CompVis/geometry-free-view-synthesis
Is a geometric model required to synthesize novel views from a single image?
Leverages transformer architecture trained in VQGAN's discrete latent space to generate novel viewpoints without explicit 3D geometric priors or depth estimation. Supports multiple training configurations (explicit/implicit depth, feature-based, embedding-based) on RealEstate10K and ACID datasets, with pre-trained checkpoints available for interactive scene exploration via the `braindance` demo tool.
379 stars. No commits in the last 6 months.
Stars
379
Forks
34
Language
Python
License
MIT
Category
Last pushed
Apr 16, 2023
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/diffusion/CompVis/geometry-free-view-synthesis"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Related models
kyegomez/Gamba
Implementation of PyTorch: "GAMBA: MARRY GAUSSIAN SPLATTING WITH MAMBA FOR SINGLE-VIEW 3D RECONSTRUCTION"
perceivelab/trf-sg2im
Official PyTorch implementation of the paper Transformer-Based Image Generation from Scene...
tthinking/DATFuse
[IEEE TCSVT 2023] Official implementation of DATFuse: Infrared and Visible Image Fusion via Dual...
tthinking/YDTR
[IEEE TMM 2022] Official implementation of YDTR: Infrared and Visible Image Fusion via Y-shape...
tthinking/ITFuse
[PR 2024] Official implementation of ITFuse: An Interactive Transformer for Infrared and Visible...