Do you make a huffyuv file for each resolution and then process those or do you do it all in one process chain?
Either way I'm sure a lot of time could be saved by using avisynth, though it sounds like you have your virtualdub filter chain all worked out so whatever works I guess