The two output images are different. Maybe because the PIL module renormalize the output while our C++ code just use a cutoff to get rid off the channels with value greater than 255. A good idea should be to renormalize the pixel of the output image after the transfer.
The two output images are different. Maybe because the PIL module renormalize the output while our C++ code just use a cutoff to get rid off the channels with value greater than 255. A good idea should be to renormalize the pixel of the output image after the transfer.