In the paper, you state: "We adopt the first-order degradation pipeline from ResShift [54], which better simulates real-world degradation than the second-order version used in RealESRGAN." However, ResShift actually adopts the RealESRGAN degradation pipeline. Could you clarify what you mean by this? Does this imply that the second-order probability is set to 0?
In the paper, you state: "We adopt the first-order degradation pipeline from ResShift [54], which better simulates real-world degradation than the second-order version used in RealESRGAN." However, ResShift actually adopts the RealESRGAN degradation pipeline. Could you clarify what you mean by this? Does this imply that the second-order probability is set to 0?