What is Stable Fast 3D?
Stable Fast 3D(SF3D), the latest innovation from Stability AI, transforms single images into detailed 3D assets in just 0.5 seconds. This groundbreaking technology sets a new standard for speed and quality in 3D reconstruction.
- Generate high-quality 3D assets from a single image
- Unprecedented speed: 0.5 seconds per asset
- UV unwrapped mesh and material parameters
- Reduced illumination bake-in for better textures
Summary Video of Stable Fast 3D
Key Features
Lightning-Fast Processing
Generate 3D assets in just 0.5 seconds on a GPU with 7GB VRAM.
High-Quality Output
Enjoy detailed UV unwrapped meshes and accurate material parameters.
Versatile Applications
Perfect for gaming, VR, e-commerce, and architectural visualization.
Advanced Technology
Built on TripoSR with significant architectural improvements.
Licensing Information
Stable Fast 3D is released under the Stability AI Community License .
This license allows for:
- Non-commercial use
- Commercial use for individuals or organizations with up to $1M in annual revenue
For organizations with annual revenue exceeding $1M, contact Stability AI for Enterprise Licenses.
Note: Always refer to the official license documentation for the most up-to-date and detailed licensing information.
Alternatives to Stable Fast 3D
Google's DreamFusion
Uses text-to-3D generation, offering a different approach to creating 3D assets.
NVIDIA's GET3D
Generates textured 3D meshes from images, with a focus on high-quality results.
Point-E by OpenAI
Generates 3D point clouds from text descriptions, offering a unique perspective on 3D generation.
Magic3D by NVIDIA
Another text-to-3D model that aims to create high-quality 3D models from text descriptions.
While these alternatives offer various approaches to 3D generation, Stable Fast 3D stands out for its exceptional speed and the quality of its output from single images.
Try Stable Fast 3D
If the demo doesn't load, you can visit it directly on Hugging Face .
Try Stable Fast 3D on Hugging Face
Experience the power of Stable Fast 3D for yourself. Upload an image and watch as it's transformed into a detailed 3D asset in seconds.
Launch Demo on Hugging FaceAPI Access
Integrate Stable Fast 3D into your own applications with our easy-to-use API.
View API DocumentationAPI Usage Guide
How to Use the Stable Fast 3D API
To use the Stable Fast 3D API, follow these steps:
- Make a POST request to the API endpoint.
- Include your API key in the
authorization
header. - Set the
content-type
header tomultipart/form-data
. - In the request body, include:
image
(required): The input image file (JPEG, PNG, or WebP)texture_resolution
(optional): 512, 1024 (default), or 2048foreground_ratio
(optional): A number between 0 and 1 (default: 0.85)
Input Image Requirements:
- Minimum side length: 64 pixels
- Total pixel count: 4,096 to 4,194,304 pixels
Output:
The API returns a binary blob containing a glTF asset, including JSON, buffers, and images in GLB format.
Credit Usage:
Each successful generation costs 2 credits. Failed generations are not charged.
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'
For more detailed information, please refer to the official API documentation.
Stable Fast 3D References & Availability
Stable Fast 3D is available through various channels:
- Model code on Github
- Model weights on Hugging Face
- Demo space on Hugging Face
Stable Fast 3D is released under Stability AI Community License. This allows non-commercial use and commercial use for individuals or organizations with up to $1M in annual revenue.
For enterprises above this threshold, contact us for Enterprise Licenses.
The model is also accessible via our API, and on Stable Assistant.
The Project page is available here.
Frequently Asked Questions (FAQ)
Q: What file formats can I use as input for Stable Fast 3D?
A: Stable Fast 3D accepts JPEG, PNG, and WebP image formats as input.
Q: What are the image size requirements for Stable Fast 3D?
A: Every side of the input image must be at least 64 pixels, and the total pixel count must be between 4,096 and 4,194,304 pixels.
Q: What is the texture resolution option?
A: The texture resolution determines the detail level of the albedo (color) and normal maps. Options are 512, 1024 (default), or 2048 pixels. Higher resolution provides more detail but results in larger asset sizes.
Q: What does the foreground ratio parameter do?
A: The foreground ratio controls the amount of padding around the object in the frame. A higher ratio (closer to 1) means less padding and a larger object, while a lower ratio increases padding. The default is 0.85.
Q: What format is the output in?
A: The output is a binary blob that includes a glTF asset, containing JSON, buffers, and images in the GLB file format.
Q: How many credits does each generation cost?
A: Each successful generation costs 2 credits. Failed generations are not charged.