Image to 3D
From 10.8 credits
Turn one image into a textured 3D model. Rotate it in your browser and download GLB, OBJ, or USDZ.
Complete to generate
Runs on
-
ByteDance
Examples
Pyra as a collectible figurine
Seed3D 2.0. One studio photo of a Pyra figurine becomes a textured 500K-face model. Turn it to see the wings and the tail curled around its feet.
How to use Image to 3D
A compact decision guide for placing this pipeline in a workflow.
Best for
- Product, prop, figurine, and character models from one photo or render
- 3D assets for Blender, game engines, web viewers, and AR
Image to 3D API
Call this pipeline from your own code. One request dispatches a run; the model is an input field, not a separate endpoint.
Get an API token- inputs
- 3
- required
- 1
- Pricing
- 10.8 credits
| Input | Type | default | Description |
|---|---|---|---|
image_url | string | default— | Upload one image of the object. JPEG, PNG, WebP, BMP, TIFF, or GIF; 300–6000 px per side; up to 20 MB. |
face_count | enum | default500000 | More faces keep finer surface detail and make larger files.100000 · 500000 · 10000001000005000001000000 |
output_format | enum | defaultglb | GLB suits most 3D apps, game engines, and the web. OBJ downloads as a ZIP with its material and texture. USDZ opens in AR on iPhone and iPad.glb · obj · usdzglbobjusdz |
Point any MCP client at pipe2 and your agent gets these tools. It finds this pipeline, reads the same schema above, then runs it.
list_pipelinesget_pipeline_schemarun_pipelineget_pipeline_run_statusrequest_upload
https://mcp.pipe2.ai/mcp {
"mcpServers": {
"pipe2ai": {
"url": "https://mcp.pipe2.ai/mcp",
"headers": { "Authorization": "Bearer YOUR_TOKEN" }
}
}
}Frequently Asked Questions
What do I get?
One textured 3D model with PBR materials in the format you choose. GLB and USDZ are single files. OBJ downloads as a ZIP with the mesh, its material file, and a color texture.
Which images work best?
One object that is fully visible, sharp, and evenly lit, such as a product, figurine, toy, piece of furniture, or character, on a plain background. Upload one JPEG, PNG, WebP, BMP, TIFF, or GIF image, 300 to 6000 pixels per side, with a width-to-height ratio from 2:5 to 5:2, up to 20 MB.
Which polygon count should I choose?
100K faces makes the smallest files for web, AR, and real-time use. 500K, the default, suits most uses. 1M keeps the finest surface detail and makes the largest files.
Which file format should I choose?
Choose GLB for Blender, game engines, and the web; it keeps the full PBR material. Choose OBJ when your tool needs it; its material carries only the color texture. Choose USDZ for AR Quick Look on iPhone and iPad.
Can I preview the model before downloading it?
Yes. The result opens in an interactive 3D viewer where you can rotate and zoom it.
How long does it take?
Usually a few minutes, and sometimes more than ten when demand is high. You can leave the page; the model is saved to your library when it is ready.
How are credits charged?
Each model has a fixed price that does not depend on the polygon count or file format. If generation fails, your credits are returned.
AI Image to 3D Model Generator
Upload one image of an object and get a textured 3D model with PBR materials. Rotate and zoom the result in your browser, then download it for Blender, game engines, the web, or AR.
What you get
- GLB: one file with the mesh and PBR textures. It opens in Blender, Unity, Unreal Engine, Godot, three.js, and most web viewers.
- OBJ: a ZIP with the mesh, its material file, and a color texture, for 3D tools that prefer OBJ.
- USDZ: one file for AR Quick Look on iPhone and iPad and for Apple 3D tools.
How it works
- Upload an image: one object, fully in frame, on a plain background works best.
- Choose a polygon count: 100K for web and AR, 500K for most uses, or 1M for the finest surface detail.
- Choose a file format and run. A model usually takes a few minutes.
Tips for a better model
- Show the whole object, sharp and evenly lit, without heavy shadows.
- Remove a busy background first with AI Background Remover.
- The sides you cannot see in the image are inferred, so check the back of the model before you publish it.