Video input is planned but it'll require a lot of OSD-layer work. Since ComputerEyes is a slow scan device (it takes 6 seconds to do a full capture and you had to hold still for that time if you were using a video camera back in the day) "scanning" .PNGs seemed like a better solution.

When we emulate more capable stuff like the SGI Indy's live video input then having video in will be more useful.