Giters

lyndonzheng / TFill

[CVPR 2022]: Bridging Global Context Interactions for High-Fidelity Image Completion

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Same outputs for multiple results

Jackieqfh143 opened this issue · comments

commented

When testing the model on CelebA-HQ dataset, we got exactly same outputs for multiple results. Is there anything wrong?

commented

Hi @Jackieqfh143 For existing models, we do not provide multiple and diverse results as our previous PIC. Our improved version provides the multiple results using VQ, but it is still under review. Therefore, we do not update the corresponding code and model.

commented

Okay. Thanks for your reply!