Telecine

This is a placeholder page including notes on the Telecine I've been working on for several years now. It will eventually be a fully-detailed page, but for now it contains thoughts and facts I need to remember.

Problems with Other Designs

The low-end, under $10,000 market for telecine machines seems entirely occupied by either old machines designed for analog television, or hobbyists. Every so often, one comes across Reddit or design blogs for being interesting, but these are the problems I've seen with them:

• Strain on sprocket holes
  • many use only one gear, which moves the film one frame, stops, then moves it again. The other ends of the film are attached to the reels, which have mass and inertia of their own -- so each frame move is either pulling, or being pulled by, the reels. This has the potential to tear sprocket holes, or leave a lot of slack in the mechanism causing poor feeding. Also, many I've seen only have a couple sprockets engaging with the film at any time, which further adds strain on the film by not spreading out the resistance across several sprocket holes.

• Mechanical shutters
  • it may sound nice to use a 22MP DSLR to take the frame images, but there's a bunch of issues here:
    • Mechanical shutters take time, if you're only getting 1 or 2 frames a second then you're taking 10-20x as long to image your film as the film is long.
    • Mechanical shutters wear out if they are fired once a second for three hours. At work, I know this for a fact; we went through three Canon DSLRs on a planetary camera project -- the shutters don't hold up being fired constantly. Ten minutes of film = 600 seconds x 24 frames = 14,400 frames. That's a lot. Maybe you can get through the first batch of scanning, but if you have an actual collection of films, that's a drop in the bucket.
    • As an aside, 2MB per image is almost 50MB per second. That's a pretty big camera card to hold 10min of those images, also figuring in saving time to write to the camera card.
    • a 22MP image is much larger than even a 4K frame -- there may be advantage to crop without loss, but I don't see the advantage.

• lack of feedback
  • particularly Arduino ones, they don't seem to validate data -- move a frame, take a picture, but it doesn't check that either were successful.
  • Those not using shutters seem to be filming digital video and hoping the frames sync up, without the mechanical advantages of an original-style telecine that was designed to sync frames. This is how I've converted films in the past at analog resolution, it take a bit of finesse. I want to capture the data and massage frame rates, exposure, and framing after the fact rather than getting it right while the filming is going on.

Telecine Mark I

My first thought was to make something mechanical, so I learned 3D design and designed a machine full of gears, including a Geneva Gear for frame-by-frame movement.

Features realized in this version:

• keeping as many sprockets in contact with the film as possible;
• Separate sprockets for constant feeding from the reels, so that there is near-zero tension on the film at the frame-by-frame motion point.

Telecine Mark II

The mechanical system seemed to be:

• impractical from a 3D printing standpoint;
• have poor performance at high speeds with the geneva gear;
• difficulty syncing with camera.

I purchased some stepper motors, a 5MP digital microscope, and a 3-axis CNC microcontroller, with the intention of just printing the sprockets and programming the microcontroller for movement, and writing code to capture from the camera after the 'move' command finished.

Features realized in this version:

• calling the digital camera directly from the computer
• moving to all electronic system.

I didn't actually assemble anything; due to moving and a variety of other issues the parts sat in a box waiting for some free time.

Telecine Mark III

When my wife learned what 3D printers do, she decided we need to buy one. So, that gave me incentive to pick up this project.

Mark II's problems:

• The 3-axis CNC controller isn't really designed for the application I'm using; it would take a lot of tweaking to program it to do what I want.
• The digital microscope wasn't the greatest; OK for microscoping, but it had a hard vignette and circular distortion at the edges.

However, I acquired a second-hand Canon MS400, more to be used as a viewer but it has a SCSI card in it.

The lens on the MS400 is exactly what I need -- I haven't played with it, but it should be much more effective than the tiny lens in the digital microscope. So, I purchased an 8MP "tinkerer" digital camera -- essentially a small circuit board with a color CMOS sensor on it -- which I will mount using the MS400 lens. The advantage of the MS400 lens is it both zooms and focuses, so I can tweak for the best image possible.

I have ordered an Arduino, because I can now see why everyone else uses them -- they can be programmed to do a lot of things I was trying to 'trick' the microcontroller, the serial port, the parallel port, etc., to do, and the Arduino can do it better. I originally leaned away from it because it seemed to be more of a "set it and forget it", but on further review it can also provide the level of feedback I was hoping for.