search

MrTornado24 / FENeRF

[CVPR 2022] FENeRF: Face Editing in Neural Radiance Fields
MIT License
225 stars 19 forks source link

poor performance of the Inversion #15

Closed s1752729916 closed 1 year ago

s1752729916 commented 1 year ago

请问在用新的portrait进行Inversion的时候，渲染出来的结果很差： python inverse_render_double_semantic.py exp_name 、/checkpoint/315000_generator.pth --image_path /dataset/cnn/0/com_imgs/0.jpg --seg_path /dataset/cnn/0/parsing_celebahq/masks1024x1024/0.jpg --background_mask --image_size 128 --latent_normalize --lambda_seg 1. --lambda_img 0.2 --lambda_percept 1. --lock_view_dependence True --recon 其中的image和mask如下图所示：

cat-shapecat-shapecat-shape

但是渲染出来的结果

是我哪里搞错了吗？ cat-shapecat-shape

MrTornado24 commented 1 year ago

Hi, you need to align and crop raw images into the CelebA format before inverse rendering. Please refer to https://github.com/LynnHo/HD-CelebA-Cropper to preprocess your images.