Open panwangjun opened 2 years ago
Thank you! We used some SOTA SISR algorithms to up-sample the 2K images into higher resolution.
Now we note that there are DIV8K datasets, we recommend you could try the higher resolution images for evaluation.
Thank you! I have noticed that.
In the degradation extraction stage, I used the two PNG images and the original code provided by you, and the extracted PSF was very strange and felt abnormal. How does this approach ensure that the GT graph is consistent with the print? Thank you!
A very good job! Could you tell me, How does the DIV2K dataset match the resolution of the Sensor?