Generate 3D room layouts from RGB panoramas
Generate 3D models from images
generate any 3d looking art in seconds.
Explore Minnesota with a 3D video map
Select and view 3D objects from a dataset
Generate a dynamic 3D scene with rotating lights and knots
Generate dynamic 3D scenes with torus knots
Generate 3D models of various objects
Explore and vote on 3D arenas in a leaderboard
Gradio demo of CharacterGen (SIGGRAPH 2024)
Reconstruct 3D Gaussians from unposes images.
Play interactive 3D Pyramids game
MoGe live demo
3D Room Layout Estimation LGT-Net is a deep learning-based model designed to predict 3D room layouts from RGB panoramas. It leverages advanced neural networks to infer spatial structures, walls, floors, and ceilings from 2D panoramic images, providing a 3D reconstruction of indoor environments. This technology is particularly useful in fields like virtual reality, real estate, and gaming.
• Deep Learning Architecture: Utilizes neural networks to process and analyze RGB panoramas. • 3D Layout Generation: Converts 2D panoramic images into editable 3D room layouts. • Automatic Room Size Calculation: Estimates room dimensions and spatial relationships. • High Accuracy: Delivers precise wall and floor detection for accurate 3D models. • Integration Capability: Works seamlessly with 3D modeling tools for further customization.
What input format does LGT-Net accept?
LGT-Net requires RGB panoramic images as input, typically in formats like PNG or JPEG.
Do I need 3D modeling expertise to use LGT-Net?
No, the model is designed to be user-friendly. Even novice users can generate 3D layouts without advanced knowledge of 3D modeling.
What are the supported output formats for the 3D layouts?
The tool supports multiple formats, including OBJ, PLY, and FBX, making it compatible with various 3D software.