3D Talking Head Generation on User Study
96Lip Sync Accuracy (S)CapTalk
Evaluation Results
| Method | Links | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| CapTalkComparison Baseline=UniTalker (Fan et al., 2024)2026.05 | 96 | — | — | — | — | — | — | — | — | — | 92 | 94 | |
| CapTalkComparison Baseline=MultiTalk (Sung-Bin et al., 2024)2026.05 | 94 | — | — | — | — | — | — | — | — | — | 94 | 88 | |
| CapTalkComparison Baseline=SelfTalk (Peng et al., 2023a)2026.05 | 92 | — | — | — | — | — | — | — | — | — | 96 | 98 | |
| CapTalkComparison Baseline=CodeTalker (Xing et al., 2023)2026.05 | 83 | — | — | — | — | — | — | — | — | — | 77 | 87 | |
| CapTalkComparison Baseline=DiffPoseTalk (Sun et al., 2024)2026.05 | 79 | — | — | — | — | — | — | — | — | 72 | 78 | 79 | |
| CapTalkComparison Baseline=ARTalk (Chu et al., 2025)2026.05 | 68 | — | — | — | — | — | — | — | — | 67 | 66 | 68 | |
| CapTalkComparison Baseline=FaceFormer (Fan et al., 2022)2026.05 | 63 | — | — | — | — | — | — | — | — | — | 75 | 73 | |
| DiffPoseTalk2026.01 | 4.4 | 2.3 | 3.4 | 3.5 | 3.8 | 3.8 | 3.9 | 1.7 | 2.1 | 3.5 | — | — | |
| MANGO2026.01 | 4.3 | 3.9 | 3.9 | 4 | 4.1 | 3.9 | 4 | 3.7 | 3.3 | 3.7 | — | — | |
| DualTalk2026.01 | 4 | 3.5 | 3.8 | 3.2 | 3.5 | 3.6 | 3.7 | 2.9 | 3 | 3.2 | — | — | |
| CodeTalker2026.01 | 2.1 | 2.6 | 2.6 | 2.8 | 1.8 | 2.2 | 1.3 | 1.2 | 1.5 | 2.8 | — | — |