xinntao / ESRGAN

ECCV18 Workshops - Enhanced SRGAN. Champion PIRM Challenge on Perceptual Super-Resolution. The training codes are in BasicSR.
https://github.com/xinntao/BasicSR
Apache License 2.0
5.91k stars 1.05k forks source link

Why there is no artifacts when training ESRGAN? #90

Open zzhan127 opened 4 years ago

zzhan127 commented 4 years ago

Other models, such as pix2pix, cycleGAN are common to get a result with artifacts. But the ESRGAN model has a great result without any artifacts? How did you achieve this? Thank you!

xinntao commented 4 years ago

How do the artifacts look like?

ESRGAN will also produce artifacts when the assumed downsampling kernel is different from the real one.