Tweet
I've been attempting to digitize 8mm films for years, starting from projecting the film and using camcorder to record it (like everyone else back then
). Then progress it along according to my experience. The ultimate goal is of course finding ways to capture the film, each individual frames at a time, at ample resolution - and with as simplest & cheapest possible solution.
I had eventually achieved it, a dead simple but reliable scanning setup.
As you can see it's pretty straightforward. A modified projector, a digital camera, some wirings, and that's it! No electronics, no arduino, no plc. Nothing complicated whatsoever.
The film transport is based on a Sankyo Dualux 1000 bought from a flea market for less than 10$. I believe this is the basic version made for domestic Japanese market (100v only, ac line voltage projection lamp). However that doesn't matter as most of its internals would have to be gutted out anyway.
Mechanically intact, but most of its electronics had been removed. Only the main control switch survives.
The main drive is now a simple DC 12v geared motor. It is rated output speed is 450 rpm, which would be reduced to around 120-150 rpm (or 2-2.5 fps) at the shutter drive - a perfect fit to my need.
And now the tricky part - the trigger system, to tell the camera when to fire the shutter and take the picture.
Since I'd need nothing more than a contact closure to release the camera's shutter so I thought of the simplest possible solution. And it really is - a reed switch activated by a magnet glued to the main drive pulley. Ridiculously simple, but WORKS so reliably - it NEVER missed a shot!
The rest is just a matter of "timing" the trigger so that it would activate only when the film is stationary in the gate, not while being pulled down.
On the extreme left of the pic is where the shuttle cam and the shutter blade would be. I've found that the shutter blade can be entirely removed without sawing/filing off anything, sweet.
The light source is nothing but an ordinary 6500k 12v LED found in any hardware store. If the original projector's transformer is still there you can directly run the led from it. But since this projector uses line voltage projection lamp I'd have to use external power supply in this case.
A diffuser is added between the led & the film gate. Just a sheet of white plastic ripped out from a dead ceiling lamp.
Here's about the optical path. Still using my tried & trust formula - a 16mm projector lens (50mm f1.6, the most common ones) used as a "magnifier" to enlarge the image large enough to be easily "seen" by the digital camera. With long enough telephoto lens the camera would be able to look directly into the enlarged image of the film via this 16mm projector lens, and fill the image sensor.
By the way the Sankyo's original lens barrel is just a bit too small to have this lens inserted. I ended up replacing it with a 3D-printed part instead. Not perfect, but it gets the job done.
The camera used is just an old mirrorless ones - Panasonic DMC G-7 (bought second hand for about 280$ a few years back). But any camera with full manual control, electronic shutter function, and wired shutter remote socket should work OK as well.
The camera works Full Manual mode / native ISO setting / white balanced to the clear film base in the gate, and with digital teleconverter function to fill out the sensor. Capturing at 3 Megapixel, 3:2 aspect ratio seems more than enough resolution to capture at least 95% of details on the film, and to finish the work at HD 1080p resolution.
(4K & beyond is wayyyyy overkill for 8mm film, I insist.)
Some wiring mess behind the scene. On the right is the variable voltage PSU driving the main drive motor. Another fixed 12v ones on the power rail will supply power to the LED. The camera is powered by that small wall adaptor at the far right. The wired remote shutter sits in the middle, connecting between the projector and the camera.
The only gotcha of this setup is that it's quite slow - around 2 frames per second. Tried speeding up beyond that but it started to skip frames - still not sure whether it's camera itself running out of buffer, or the SD card used can't keep up the writing with the camera, or both. Therefor I'd set everything up and let it run on its own while I'm doing something else, then come back later to stop the process when it's done.
So how's the result? Does it worth the wait?
This is the actual what "raw" capture (negative stock in this case, Kodak Vision 3 500T I think) from the camera looked like - the entire frame (plus some borders) at 2,272*1,520 resolution. And it's pretty sharp down to all corners.
You'll get some 3,600 images for every 50 ft of Super 8 reel or around 4,000 for Double 8 films. For easier handling it would recommend to do no more than 150-200 ft at a time unless your PC is powerful enough. Usually I'd use Sony Vegas Pro for subsequent image flip/crop, color correction, editing, and finally export it as .mp4 file at 1080p resolution, bout YMMV.
The very same film frame as the above, after vertical flipping & cropping, color inversion & correction.
Does it really work?
I've been using this setup for a little over a couple of years, and get enough regular customers to permanently have this as part of my service(s). According to the logbook I've been scanning about 62 of 50' cartridge, or 3,100' (or 230,000 frames) of Super8 this year. So yes it's practical, and works well enough to my need.
For people who's more on programming/3D printing parts this is definitely NOT for you. But for people with slight mechanical skill / electronically idiot & knows nothing about programming (like me), you may find this approach interesting.
). Then progress it along according to my experience. The ultimate goal is of course finding ways to capture the film, each individual frames at a time, at ample resolution - and with as simplest & cheapest possible solution.I had eventually achieved it, a dead simple but reliable scanning setup.
As you can see it's pretty straightforward. A modified projector, a digital camera, some wirings, and that's it! No electronics, no arduino, no plc. Nothing complicated whatsoever.
The film transport is based on a Sankyo Dualux 1000 bought from a flea market for less than 10$. I believe this is the basic version made for domestic Japanese market (100v only, ac line voltage projection lamp). However that doesn't matter as most of its internals would have to be gutted out anyway.
Mechanically intact, but most of its electronics had been removed. Only the main control switch survives.
The main drive is now a simple DC 12v geared motor. It is rated output speed is 450 rpm, which would be reduced to around 120-150 rpm (or 2-2.5 fps) at the shutter drive - a perfect fit to my need.
And now the tricky part - the trigger system, to tell the camera when to fire the shutter and take the picture.
Since I'd need nothing more than a contact closure to release the camera's shutter so I thought of the simplest possible solution. And it really is - a reed switch activated by a magnet glued to the main drive pulley. Ridiculously simple, but WORKS so reliably - it NEVER missed a shot!
The rest is just a matter of "timing" the trigger so that it would activate only when the film is stationary in the gate, not while being pulled down.On the extreme left of the pic is where the shuttle cam and the shutter blade would be. I've found that the shutter blade can be entirely removed without sawing/filing off anything, sweet.
The light source is nothing but an ordinary 6500k 12v LED found in any hardware store. If the original projector's transformer is still there you can directly run the led from it. But since this projector uses line voltage projection lamp I'd have to use external power supply in this case.
A diffuser is added between the led & the film gate. Just a sheet of white plastic ripped out from a dead ceiling lamp.
Here's about the optical path. Still using my tried & trust formula - a 16mm projector lens (50mm f1.6, the most common ones) used as a "magnifier" to enlarge the image large enough to be easily "seen" by the digital camera. With long enough telephoto lens the camera would be able to look directly into the enlarged image of the film via this 16mm projector lens, and fill the image sensor.
By the way the Sankyo's original lens barrel is just a bit too small to have this lens inserted. I ended up replacing it with a 3D-printed part instead. Not perfect, but it gets the job done.
The camera used is just an old mirrorless ones - Panasonic DMC G-7 (bought second hand for about 280$ a few years back). But any camera with full manual control, electronic shutter function, and wired shutter remote socket should work OK as well.
The camera works Full Manual mode / native ISO setting / white balanced to the clear film base in the gate, and with digital teleconverter function to fill out the sensor. Capturing at 3 Megapixel, 3:2 aspect ratio seems more than enough resolution to capture at least 95% of details on the film, and to finish the work at HD 1080p resolution.
(4K & beyond is wayyyyy overkill for 8mm film, I insist.)
Some wiring mess behind the scene.
On the right is the variable voltage PSU driving the main drive motor. Another fixed 12v ones on the power rail will supply power to the LED. The camera is powered by that small wall adaptor at the far right. The wired remote shutter sits in the middle, connecting between the projector and the camera.The only gotcha of this setup is that it's quite slow - around 2 frames per second. Tried speeding up beyond that but it started to skip frames - still not sure whether it's camera itself running out of buffer, or the SD card used can't keep up the writing with the camera, or both. Therefor I'd set everything up and let it run on its own while I'm doing something else, then come back later to stop the process when it's done.
So how's the result? Does it worth the wait?
This is the actual what "raw" capture (negative stock in this case, Kodak Vision 3 500T I think) from the camera looked like - the entire frame (plus some borders) at 2,272*1,520 resolution. And it's pretty sharp down to all corners.
You'll get some 3,600 images for every 50 ft of Super 8 reel or around 4,000 for Double 8 films. For easier handling it would recommend to do no more than 150-200 ft at a time unless your PC is powerful enough. Usually I'd use Sony Vegas Pro for subsequent image flip/crop, color correction, editing, and finally export it as .mp4 file at 1080p resolution, bout YMMV.
The very same film frame as the above, after vertical flipping & cropping, color inversion & correction.
Does it really work?
I've been using this setup for a little over a couple of years, and get enough regular customers to permanently have this as part of my service(s). According to the logbook I've been scanning about 62 of 50' cartridge, or 3,100' (or 230,000 frames
) of Super8 this year. So yes it's practical, and works well enough to my need.For people who's more on programming/3D printing parts this is definitely NOT for you. But for people with slight mechanical skill / electronically idiot & knows nothing about programming (like me
), you may find this approach interesting.
Attached Files
). That way I can just swap out the card when finished scanning a reel and continue the work instead of transferring all images via a (slow) data cable as of now.
Comment