Hi. I've taken a look on Avisynth source (2.02) and it seemed to me that the Built-in BitBlt was based on memcpy() . Right ?? Since it's used in many,many filters i think it could be usefull to make an heavy optimised version. even if the speed gain is small (ie 10%) the overall result would be a good speedup for avisynth... if it's already heavy optimised please tell me... if not i'm going to create some optimised BitBlts. Tell me if you think it's a good idea to integrate them in avisynth. (but maybe i totally messed up something ???)