2 Text to 3D
This community Space adds a text-to-image step before TRELLIS.2 reconstruction. A prompt first produces a reference image using Z-Image; TRELLIS.2 then uses that image to generate the 3D asset.
You can also skip text generation by uploading an image directly. The interface provides resolution, target-face-count, texture-size, and sampler settings, plus a Rerun preview and GLB download. This is a separate Space maintained by prithivMLmods, rather than Microsoft’s original demo.
How to use TRELLIS.2 Text to 3D
- Enter a prompt and select Generate Image, or upload an image in Image-to-3D mode.
- Review the reference, adjust the 3D settings if needed, and select Generate 3D.
- Inspect the result and select Download GLB File.
What to know before you start
- Text-to-3D runs two model stages. Check the generated reference before spending resources on reconstruction.
- An uploaded image bypasses the text-to-image stage but still uses the 3D generation service.
The provider controls availability, sign-in, processing, and downloads. Save any result before reloading or closing the demo.
Does TRELLIS.2 itself read the text prompt here?
In this Space, the text prompt first becomes an image. TRELLIS.2 reconstructs the asset from that image.
