How to train the model for x8 scale?

Thanks for creating this repo and very helpful blog post. I tried inference x2 and x4 of EDSR model and it is good. I’m changing source code to train for x8 model. In the upsampling method of model/edsr.py, I change it like this The training code of x8 run for 10000 step but the PSNR is just around 15. How should I update the EDSR training code for x8? Thanks in advance.

[krasserm]

I’d expect it is able to learn that to some extend.

[krasserm]

I fixed this issue by resizing the upscaled image to be the same shape as HR image.

A proper solution would be to center-crop the HR image, using the SR image size as crop window. Resizing recomputes all pixel values via (bilinear, bicubic, …) interpolation which may explain the low PSNR values you see during validation.

About reusing weights of lower factors, layers have to be named consistently so that weights from lower resolution models can be loaded

Exactly, this must be done for all model implementations. What you proposed for up-sampling looks good to me (didn’t see the “wrong” names in your first proposal).