As far as I know it's possible to reuse the weights of a convolution in a transposed convolution to upsample an image. However when reusing the weights, the resulting restored images aren't even close to the original ones.
Examples:
- https://naokishibuya.medium.com/up-sampling-with-transposed-convolution-9ae4f2df52d0
- https://www.machinecurve.com/index.php/2019/09/29/understanding-transposed-convolutions/
Why does it make sense to reuse the weights of a normal convolution in a transposed convolution?
