Why they use such networks, even through GANs and diffusion models can also be used for image upscaling? Examples of such models include Nvidia DLSS and AMD FidelityFX Super Resolution (FSR).
Asked
Active
Viewed 46 times
1 Answers
0
CNN have a neighborhood bias. That is, inputs adjacent to the present input are likely more important than far away inputs. In domains such as images this often holds true and thus makes CNNs particularly effective.
foreverska
- 2,347
- 4
- 21