Results for FATE

Submission info

Full name Full-head Gaussian Avatar with Textural Editing from Monocular Video
Publication title FATE: Full-head Gaussian Avatar with Textural Editing from Monocular Video
Publication authors Jiawei Zhang, Zijian Wu, Zhiyang Liang, Yicheng Gong, Dongfang Hu, Yao Yao, Xun Cao, Hao Zhu
Publication venue CVPR 2025
Publication url https://arxiv.org/abs/2411.15604
Project link https://zjwfufu.github.io/FATE-page/
Hardware RTX 3090 x 1
Code Code available with license: MIT
Use other data No
Last edited 24 May, 2025

Monocular FLAME Avatar results

Unseen expressions, unseen viewpoints

PSNR SSIM LPIPS JOD AKD CSIM
Left camera 17.5653 0.7897 0.2685 5.0302 3.4836 0.7384
Center camera 21.3639 0.8298 0.1730 6.1113 2.6065 0.8727
Right camera 18.3933 0.8414 0.2198 5.5382 4.4751 0.6978
Total 19.1075 0.8203 0.2204 5.5599 3.5217 0.7696

Unseen expressions, train viewpoint

PSNR SSIM LPIPS JOD AKD CSIM
Train viewpoint 23.7410 0.8527 0.1348 6.6586 2.0433 0.8987

Generated Images

Subject ID Left camera Center camera Right camera Train camera Ground truth
393
404
461
477
486