Video Generation on SkyTimelapse
39.5FVDFrameDiT-H
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| FrameDiT-HNumber of frames=16, Resolution=256 × 2562026.03 | 39.5 | — | — | — | |
| AR-DiffusionNumber of frames=16, Resolution=256 × 256, Checkpoint=Official2026.03 | 40.2 | — | — | — | |
| FrameDiT-GNumber of frames=16, Resolution=256 × 2562026.03 | 40.6 | — | — | — | |
| AR-DiffusionNumber of frames=16, Resolution=256 × 2562026.03 | 40.8 | — | — | — | |
| LatteNumber of frames=16, Resolution=256 × 2562026.03 | 42.7 | — | — | — | |
| PVDMNumber of frames=16, Resolution=256 × 2562026.03 | 55.4 | — | — | — | |
| VIDMNumber of frames=16, Resolution=256 × 2562026.03 | 57.4 | — | — | — | |
| DDMIRepresentation Type=Continuous, Domain Specificity=Domain-agnostic2024.01 | 66.25 | — | — | — | |
| PVDMRepresentation Type=Discrete2024.01 | 71.46 | — | — | — | |
| StyleGAN-VNumber of frames=16, Resolution=256 × 2562026.03 | 79.5 | — | — | — | |
| StyleGAN-VRepresentation Type=Continuous, Domain Specificity=Domain-specific2024.01 | 79.52 | — | — | — | |
| DIGANNumber of frames=16, Resolution=256 × 2562026.03 | 83.1 | — | — | — | |
| DIGANRepresentation Type=Continuous, Domain Specificity=Domain-specific2024.01 | 83.11 | — | — | — | |
| LVDMRepresentation Type=Discrete2024.01 | 95.2 | — | — | — | |
| LVDMNumber of frames=16, Resolution=256 × 2562026.03 | 95.2 | — | — | — | |
| FVDMNumber of frames=16, Resolution=256 × 2562026.03 | 106.1 | — | — | — | |
| MOCOGAN-HDRepresentation Type=Discrete2024.01 | 164.1 | — | — | — | |
| MOCOGANRepresentation Type=Discrete2024.01 | 206.6 | — | — | — | |
| MoCoGANNumber of frames=16, Resolution=256 × 2562026.03 | 206.6 | — | — | — | |
| VideoGPTRepresentation Type=Discrete2024.01 | 222.7 | — | — | — | |
| VideoGPTNumber of frames=16, Resolution=256 × 2562026.03 | 222.7 | — | — | — | |
| Diffusion ForcingNumber of frames=16, Resolution=256 × 2562026.03 | 251.9 | — | — | — | |
| DIGANTraining cost (GPU-days)=162021.12 | — | 83.11 | 196.7 | — | |
| DIGANResolution=256x2562022.12 | — | 83.1 | 196.7 | — | |
| DIGAN2023.08 | — | 83.1 | 196.7 | — | |
| LongVideoGANResolution=256x2562022.12 | — | 116.5 | 152.7 | — | |
| LongVideoGAN2023.08 | — | 116.5 | 152.7 | — | |
| MoCoGANResolution=256x2562022.12 | — | 206.6 | 575.9 | — | |
| MoCoGANBackbone=StyleGAN2, Resolution=256x2562022.12 | — | 85.9 | 272.8 | — | |
| MoCoGAN2023.08 | — | 85.9 | 272.8 | — | |
| MoCOGANTraining cost (GPU-days)=52021.12 | — | 206.6 | 575.9 | — | |
| MoCOGAN + StyleGAN2 backboneTraining cost (GPU-days)=82021.12 | — | 85.88 | 272.8 | — | |
| MoCoGAN-HDResolution=256x2562022.12 | — | 164.1 | 878.1 | — | |
| MoCOGAN-HDTraining cost (GPU-days)=7.5 + 92021.12 | — | 164.1 | 878.1 | — | |
| StyleGAN-VTraining cost (GPU-days)=82021.12 | — | 79.52 | 197 | — | |
| StyleGAN-VResolution=256x2562022.12 | — | 73.9 | 248.3 | 40.8 | |
| StyleGAN-V2023.08 | — | 73.9 | 248.3 | 40.8 | |
| StyleSVResolution=256x2562022.12 | — | 49 | 135.9 | 49.9 | |
| StyleSV2023.08 | — | 49 | 135.9 | 49.9 | |
| StyleSV + MTMModule=MTM2023.08 | — | 42.3 | 124.6 | 44.5 | |
| TATS-baseResolution=256x2562022.12 | — | 132.6 | — | — | |
| VideoGPTTraining cost (GPU-days)=16 + 162021.12 | — | 222.7 | — | — | |
| VideoGPTResolution=256x2562022.12 | — | 222.7 | — | — |