Open exzombie opened 4 years ago
For simplicity, QArv always used contiguous cv::Mat. This reduces performance of swscale decoder. Consider lifting this limitation. Most of the code should not really depend on this IIRC, but this needs to be checked and tested.
cv::Mat
I'm also wondering whether going to cv::UMat would make sense …
cv::UMat
For simplicity, QArv always used contiguous
cv::Mat
. This reduces performance of swscale decoder. Consider lifting this limitation. Most of the code should not really depend on this IIRC, but this needs to be checked and tested.