Photofinish and a home-brew digital camera PDF
Written by Guilherme Maranhão   

Very likely you already seen a photofinish image, either of a horse track or from the Olympic Games. They all look alike, the background is usually of a single color and seems stretched with a bunch of athletes or horses as they cross the finish line of a race. There's a timeline either at the bottom or at the top of the image, numbers that represent elapsed seconds from the beginning of the race. These images were generated by a camera equipped with a slit instead of a regular shutter and 35mm film moving behind this slit. Nowadays the linear CCD substituted the slit and made this cameras very similar to most scanners we have (or maybe scanners are similar to those cameras).


Late 19th century the Ermanox camera introduced the focal plane shutter, at that point the shutter was basically a moving slit. Later there's a report about the British Museum using a fixed slit with glass plates to record the entire surface of ceramic pots in one picture. Kodak made the slit static and moved flexible film for the first time in Circut panoramic cameras. In the 1930's, Lorenzo del Riccio, italian immigrant living in California, invented the photofinish camera. At that time he was working for the Paramount Pictures photo lab, probably with access to lots of 35mm film and equipment (this part is very badly documented) but those who hired him to develop this idea apparetly were the Hollywood horse track. He tried many approaches with regular cameras and flash exposures but without sucess. With a flash you document who wins the race, but unless you fire a camera for each horse passing the finish line, you cannot get the entire result of a race on images.


Lorenzo opted for moving the film in the opposite direction of the running horses, behind a slit. He placed this contraption in front of the finish line on a high spot. From there all the camera saw was the finish line itself. This is the reason why on the Olympic Games although the race tracks are terracota colored the photofinish images have white background, because all the camera sees is the finish line.

Lorenzo would turn the camera on just before the winner finished the race and turn it off after the last one had crossed the line. Finding out the speed the film has to move to record a good image was the trick that made Lorenzo famous.


Slit photography isn't a kind of high speed photography. Instead of registering a range in space for a fraction of time, the slit records a range of time in a fraction of space. All the results of a single race could fit into one image this way. And even though the resulting image has a rectangular shape, the area that the camera sees is just a line.


Most photographers never did get close to this kind of technology because it was so complicated and demanded so much film to make things work. This knowlodge ended up kept inside the photofinish photography industry. Digital photography brought some of it closer to us in the form of the flatbed scanner which is basically a slit camera inside a box.


Professor Andrew Davidhazy from RIT in Rochester, published in the web many articles showing how to experiment with the linear CCD from hand held scanners by making a digital slit camera. The whole idea is very simple, all you have to do is free your scanner from its box and maybe make a few modifications to it if you must. If you're curious enough, I'm pretty sure you can do it, but having the guts to take it apart and risk loosing it helps.




Through the years I came up with what I think is the best combination for the things I do with such a machine, I use a HP 2200C scanner, with Vuescan software (sort of a universal driver for scanners) running on Mac OS plataform. When I started I had a Pentium 150Mhz, running Windows 95 and a scanner I couldn't tell which brand it was. Both setups gave great results, the trick is to find a combination of things that leads to a working machine. Most parallel or SCSI scanners have a manufacturer's driver that is able to check if the lamp is working, for example, and that spoils the setup (you end up with a message on screen telling you to visit the nearest service center).


This is the main reason why I picked Vuescan, it doesn't care if the scanner is in one piece or not. My experience with Sane on Linux is that is tries to calibrate for every scan and that is annoying. Slim scanners are of no use to this project, they can however be used to create a kind of scanning back for a 8x10" camera, but that's another story for the future. I usually find scanners at the place where they sell scrap computer parts, a broken glass or a burnt lamp usually end the life of a scanner, since we don't need them one can use such scanners for this.




At this point I should do a little disclaimer: this may cause extensive damage to your scanner and your computer, do this at your own risk. I'm no expert in eletronics, this is based in my readings and experience. If you decide to go ahead, good luck and keep your fingers away from the circuit board to which the scanner lamp is attached, voltage there goes very high and causes a strong burned skin smell when it touches you! There's also a lot of glass inside a scanner, be careful with that.


A typical scanner is a box containing three circuit boards (main, sensor and lamp), a little motor, a lamp and some other hardware. Basically you want to keep two of the boards (main and sensor) and the cable linking them and get rid of the rest. To test your setup it would nice to get a working scanner and scan a full sized image with it first. Unplug it from the wall, disassemble it and unplug the lamp from its board, check to see it still works. Repeat for the entire lamp board, then the motor. If you get an error message it's time to either look for other hardware or other software. If you didn't get an error message, you're almost done (dispose of the lamp safely). You may have problems with the U-shaped sensor. Some scanners have it, some don't or some software don't care about it. This sensor tells if the moving mechanism of the scanner is in its starting position.


There's usually a plastic tab from the scanner box that entries the U-shaped sensor. So remove the two remaining boards from the box along the moving part to which the sensor board is attached and test the scanner again. If you get a message such as "waiting for scanner", it seems the software is waiting for the scanner to reset. Place the tip of a screwdriver in the U-shaped sensor to see if that makes the scanner work. Test also if you can leave it there for the entire process. If you get a negative, look for other hardware.


By now you probably noticed there's a lot of black plastic right in front of your lens, and some mirrors too. Remove the mirrors carefully and get your saw out, let's get rid of that plastic. Once you are done, you will notice that your lens is focused too near, with a gentle tap of a hammer you can probably sink it a milimeter or two in its hole sending the focus back a couple of centimeters. Nice. You can build a whole new body for your sensor, with a lens that focuses and has an iris, but that's not a requirement. Just tape the main circuit board to the side of be camera and you're done. Plug it to the computer and hit scan.




Keep in your head the principle of the photofinish photography, things that remain static render lines at the background of the photo, things that move across the view of the camera get to appear in the final image. Go out, experiment, go as far as you can get from an eletrical outlet!


Guilherme Maranhão
This e-mail address is being protected from spambots. You need JavaScript enabled to view it
November 2007



Share This