notesum.ai
Published at November 25Generative Omnimatte: Learning to Decompose Video into Layers
cs.CV
Released Date: November 25, 2024
Authors: Yao-Chih Lee1, Erika Lu2, Sarah Rumbley2, Michal Geyer2, Jia-Bin Huang1, Tali Dekel2, Forrester Cole2
Aff.: 1University of Maryland College Park; 2Google DeepMind
![[Uncaptioned image]](https://arxiv.org/html/2411.16683v1/x1.png)
| Scene | Movie | Kubric | Average | |||
|---|---|---|---|---|---|---|
| Metric | PSNR | LPIPS | PSNR | LPIPS | PSNR | LPIPS |
| ObjectDrop [51] | 28.05 | 0.124 | 34.22 | 0.083 | 31.14 | 0.104 |
| Propainter [64] | 27.44 | 0.114 | 34.67 | 0.056 | 31.06 | 0.085 |
| Lumiere inpainting [6] | 26.62 | 0.148 | 31.46 | 0.157 | 29.04 | 0.153 |
| Ominmatte [34] | 21.76 | 0.239 | 26.81 | 0.207 | 24.29 | 0.223 |
| LNA [21] | 23.10 | 0.129 | - | - | - | - |
| Omnimatte3D [45] | - | - | - | - | - | - |
| OmnimatteRF [27] | 33.86 | 0.017 | 40.91 | 0.028 | 37.38 | 0.023 |
| Ours | 32.69 | 0.030 | 44.07 | 0.010 | 38.38 | 0.020 |