Generate 3D scenes from one or two images
Text-to-3D and Image-to-3D Generation
Generate protein structures from specified lengths and seeds
Create a dynamic 3D scene with random shapes and lights
Create 3D models from images using depth estimation
Explore a 3D map of Minnesota
Create a 3D model from an image in ~10 seconds!
Create an immersive 3D scene with moving lights
Generate dynamic 3D torus knot shapes
Explore Minnesota in a 3D VR environment
MoGe live demo
text-to-3D & image-to-3D
Create a 3D scene with random torus knots and lights
Splatt3R is a cutting-edge AI tool designed for 3D modeling that enables users to generate 3D scenes from one or two uncalibrated images. It leverages zero-shot Gaussian splatting technology to create immersive 3D environments without requiring calibrated camera setups or extensive training data. This innovative approach makes it accessible for users to transform 2D imagery into 3D models effortlessly.
• Zero-shot capability: Generate 3D scenes without prior training or calibration. • Gaussian splatting: Advanced technique for creating smooth and detailed 3D models. • Uncalibrated image support: Works with uncalibrated image pairs or even a single image. • User-friendly interface: Simplified workflow for seamless 3D scene generation. • High-quality output: Produces detailed and accurate 3D representations.
What is Gaussian splatting?
Gaussian splatting is a technique used to create 3D models by projecting features from 2D images into a 3D space using Gaussian distributions.
Can I use Splatt3R with only one image?
Yes, Splatt3R supports zero-shot generation from a single image, though using two images may improve accuracy.
How long does the generation process take?
The process is typically fast, but processing time depends on the complexity of the scene and the system's hardware capabilities.