---

name: services-and-integrations description: "Operate Hunyuan3D-2 FastAPI, Gradio, Blender add-on, and client integration workflows." metadata: disco-role: operating disable-model-invocation: true license: NOASSERTION

Hunyuan3D-2 Services and Integrations

Use this sub-skill when the task is about serving Hunyuan3D-2 through its FastAPI worker, launching the Gradio app, sending client requests, integrating with Blender, or understanding REST payloads and service flags.

Route here for

• api_server.py launch flags, /generate, /send, and /status/{uid} behavior.
• Building base64 JSON payloads for image-to-3D, text-to-3D, existing-mesh texturing, and textured output.
• gradio_app.py model/subfolder flags, low-VRAM mode, FlashVDM mode, texture toggle, and cache/export behavior.
• Blender add-on installation/use at the level of API URL, image path, selected mesh, texture flag, and job status.
• Debugging local server connectivity, ports, CORS, binary GLB responses, status polling, and service-side CUDA failures.

Do not route here for

• Low-level Hunyuan3D-DiT generation parameters except request payload routing: use ../shape-generation/.
• Hunyuan3D-Paint internals, custom rasterizer, mesh cleanup, and texture quality: use ../texture-and-mesh/.
• Installation, CUDA extension build order, and model cache setup: use ../environment-and-model-setup/.

Essential references

• Gradio and API for service launch commands, flags, endpoint schemas, and client workflows.
• Blender add-on for Blender properties, payload behavior, and limitations.
• Troubleshooting for service startup, HTTP, payload, queue, and CUDA/runtime errors.

Bundled helpers

• scripts/request_api.py builds or sends Hunyuan3D-2 API requests. It supports --dry-run for safe payload inspection.
• scripts/launch_api_server.py is a compact self-contained FastAPI launcher adapted from the repository server semantics.
• scripts/launch_gradio_app.py is a compact self-contained single-image Gradio launcher for installed Hunyuan3D environments.

Example dry-runs:

python scripts/request_api.py --image input.png --output mesh.glb --dry-run
python scripts/request_api.py --mesh mesh.glb --image ref.png --texture --dry-run
python scripts/request_api.py --mode status --uid 00000000-0000-0000-0000-000000000000 --dry-run
python scripts/launch_api_server.py --dry-run
python scripts/launch_gradio_app.py --dry-run

Omit --dry-run only when model loading, CUDA use, and possible checkpoint downloads are intended.