While I had those old Super 8 films out, I thought I'd have a go at scanning them with the flatbed scanner I bought last year. The Canon Lide 500F has a highest resolution of 2400x4800 dpi, so assuming the Super 8 frame to be around 5mm wide, then this ought to give a scanned image that is around 500-900 pixels in size. The scanner comes with a Film Adaptor Unit (FAU) that scans with additional infra-red to perform dust and scratch fixing automatically, and the images came out pretty well.
The strip scanned here (see left) is as much as the diminutive Lide 500F FAU can scan in one go; it's designed for scanning 35mm stills of course. Scanning a movie this way could become very tedious.
Some folks out there have tried this already; some have even devised home-made frame advance mechanisms to automate the process, scanning a strip of film at a time, then using software to break the strips into separate frames. My scanner only scans a short strip of around 10 frames at a time, so it might be better to scan individual frames (avoiding the need for custom-made software to split-up the images). I might possibly then need to use After Effects to stabilize the images afterwards.
I've made a few first moves on this project, but it's a long way from working yet. Firstly I've acquired a stepper motor from an old Microtek flatbed scanner. The stepper motor has a step size of 1.8 degrees and came with a handy gearing mechanism which ratios this down even more. (I looked online to find out how the 6 wires should be connected up). I've made a film channel to fit inside the 35mm film tray of the FAU from strips of card that guide the 8mm film down the centre of the film unit. I've glued an old 35mm film core onto the stepper motor's output gear and this happens to fit very snuggly onto the FAU tray such that a rubber band on the film core contacts the film and slides it across the scanner glass. Now I've just bought a USB stepper motor driver board (StepperBee) and have used the Autostep software it came with to successfully transport the film through the unit.
The mechanism actually does advance the film! It takes approximately 13 steps of the motor to move the film one frame. The problem though is that; 'approximately'... Unfortunately the mechanism isn't accurate enough to transport the film repeatedly by the required distance. The distance the film should move is somewhere inbetween 13 and 14 steps (4.23mm to be precise). The 35mm core I have used has a large diameter of 75mm so transmits too much rotational movement with each step; a much smaller drive wheel might just work though...