The average video is converted from YV12 to RGB24 (or RGB32) when it's imported into the editing program and back to YV12 when it's encoded for distribution, but are there any extra conversions when the fake avis are used? As far as I can tell there's no way to run a filtering script before Vfapi returns the RGB24 video, so (assuming that your script requires no conversion to YV12) the final video would have been converted four times. I did those four conversions on the original d2v using AviSynth and there was no visible difference, so this is just to satisfy my curiosity.
If I convert the d2v using Vfapi (I followed the instructions in the latest version of DGDecode), the resulting fake avi registers in VDM as uncompressed RGB24 and requires converting to YV12 before my script works. So what am I doing wrong?