Stable Fast 3D (SF3D) is a rapid 3D asset generation tool by Stability AI that transforms a single 2D image into a detailed 3D asset in about 0.5 seconds. It emphasizes speed and quality for 3D reconstruction, enabling high-quality UV-unwrapped meshes and material parameters from a single input image. SF3D is suited for gaming, VR, e-commerce, and architectural visualization, and is built on architectural improvements over the TripoSR framework. Licensing follows the Stability AI Community License, allowing non-commercial use and commercial use up to $1M annual revenue for individuals or organizations. For larger enterprises, Enterprise Licenses are available upon request.

What is Stable Fast 3D?

Converts a single image into a textured 3D asset in under a second (0.5 seconds per asset on a GPU with 7GB VRAM).
Produces UV-unwrapped meshes with accurate material parameters.
Designed for fast workflows in gaming, VR, e-commerce, and architectural visualization.
Built on the TripoSR architecture with significant improvements for speed and quality.

How to Use Stable Fast 3D

Upload an image (JPEG, PNG, or WebP) to generate a 3D asset.
The system returns a GLB/GLTF-style binary blob containing a glTF asset with JSON, buffers, and textures.
Use the generated 3D asset in your pipeline or applications.

API & Usage

API Access: Integrate Stable Fast 3D into your applications via API.
How to Use the Stable Fast 3D API:

Make a POST request to the API endpoint.
Include your API key in the Authorization header.
Set Content-Type to multipart/form-data.
In the request body, include:

image (required): input image (JPEG, PNG, WebP)
texture_resolution (optional): 512, 1024 (default), or 2048
foreground_ratio (optional): 0 to 1 (default: 0.85)
Input image requirements:
Minimum side length: 64 pixels
Total pixel count: 4,096 to 4,194,304
Output: A binary blob containing a glTF asset (GLB format).
Credits: Each successful generation costs 2 credits; failed generations are free.

Example API call:

curl -X POST 'https://api.stability.ai/v1/generation/stable-fast-3d' \
 -H 'Content-Type: multipart/form-data' \
 -H 'Authorization: Bearer YOUR_API_KEY' \
 -F 'image=@path/to/your/image.jpg' \
 -F 'texture_resolution=1024' \
 -F 'foreground_ratio=0.85'

Licensing & Availability

Model is released under the Stability AI Community License.
Allowed use:
Non-commercial use
Commercial use for individuals or organizations with up to $1M in annual revenue
For enterprises above $1M revenue, contact stability AI for Enterprise Licenses.
Access methods:
API
Hugging Face demo and spaces
Model code on GitHub and weights on Hugging Face

Alternatives

Google’s DreamFusion: Text-to-3D generation alternative.
NVIDIA’s GET3D: Textured 3D meshes from images.
OpenAI’s Point-E: 3D point clouds from text descriptions.
NVIDIA’s Magic3D: Text-to-3D model for high-quality results.

Note: Stable Fast 3D emphasizes speed and quality of output from single images, making it competitive for rapid 3D asset creation.

How It Works

Input an image; the system generates a 3D asset with a UV-unwrapped mesh and materials.
Output is a GLB/GLTF-compatible binary blob containing the 3D asset.
The approach focuses on fast processing (0.5 seconds per asset) with high-quality textures and materials.

Safety & Legal Considerations

Ensure proper licensing for any assets used as input and comply with usage rights for generated content.

Core Features

Ultra-fast 3D asset generation: ~0.5 seconds per asset on suitable GPUs
Single-image input → detailed UV-unwrapped 3D meshes with material parameters
Output format: GLB/GLTF-compatible binary blob
Texture resolution options: 512, 1024 (default), 2048
Foreground ratio control for padding around the object (default 0.85)
API access for integration into apps and pipelines
Stability AI Community License with clear commercial/non-commercial terms
Availability via API, Demo spaces, and open-source/community channels

Stable Fast 3D by Stability AI

Introduction

Tags

Featured

Dora Studio

n8n

Lovable

Wan AI