Video Generation on Kinetics-600
4.3FVDDFoT
Evaluation Results
| Method | Links | |
|---|---|---|
| DFoTType=Diffusion, Resolution=128 x 1282025.10 | 4.3 | |
| MAGVIT-v2Type=Autoregressive, Resolution=64 x 642025.10 | 4.3 | |
| Rolling DiffusionType=Diffusion, Resolution=64 x 642025.10 | 5.2 | |
| CanvasMARType=Autoregressive, Resolution=64 x 64, NAR=122025.10 | 6.2 | |
| CanvasMARType=Autoregressive, Resolution=64 x 64, NAR=102025.10 | 6.3 | |
| CanvasMARType=Autoregressive, Resolution=64 x 64, NAR=112025.10 | 6.3 | |
| MAGVITType=Autoregressive, Resolution=64 x 642025.10 | 9.9 | |
| MAGIType=Autoregressive, Resolution=64 x 642025.10 | 11.5 | |
| VDMType=Diffusion, Resolution=64 x 642025.10 | 16.2 | |
| RaMViDType=Diffusion, Resolution=64 x 642025.10 | 16.5 | |
| TransframerType=Autoregressive, Resolution=64 x 642025.10 | 25.4 | |
| TrIVD-GAN-FPType=GAN, Resolution=64 x 642025.10 | 25.74 | |
| PhenakiType=Autoregressive, Resolution=128 x 1282025.10 | 36.4 | |
| CCVSType=Autoregressive, Resolution=64 x 642025.10 | 55 | |
| Video VQ-VAEType=Autoregressive, Resolution=64 x 642025.10 | 64.3 | |
| DVD-GAN-FPType=GAN, Resolution=64 x 642025.10 | 69.1 | |
| CogVideoType=Autoregressive, Resolution=64 x 642025.10 | 109.2 | |
| Video TransformerType=Autoregressive, Resolution=64 x 642025.10 | 170 | |
| Video-GPTParameters=3.8B, Architecture=Vanilla Transformer2025.05 | 315.4 | |
| SeineParameters=0.9B, Architecture=U-Net, Text Encoder=CLIP-VIT-L [59]2025.05 | 332.8 | |
| LVMParameters=7B, Architecture=Vanilla Transformer2025.05 | 356.48 | |
| Open-Sora-Plan v1.3.0Parameters=2B, Architecture=DiT, Text Encoder=mT5-XXL [86]2025.05 | 384.27 |