3D Generation
Generate 3D meshes (GLB/OBJ) from an input image, on the generative backend.
No active public 3D endpoint today
There is currently no active 3D endpoint in the catalog. The image-to-3D model
hf-tencent-Hunyuan3D-2.1 (tencent/Hunyuan3D-2.1) is registered but inactive. To use 3D
generation now, import an image-to-3D model from HuggingFace
with task_type matching the model, then promote it after a successful test.
Verify current 3D availability in the live catalog:
curl "https://api.colabhive.com/api/builder/v1/endpoints?visibility=public&task_type=image-to-3d"
How image-to-3D works here
When a 3D endpoint is active, you upload a source image, run inference, and download the resulting mesh (plus optional PBR textures):
from colabhive import ColabHive
client = ColabHive(api_key="hive_...", account_id="YOUR_ACCOUNT_ID")
image_url = client.endpoints.upload_input("YOUR_3D_ENDPOINT_ID", "./chair.png")
result = client.endpoints.infer(
endpoint_id="YOUR_3D_ENDPOINT_ID",
input_data={"image_url": image_url, "format": "glb"},
)
for art in result.get("result", {}).get("output_artifacts", []):
client.endpoints.download_artifact(art["url"], f"./{art['filename']}")
Input parameters (image-to-3d)
| Parameter | Type | Default | Description |
|---|---|---|---|
image_url | string | required | Source image URL (use upload_input first) |
format | string | glb | Output format: glb or obj |
texture | boolean | true | Generate PBR textures (model-dependent) |
Output format
{
"output_artifacts": [
{ "filename": "model.glb", "url": "https://storage.colabhive.com/...",
"content_type": "model/gltf-binary", "size_bytes": 5242880, "modality": "3d" }
],
"metadata": { "format": "glb" }
}
.glb files open in Blender, Unity, Unreal, or any glTF viewer.
Related
Authors: José Luis Minich, Maximiliano Lucius.