3D Camera-Controlled Video Synthesis on RealEstate10K (unseen camera trajectories)
0.358TransErrAC3D
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| AC3DArchitecture=VDiT2024.11 | 0.358 | 0.035 | 1.18 | 36.55 | 28.76 | |
| AC3DArchitecture=CogVideoX2024.11 | 0.374 | 0.039 | 1.27 | 38.2 | 28.62 | |
| VD3DArchitecture=FIT2024.11 | 0.409 | 0.043 | 1.4 | 42.43 | 28.07 | |
| VD3DArchitecture=VDiT2024.11 | 0.421 | 0.056 | 1.21 | 38.57 | 28.34 | |
| CameraCtrlArchitecture=U-Net2024.11 | 0.465 | 0.089 | 2.48 | 55.64 | 26.81 | |
| VD3DArchitecture=CogVideoX2024.11 | 0.467 | 0.063 | 1.66 | 43.14 | 28.08 | |
| MotionCtrlArchitecture=U-Net2024.11 | 0.477 | 0.094 | 2.99 | 61.7 | 26.46 | |
| MotionCtrlArchitecture=VDiT2024.11 | 0.504 | 0.126 | 1.74 | 43.81 | 27.69 | |
| CameraCtrlArchitecture=VDiT2024.11 | 0.513 | 0.138 | 1.62 | 42.1 | 27.73 |