Enterprise Plans
AI Application Development
AI Translator API
AI SEO/GEO Service
Web Scraping Service
Log in
const fs = require('fs');
const { Readable } = require('stream');
const { finished } = require('stream/promises');
const main = async () => {
const {
model_mesh: { url, file_name },
} = await fetch('https://api.ai.cc/v1/image/generations', {
method: 'POST',
headers: {
Authorization: 'Bearer ',
'Content-Type': 'application/json',
},
body: JSON.stringify({
model: 'triposr',
image_url: 'https://upload.wikimedia.org/wikipedia/commons/7/7a/Basketball.png',
}),
}).then((res) => res.json());
const { body } = await fetch(url);
const stream = fs.createWriteStream(`./${file_name}`);
await finished(Readable.fromWeb(body).pipe(stream));
};
main();
import requests
def main():
response = requests.post(
"https://api.ai.cc/v1/images/generations",
headers={
"Authorization": "Bearer ",
"Content-Type": "application/json",
},
json={
"model": "triposr",
"image_url": "https://upload.wikimedia.org/wikipedia/commons/7/7a/Basketball.png",
},
)
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()
-
AI Playground
Test all API models in the sandbox environment before you integrate.
We provide more than 300 models to integrate into your app.
Product Detail
Introducing TripoSR: Revolutionizing Image-to-3D Reconstruction
TripoSR is a groundbreaking transformer-based model jointly developed by Stability AI and Tripo AI. Released on March 4, 2024, this Version 1.0 model sets a new standard for rapid 3D object reconstruction. It excels at generating high-quality 3D meshes from just a single RGB image, achieving this feat in under 0.5 seconds on an NVIDIA A100 GPU.
💡 Key Capabilities & Features
- 🚀 Ultra-Fast Generation: Achieves feed-forward 3D generation in under 0.5 seconds.
- 🧠 Advanced Transformer Architecture: Utilizes an efficient transformer design for sophisticated processing.
- ✨ High-Quality 3D Meshes: Produces detailed and accurate 3D mesh outputs.
- 🖼️ Single Image Input: Simplifies the process by requiring only one RGB image.
- 🏆 State-of-the-Art Performance: Leads in Chamfer Distance and F-score metrics against competitors.
🎯 Intended Applications
TripoSR is engineered for sectors demanding swift 3D visualization directly from 2D images. Its primary applications include:
- Entertainment: For quick asset creation in film and animation.
- Gaming: Rapid prototyping and generation of game assets.
- Industrial Design: Expedited visualization of product concepts.
- Architecture: Fast conversion of architectural sketches or photos into 3D models.
Being an image-to-3D model, TripoSR operates independently of human languages, making it universally applicable.
⚙️ Technical Deep Dive
Model Architecture
TripoSR's sophisticated architecture is a blend of transformer-based components, meticulously optimized for superior 3D reconstruction:
-
Image Encoder:
- Utilizes a DINOv1 pre-trained vision transformer.
- Transforms RGB images into latent vectors, capturing both global and local features.
-
Image-to-Triplane Decoder:
- A transformer-based decoder.
- Converts latent vectors into a triplane NeRF (Neural Radiance Field) representation.
- Employs advanced attention mechanisms to learn complex relationships across triplane components.
-
Triplane-based Neural Radiance Field (NeRF):
- The final stage generating the 3D representation.
- Specifically optimized for handling intricate shapes and textures effectively.
Training Data
The model was rigorously trained on a carefully curated subset of the expansive Objaverse dataset, specifically chosen for its focus on realistic and high-quality 3D models.
Performance Metrics
TripoSR consistently surpasses other open-source alternatives in both quantitative and qualitative assessments. It demonstrates exceptional performance in critical metrics such as Chamfer Distance and F-score across a wide array of diverse datasets.
📊 Comparison to Other Models
- ✔️ Accuracy: Offers superior 3D reconstruction quality compared to other open-source solutions.
- ⚡ Speed: Generates complex 3D meshes in under 0.5 seconds on an NVIDIA A100 GPU.
- 🛡️ Robustness: Exhibits remarkable adaptability to various imaging conditions by inferring camera parameters dynamically, without relying on explicit conditioning.
🛠️ Usage & Licensing
Code Samples
For developers looking to integrate TripoSR, code samples and documentation are available. You can explore TripoSR's official announcement and resources for detailed implementation guides.
Ethical Guidelines & Licensing
TripoSR is released under the permissive MIT License. This promotes open-source development and encourages responsible implementation across various fields including AI, computer vision, and computer graphics applications. The MIT License permits broad usage, encompassing commercial, personal, and research applications.
By harnessing the advanced capabilities of TripoSR, developers are empowered to forge powerful 3D reconstruction applications with unprecedented speed and accuracy. This breakthrough opens up exciting new possibilities across diverse domains that rely on rapid 2D to 3D conversion.
❓ Frequently Asked Questions (FAQ)
- Q: What is TripoSR?
- A: TripoSR is an advanced image-to-3D reconstruction model developed by Stability AI and Tripo AI. It generates high-quality 3D meshes from a single 2D RGB image very rapidly.
- Q: How fast can TripoSR generate a 3D model?
- A: It can generate a 3D mesh in under 0.5 seconds when running on an NVIDIA A100 GPU, making it one of the fastest solutions available.
- Q: What kind of input does TripoSR require?
- A: TripoSR only requires a single RGB image to perform 3D object reconstruction.
- Q: What are the main applications of TripoSR?
- A: It is ideal for applications in entertainment, gaming, industrial design, and architecture where rapid 3D visualization from 2D images is crucial.
- Q: What is the licensing for TripoSR?
- A: TripoSR is released under the permissive MIT License, which permits its use for commercial, personal, and research purposes, promoting open and responsible AI development.
Learn how you can transformyour company with AICC APIs
