Thank you for your work. I just have one question about the implementation. I wonder if the chunk size could be reduced to one pixel, like the central pixel within the convolution. In this case, other neighboring chunks would become neighboring pixels in convolution. I am curious about the motivation of chunk within this implementation.
Thank you for your work. I just have one question about the implementation. I wonder if the chunk size could be reduced to one pixel, like the central pixel within the convolution. In this case, other neighboring chunks would become neighboring pixels in convolution. I am curious about the motivation of chunk within this implementation.