What is GT when training RealBasicVSR

[DachunKai]

Because RealBasicVSR is evaluate by non-reference metric (NIQE, PI), what is the GT or loss used for training RealBasicVSR?

[Feynman1999]

Because RealBasicVSR is evaluate by non-reference metric (NIQE, PI), what is the GT or loss used for training RealBasicVSR?

use reds dataset's HR imgs as GT, and do some degradation as LR