Frequently Asked Questions (FAQ)

Hi3D API Usage Questions

What are the prerequisites for calling the API?

Before making production API calls, sign in to the Open Platform, activate a test resource package or purchase a production resource package, and create an API Key.

How do I purchase an API resource package?

After signing in, go to Resource Packages on the platform to purchase a package.

How do I obtain an API Key?

After signing in, create and manage API Keys on the API Key Management page.

How do I make an API call?

Use HTTPS with the POST method.

• Authentication: Put your API Key in the Authorization header to call the Get Token endpoint.
• Create task: Include the Token in the Authorization header when calling the Create Task endpoint, and provide required body parameters such as request type (geometry-only / phased / one-shot), image input type and files (single-view / multi-view), model resolution, target polygon count, and output format.

See the API Reference for full parameter details.

Concurrency & rate limits?

Default 30 concurrent tasks. For higher concurrency, contact apicontact@hi3d.ai.

How to check task status and download results?

Poll the task or configure a Webhook to receive status updates. When status=success, the response includes the generated asset ID, cover/preview info, and download URLs (model + textures).

Does the API support batch generation?

Each call creates one task only.

How to troubleshoot common errors?

1. Check network connectivity
2. Verify image format/size limits
3. Ensure sufficient credit balance
4. For multi-view, include the required front view
5. If the queue is busy, reduce concurrency or retry later

If issues persist, follow the error message guidance or contact apicontact@hi3d.ai.

Are credits charged on failed calls?

No. For server faults/timeouts/parameter-validation failures, credits are refunded or not deducted (timeout threshold: 60 minutes).

See the error code for the failure type.

Hi3D Model Questions

Generation modes and types?

• Modes: Image / Multi-view to 3D; Image to 3D Relief; 3D model split; 3D Model Multicolor.
• Types:
  • Image / Multi-view to 3D: Geometry-only; Texture-only (optional PBR); All-in-One Geometry and Texture Generation (optional PBR).
  • Image to 3D Relief: All-in-One Generation only.
  • 3D model split: All-in-One Generation only.
  • 3D Model Multicolor: All-in-One Generation only.

Input image/model format and size limits?

• Images
  • Formats: png, jpeg/jpg, webp.
  • Size: Up to 20 MB per image.
• Models
  • Formats: glb, stl, obj.
  • Size: Up to 200 MB per model.

Limits on the number of input images?

Limits vary by mode:

• Single-view to 3D: 1 image.
• Multi-view to 3D: 2-4 images; the front view is required. Back, left, and right views are recommended for a more detailed and complete 3D model.
• Image to 3D Relief: 1 image.
• 3D model split: 1 model.
• 3D Model Multicolor: 1 model.

Supported resolutions?

• General models: v1.5 (512³, 1024³, 1536³, 1536³ Pro); v2.0 (1536³, 1536³ Pro); v2.1 (1536fast, 1536pro); v3.0 (2048quality, 2048master).
• Portrait models: v1.5 (1536); v2.0 (1536pro); v2.1 (profast, 1536pro).
• Depth models: Base, Pro.
• Higher resolutions produce more detail and larger files, and require longer inference time.

Polygon count?

You can set a target polygon count from 100,000 to 5,000,000. Recommended face counts vary by resolution:

• 512³: 500,000
• 1024³: 1,000,000
• 1536³, 1536³fast, 1536³Pro: 2,000,000
• 2048³quality: 2,000,000
• 2048³master: 5,000,000

Output formats?

• 3D: obj, glb, stl, fbx, usdz, 3mf.
• Textures: jpg, png.
• Relief: exr, png, stl, glb, 3mf, bmp.
• Outputs can be used directly in game engines or for 3D printing.