notesum.ai
Published at November 27CAT4D: Create Anything in 4D with Multi-View Video Diffusion Models
cs.CV
Released Date: November 27, 2024
Authors: Rundi Wu1, Ruiqi Gao1, Ben Poole1, Alex Trevithick1, Changxi Zheng2, Jonathan T. Barron1, Aleksander Holynski
Aff.: 1Google DeepMind; 2Columbia University
![[Uncaptioned image]](https://arxiv.org/html/2411.18613v1/x1.png)
| Method | Fixed Viewpoint Varying Time | Varying Viewpoint Fixed Time | Varying Viewpoint Varying Time | ||||||
| PSNR | SSIM | LPIPS | PSNR | SSIM | LPIPS | PSNR | SSIM | LPIPS | |
| 4DiM[71] | 19.77 | 0.540 | 0.195 | 18.81 | 0.428 | 0.219 | 17.28 | 0.378 | 0.256 |
| Ours | 21.97 | 0.683 | 0.121 | 21.68 | 0.588 | 0.105 | 19.73 | 0.533 | 0.155 |