Modify the improvement factor

[BOUTAYEBAnasse]

Hi, I'm wondering if it's possible to train a model with an improvement factor of 2 instead of 4. I'm talking about this modification on the script side (train.py and constants.py) ?

[remicres]

hi @BOUTAYEBAnasse

yes you can train a model with less upsampling ratio.

note that the model provides also the 1x, 2x and 4x upsampled outputs separately

[BOUTAYEBAnasse]

Dear Rémi, Thank you very much for your reply.

By the way, the fake upsampled images *1 and *2 are saved under which directory?

[remicres]

sr.py does not currently write this output. I just have checked and unfortunately it is not exported in the savedmodel.

you have to modify the code