Zj-BinXia/DiffIR

some question about model framework

Closed this issue · 1 comments

Thank you for your excellent work.

为什么将GT和LQ图像拼接在一起后需要进行四倍下采样呢?

Reduce computational consumption and make the model more practical