 |
Computer science undergrad student Joshua Bird has worked on numerous software and hardware projects, including building a 3D-printed “wigglegram” lens we covered last year. Bird’s latest project is a 3D-printed film video camera. It’s a cool project, although as Bird says, it’s ‘an expensive and unreliable way to record low-quality videos.’ However, that’s not enough to stop Bird, as he continues making DIY film products.
While film photography has enjoyed a renaissance and has carved out a small but significant niche, motion picture film doesn’t see much use outside of professional operations due to the extreme cost. ‘Motion picture film is in short supply and it’s almost impossible to find someone willing to develop and scan your videos,’ writes Bird. ‘All in all, you would probably need to wait months and pay ~$250 for just 10 minutes of footage – not exactly the best value proposition.’
Bird’s idea was to bypass motion picture film and its associated costs altogether by building a camera that uses standard film roll canisters, which are cheaper, more plentiful and easier to develop. However, a 36-shot roll only delivers 3 seconds of video at 12 frames per second.
To extract more footage from a 36-shot roll, Bird reduced the image size by a factor of 8, resulting in 288 photos per roll, or about 16 seconds of video when shot at 18fps. Bird found out while working on the project that buying and developing enough film for 10 minutes of video ended up costing around $600, more than twice the cost of the motion picture film. As he says, ‘oops.’
 |
| Joshua Bird’s initial sketches for his 3D-printed motion picture camera |
Economics aside, it’s a fascinating project. After trying four-bar linkages to pull the film down, Bird opted for a simpler tooth gear and cam design. While it worked, there were some reliability issues, likely due to tolerances with 3D printing being slightly too wide for the design. If Bird tackles the project again, he’ll reconsider the pulldown mechanism. The shutter was a special-ordered stainless steel design, which only took a week to arrive and cost around $10.
A motion picture viewfinder is quite complicated, and there are a couple of ways you could go about it. Bird wanted a through-the-lens (TTL) viewfinder, meaning that you can see precisely what’s being recorded. You can use a beam splitter prism, which splits half the light to the film and the other half to the viewfinder. This is the design used by Super 8 and consumer-oriented Super 16 cameras.
It’s a simple design, although at the cost of light. You can instead angle the shutter at 45 degrees. With this design, all the light is either sent to the viewfinder when the shutter blocks the film or sent to the film when the shutter is open. There’s no degradation with this option. If you’d like to learn more about how a motion picture camera works, you should check out this article: ‘Video: A 1000 fps look at how a 16mm motion picture camera works.’
 |
| Shutter design sketch |
Bird opted for the 45-degree design. It proved to be complicated. He needed to design and order 3D-printed nylon gears. Then, the mechanism barely fit in the camera. The shutter must also be completely flat and have a mirror finish. Finally, adjusting the shutter’s position to ensure a perfectly focused TTL viewfinder is challenging. In Bird’s case, it didn’t come together perfectly, so adjusting focus using the viewfinder was hard.
For the lens, cheapness was the goal, so Bird used the Canon EF-S 18-55mm kit lens. Unfortunately, ‘like most things with my camera, this ended up being a terrible idea,’ lamented Bird. The lens doesn’t have a focus scale, and since the viewfinder wasn’t as sharp as Bird wanted, manually focusing without a scale was hard. The lens also changes focus as you zoom, which is not ideal for video work. It also lacks a constant aperture, so the exposure changes as you zoom. If Bird were going to do it again, or if you want to try the project yourself, opt for a lens with a focus scale, parfocal design and constant aperture.
 |
| The EF-S 18-55mm lens that Bird selected lacks a physical focus scale, so he needed to make a rough scale using tape and markings on the lens to aid with focusing |
There are many more nuanced parts of the project, including rebuilding Canon’s lens protocol in C++, setting up a stepper motor and film jam sensor, using Negative Lab Pro (NLP) within Lightroom, stitching frames and using a machine learning script to perform image alignment. For the full details on the project, read Joshua Bird’s blog post and refer to the GitHub page. While perhaps not as affordable to use as Bird had hoped, the project is nonetheless open source.
Author:
This article comes from DP Review and can be read on the original site.
