Vertical Webcam with a Raspberry Pi, an HQ Camera and a Surplus Photographic Enlarger

I regularly meet with other members of the Surrey Linux User Group on Jitsi. We have been doing this rather than meeting in person every month, during the pandemic of 2020 (and beyond?).

In addition to this, of course, I often fiddle about with electronic gadgetry, and cannot always identify pins and other stuff without sighted assistance.

So, I decided I needed a way to connect up a Webcam which would look downwards onto the desk, and in which other Jitsi users in the meetings would be able to see what was under the camera.

With a lot of photography now switched from film to digital, photographic enlargers are appearing on Ebay for next to nothing, compared to what they would have cost to buy new.

A photographic enlarger, the one I bought, consists of a flat baseboard, upon which there is a vertical pole mounted at the back. An adjustable assembly can be moved up and down this pole by means of a rotating knob.

On the assembly that moves up and down, there was originally a very strong light, a 45 degree mirror directing the light down past where a photographic slide would be fitted, and a focusable object lens with an aperture adjustment and an optional red filter.

The enlarger I bought from Ebay is a Durst B30. It cost me less than twenty pounds, about 27 dollars or so. It would probably have cost the best part of £800 or so brand new.

The baseboard on this enlarger is about two feet, 60 centimetres square. The pole up and down which the assembly moves is about three feet in height, about 90 centimetres.

I removed the light, mirror and focusable lens assembly.

I made an 8mm plywood platform to replace this, upon the top of which I mounted a Raspberry Pi 3B+, by drilling holes in the plywood and using M2.5 hardware and stand-offs.

Under the plywood, I fixed a right-angle bracket which comes from an Ikea bookcase. The bracket is designed to be screwed into the wall above the bookcase and screwed to the top, to help to prevent the bookcase from toppling over.

The plywood platform mounts easily to two 4BA screw holes left by removing the original assembly, and is held horizontally.

The plywood is about eight inches from left to right, about twenty-one centimetres, and about two-and-half inches front to back, about six centimetres.

From Amazon I bought some knurled standard tripod screws which fit through the vertical portion of the angle bracket.

The 30cm ribbon cable from the camera snakes round the front of the plywood platform and plugs into the CSI port on the Pi.

The 'High Quality' Pi camera fits onto one of these tripod screws through the vertical face of the bracket. The vertical and horizontal faces of this bracket are slotted, allowing some adjustment of both the degree to which the bracket projects from the front of the plywood platform, and the height at which the knurled screw mounts the camera.

I bought a couple of different 'Arducam' lenses for the HQ camera. These cost about twenty pounds a piece on Amazon.

I also fitted a 'cold white' LED lamp with a GU4 fitting on to the plywood platform, facing down to illuminate the object below. The lamp is powered by eight AA batteries in a battery holder. I intend to replace the batteries with a more long-lasting solution. This lamp is fitted through a hole drilled in the plywood to one side of where the Raspberry Pi is mounted. The hole is the right size to make the pins end of the plastic light cone squeeze snugly into it, and the pins are exposed on the top surface of the plywood.

This is a 12 Watt lamp, so it draws an Amp from the batteries. Which would drain AA batteries quite quickly, so the temporary solution is less than ideal.

I bought an HDMI to USB converter from Amazon. This, when plugged into any Windows or Linux computer, appears on that computer as a Webcam device.

I plugged the HDMI end into the Raspberry Pi on the enlarger.

So, I now have a vertically down-facing Webcam, with plenty of illumination.

Further to this, using Kurzweil 1000 OCR software on Windows, I can very rapidly do OCR on documents, and use this method to test focusing.

I can also do this without the HDMI to USB converter, by:

  1. Running raspivid on the Raspberry Pi and making it stream to a TCP listener.

  2. Using the v4l2loopback kernel module on an Ubuntu desktop to produce a 'bogus' video device.

  3. Reading from the TCP streamed video from the Pi, on the Ubuntu machine, and dumping the stream into the v4l2loopback device, using ffmpeg.

Either method allows me to log in to the Jitsi meetings, on the Windows PC as myself, and on the Ubuntu machine as 'GizmoCam'.

Using this method sighted folks on the other side of the world could look at stuff and tell me what it is, what pins are where (as long as they can see PCB screen-printing), or identify knobs and buttons on little boxes by reading the legend over my camera.

You could also do this stuff without the HQ camera, using one of the little and much cheaper lower quality cameras. And by using something else to hold the Pi and the camera facing down.

But my assembly removes the problems of vibration when using a much more Heath Robinson (Rube Goldberg) construction, and using the HQ camera provides better quality video and the Arducam lenses provide easy focus rings and, in some cases, aperture adjustment.

This thing has probably cost about £100 overall.

I think that is very good value for what I got from it.

Not only the end result, but the hours of figuring it out, the hours of careful construction, and the help and kudos from my buddies from the Surrey Linux User Group on Jitsi

The whole assembly is very, very strong and stability is excellent.

This is the first time I have ever touched a photographic enlarger of any kind for any purpose.

I was so taken with the quality and ingenuity of the construction of its parts that I have been careful not to modify any part of the thing. I have only removed and stored away the optical parts and replaced them with the Pi, camera and light on my plywood platform.

I could, if desired, re-assemble the original enlarger and re-sell it.

Thanks are due to some folks from the Surrey Linux User Group, in particular John W and Tony W. John in particular was amused watching my hands manipulate things under the cam, and was a great help while I was going through several iterations of the platform and camera assembly to see what worked best and whether options made focus easy, or even possible.

Some of the cost could have been saved by not going through unsuccessful iterations. For instance originally using fibre-glass matrix board for the platform instead of plywood, and from buying M3 and M2.5 hardware on Amazon. But in doing so I now have plenty of all of these things in the junk box.