Poor PSNR in Denoising of Salinas

[ayush-0p1]

While i was successfully able to reproduce your results for houston dataset using the checkpoint diffusion model you provided of around 36db , but getting only 32 using the same checkpoint in salinas ? Can you help me mitigate this issue ?

[ayush-0p1]

Or i can only use houston for denoising and salinas for inpainting , i am downloading original salinas_corrected.mat file and then adding noise myself for sigma = 50 and then only i am getting 32 psnr , but while i am using your provided houston file i am getting proper 36 db can you help me improve psnr in salinas denoising . so i can replicate your results .