Talking Head Video Generation on HDTF
14.76FIDIM-Portrait w/o residual
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| IM-Portrait w/o residualRender Speed=109 FPS, Resolution=512x512, Hardware=Nvidia A100 GPU2025.04 | 14.76 | 0.03 | 0.116 | 65.75 | |
| X-PortraitRender Speed=0.1 FPS*, Resolution=512x512, Hardware=Nvidia A100 GPU, Mechanism=Diffusion-based2025.04 | 15.66 | 0.034 | 0.119 | 132.9 | |
| IM-PortraitRender Speed=109 FPS, Resolution=512x512, Hardware=Nvidia A100 GPU2025.04 | 15.93 | 0.031 | 0.118 | 69.04 | |
| Follow-your-emojiRender Speed=0.5 FPS*, Resolution=512x512, Hardware=Nvidia A100 GPU, Mechanism=Diffusion-based2025.04 | 16.09 | 0.035 | 0.115 | 117.54 | |
| IM-Portrait w/o late-stageRender Speed=109 FPS, Resolution=512x512, Hardware=Nvidia A100 GPU2025.04 | 16.7 | 0.031 | 0.123 | 70.69 | |
| Face-V2VRender Speed=22 FPS, Resolution=256x256, Hardware=Nvidia A100 GPU2025.04 | 20.57 | 0.029 | 0.132 | 78.11 | |
| EmoPortraitRender Speed=7 FPS, Resolution=512x512, Hardware=Nvidia A100 GPU2025.04 | 24.77 | 0.039 | 0.189 | 140.51 | |
| Portrait4d-v2Render Speed=10 FPS, Resolution=512x512, Hardware=Nvidia A100 GPU2025.04 | 25.74 | — | — | 139.11 | |
| LetsTalk-L# Params=0.7B, Latency (ms)=402024.11 | 31.761 | — | — | — | |
| Hallo# Params=2.4B, Latency (ms)=1152024.11 | 44.343 | — | — | — | |
| LetsTalk-B# Params=0.3B, Latency (ms)=272024.11 | 46.24 | — | — | — | |
| EchoMimic# Params=2.2B, Latency (ms)=542024.11 | 55.791 | — | — | — | |
| AniPortrait# Params=2.6B, Latency (ms)=742024.11 | 78.284 | — | — | — | |
| SadTalker# Params=138M, Latency (ms)=232024.11 | 102.371 | — | — | — |