Skip to content

Reproducing the paper's method #10

Description

@xyf3626

Thanks for your excellent work! When I try to reproduce the training method based on your paper, I encountered a problem . The compressed textures seem unable to reconstruct high-frequency details. Below are my training code, comparison images, and loss curve. I would greatly appreciate it if you could tell me what is going wrong.
In addition, I also want to know is there a warmup phase while you are training, and how do you initialize the BC1 parameters.

training.py

Image Image

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions