Raspberry PI 8mm Scanner

My family had a few roles of 8mm films they wanted to scan, and I decided 3d print a version I found on Cults3D. While it kinda worked, it had plenty of drawbacks. Slow; only worked for 8mm not Super 8 without modification; and my print was jenky as all get out. So I designed my own, and this is that story, and preview 8mm scan at end (not an instructional step by step!)

How it started

After two weeks of printing, readjusting in blender (my 3D printer has an opinionated idea of clearances), and assembly, I finally had my first 8mm scanner!

It took about 9 seconds for each frame. A five inch reel could have around 10,000 frames…so a solid 25 hours of scanning time. That’s if everything goes right (Spoiler, it never did.)

Things only got worse when I ran into damaged film, or ones that were spliced together with tape. Even just trying to make it work smooth with Super 8 was a pain. I needed a solution that was more universal.

Just to add, I don’t mean to throw any shade at the designer of the 8mm film grabber. It’s amazing they put such effort into it and shared it to begin with! It was still worth it for $50 worth of parts vs a $300 retail scanner.

Designing it myself

First obstacle: How to make the 8mm scanning faster?

It was excruciating waiting for a full revolution of a gear for each shot, so I needed a way to use only incremental amounts of the motor. Thankfully it’s a step motor that allows for rather fine grain control.

So I made a large gear and stepped down to it. It failed badly. There was no way to keep proper pressure across all the points into the film and have it feed smoothly. Would need a metal gear, not a 3D printed one.

However I knew this concept could work. I just needed to basically use only an 8th of a turn instead of a full turn.

Second obstacle: How to grab the film, even if it’s damaged?

So I needed something that could go around in partial circles, and also grab the film even if the place where you’re supposed to grab them is all torn up.

I thought of tennis ball launchers of two wheels that would press against each other and shoot something between them. But what would be soft enough for film to not destroy it, but catchy enough to grab? My wife had the answer, LEGO wheels.

Third obstacle: How will I align the frame for each picture?

That’s my secret, I don’t. I can have the wheel rotate “close enough” for the next frame to be in shot somewhere, as well as each frame on either side of it.

That way I can do some template matching in opencv in software later to figure out exactly where to crop!

I basically just take known good placement, aka the grabber holes on the left hand side, and make a black and white template off of them to search for in every image, then align them.

Forth obstacle: How to turn the receiving reel?

Technically this isn’t really needed, you’ll just have a pile of film spilling onto the floor you’ll have to cleanup after the scan. To avoid that, I decided to use the same motor as the film moving motor so that we never had to wait for two different motor’s movements.

My first attempt failed pitifully by trying to have a rubber band system. Instead, it was time to go back to gears….but I actually haven’t even done that yet. As I said, not actually needed.

How it’s going

Well, it’s almost 10 times faster now!

Throughout the process I have also upgraded the camera to the pi camera pro, and the light to a video light I could manually set the white balance on.

The Results

Well I don’t have any scans yet that I would want to share on here from the new setup (those clips so far all involve family travel or gatherings), so here is a much more fun clip from the first scanner! (Had to be digitally stabilized it was so bad with that scanner!)

Hard bumps in the Tank Corps

Hope you enjoyed!