Camera-controlled Video Generation on DL3DV 160 samples subset 10K (val)
0.984CLIP ConsistencyGWTF
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| GWTFgamma (γ)=0.72026.03 | 0.984 | 36.45 | 0.457 | 21.25 | 0.173 | 145.2 | |
| TTMtw (time window)=4, ts (time stride)=82026.03 | 0.98 | 23.5 | 0.382 | 15.69 | 0.147 | 157.2 | |
| TTMtw (time window)=4, ts (time stride)=92026.03 | 0.98 | 23.15 | 0.38 | 15.59 | 0.147 | 158.8 | |
| GWTFgamma (γ)=0.52026.03 | 0.975 | 26.08 | 0.36 | 15.31 | 0.149 | 118.5 | |
| GT2026.03 | 0.974 | — | — | — | — | — | |
| Coarse Video2026.03 | 0.973 | 11.46 | 0.276 | 15.55 | 0.22 | 41.3 | |
| Weighted h-Transform Sampling2026.03 | 0.972 | 11.45 | 0.272 | 13.26 | 0.143 | 38.7 |