← All models
TripoSplat is an open-source model from TripoAI / VAST AI Research that converts a single 2D image into high-quality 3D Gaussians using a novel learned density-control approach
Details
- Accepts
- image
Pricing
- Price
- 6.25 cr / request
Prices in credits (1 credit = $0.01).
Data schema
Input
| Field | Type | Description |
|---|---|---|
| seed | — | Random seed for reproducibility. If None, a random seed is generated. |
| image_urlrequired | string | URL of the input image to convert into a 3D Gaussian splat. |
| num_gaussians | integer | Target number of Gaussians in the output splat. Higher means more detail but larger files and slower rendering. Rounded to the nearest multiple of 32 internally. |
| output_format | stringenum: ply, splat | File format for the generated Gaussian splat. |
| guidance_scale | number | Classifier-free guidance strength. Values at or below 1.0 disable guidance; higher values increase detail and adherence to the input image, but too high can oversaturate colors. |
| num_inference_steps | integer | Number of flow-matching sampler steps. More steps improve fidelity with a roughly linear runtime cost. |
| enable_safety_checker | boolean | Enable safety checking of the input image. |
Output
| Field | Type | Description |
|---|---|---|
| seed | integer | The seed used for generation. |
| timings | object | Per-stage timings in seconds. |
| model_mesh | — | Generated 3D Gaussian splat file (.ply or .splat). |
| num_gaussians | integer | The actual number of Gaussians produced. |
| preprocessed_image | — | The background-removed RGB composite the model actually saw, rendered on a black background. |