Search papers, labs, and topics across Lattice.
Sorbonne University
1
0
3
Achieve 4K image-to-video generation with diffusion models without training by cleverly fusing tiled denoising with a low-resolution latent prior, balancing detail and global coherence.
