CodeFormer
CodeFormer copied to clipboard
sczhou
[NeurIPS 2022] Towards Robust Blind Face Restoration with Codebook Lookup Transformer
@sczhou @JohanAR @chenxwh Truly excellent work! I am trying to reproduce the stage1,but the result can not reach the official vqgan checkpoint, official vqgan checkpoint's eye's or face's detail are...
Hello, I am getting this error when I run the command to process an image. Any solutions would be great! Thanks in advance.
Hello, dear author. Your work has been very impressive, and I greatly appreciate the contributions you have made. While using your code for the third stage of training, I encountered...
 As shown in the figure above, the authors claimed they adopted the sinusoidal positional embedding in their project. However, as shown in the code below, they use "nn.Parameter" instead....
when not using pre-computed latent codes, the hq_vqgan_fix is used for generating code. However, it is randomly initialized. There should be a model_path config in https://github.com/sczhou/CodeFormer/blob/master/options/CodeFormer_stage2.yml#L73.
Dear sczhou, I hope this message finds you well. I am writing to report a potential issue with the inference_codeformer.py script in the [Project/Tool Name] repository. Description of the Issue:...
Error: ``` Traceback (most recent call last): File "scripts/crop_align_face.py", line 205, in size_ = align_face(in_path, out_path) File "scripts/crop_align_face.py", line 177, in align_face img = img.resize((output_size, output_size), PIL.Image.ANTIALIAS) AttributeError: module 'PIL.Image'...
