triposr

This documentation is valid for the following list of our models:

• triposr

Model Overview

A transformer-based model designed for rapid 3D object reconstruction from a single RGB image, capable of generating high-quality 3D meshes in under 0.5 seconds on an NVIDIA A100 GPU.

Setup your API Key

If you don’t have an API key for the AI/ML API yet, feel free to use our Quickstart guide.

Submit a request

API Schema

post

Authorizations

Body

modelundefined · enumrequired

Options: triposr

image_urlstring · urirequired

output_formatstring · enum · default: glb

Options: glb, obj

do_remove_backgroundboolean

foreground_rationumber · max: 1 · default: 0.9

mc_resolutioninteger · default: 256

Responses

cURL

JavaScript

Python

HTTP

cURL

curl -L \
  --request POST \
  --url 'https://api.aimlapi.com/v1/images/generations' \
  --header 'Authorization: Bearer <YOUR_AIMLAPI_KEY>' \
  --header 'Content-Type: application/json' \
  --data '{
    "model": "triposr",
    "image_url": "https://example.com",
    "output_format": "glb",
    "do_remove_background": true,
    "foreground_ratio": 0.9,
    "mc_resolution": 256
  }'

201

201

No body

Example

import requests


def main():
    response = requests.post(
        "https://api.aimlapi.com/v1/images/generations",
        headers={
            # Insert your AIML API Key instead of <YOUR_AIMLAPI_KEY>:
            "Authorization": "Bearer <YOUR_AIMLAPI_KEY>",
            "Content-Type": "application/json",
        },
        json={
            "model": "triposr",
            "image_url": "https://upload.wikimedia.org/wikipedia/commons/thumb/2/22/Fly_Agaric_mushroom_05.jpg/576px-Fly_Agaric_mushroom_05.jpg",
        },
    )

    response.raise_for_status()
    data = response.json()
    url = data["model_mesh"]["url"]
    file_name = data["model_mesh"]["file_name"]

    mesh_response = requests.get(url, stream=True)

    with open(file_name, "wb") as file:
        for chunk in mesh_response.iter_content(chunk_size=8192):
            file.write(chunk)


if __name__ == "__main__":
    main()

Response:

The example returns a textured 3D mesh in GLB file format. You can view it here.

For clarity, we took several screenshots of our mushroom from different angles in an online GLB viewer. As you can see, the model understands the shape, but preserving the pattern on the back side (which was not visible on the reference image) could be improved:

Compare them with the reference image:

Try to choose reference images where the target object is not obstructed by other objects and does not blend into the background. Depending on the complexity of the object, you may need to experiment with the resolution of the reference image to achieve a satisfactory mesh.