Jianyi Wang
Jianyi Wang
Hi. I think your results are reasonable. StableSR sometimes may not show 100% fidelity given the powerful generation capability of the fixed Stable Diffusion. And I am also not sure...
Hi. I think it is because 4x enables more generalization while 1x provides more limits on the results.
Hi. First, CFW is used for further improve the quality of the latents from diffusion Unet. The info of LR is introduced by the encoder of autoencoder. Second, w is...
Hi~thanks for your interest of our work. For data generation, we do not have CFW so we actually set w=0. During training, you just need to set w=1.
For PSNR and SSIM, did you test on the **Y** channel? It seems you tested on **RGB** channel. For the second question: yes. Set W=1 during training.
> > For PSNR and SSIM, did you test on the **Y** channel? It seems you tested on **RGB** channel. > > Yes, we use RGB channel, you mean we...
I do not think there is a big problem with your data generation code and I am sorry that I am currently on vacation and cannot access the data on...
> > I do not think there is a big problem with your data generation code and I am sorry that I am currently on vacation and cannot access the...
Hi. The reason is that your data path is not correct. You need to change the path in the config file accordingly, e.g., [here](https://github.com/IceClear/StableSR/blob/main/configs/stableSRNew/v2-finetune_text_T_512.yaml#L138). For data details, you may refer...
Hi. The inputs should be the same size as gts.