Can we know the mapping (pixel mapping) from the output to the input image ?
Can we know the mapping (pixel mapping) from the output to the input image ?