Extremely interested in your work

[xiewende]

Extremely interested in your work.
I noticed that the paper integrated MDIQA as a loss function into the Real-ESRGAN restoration model and conducted a lot of experiments.
I'd like to know how the MDIQA-based loss function was implemented. I don't see where the code is.
I look forward to your reply.

[YaoShunyu19]

Thank you for your interest!
You can try integrating MDIQA directly as a loss function in Real-ESRGAN and adjust the dimension weight ratios to achieve your desired visual outcomes. For instance, you might set the sharpness weight ratio to 2.0 while keeping the other weights at their default value of 1.0.
I plan to release the restoration code when I have some time in the near future.

[xiewende]

Thanks
I will try the method you mentioned.
In the meantime, I look forward to your code.

[Lettie-LIU]

Thank you for your interest!
You can try integrating MDIQA directly as a loss function in Real-ESRGAN and adjust the dimension weight ratios to achieve your desired visual outcomes. For instance, you might set the sharpness weight ratio to 2.0 while keeping the other weights at their default value of 1.0.
I plan to release the restoration code when I have some time in the near future.
I am interested and looking forward this code as well. Thanks!