Large-format scanners come with an incredible cost and maintenance issues. Consumer flat-bed scanners won't cut it. A DIY solution is hoping to fill the gap.
Many photographers will tell you that the image resolution of even the latest digital cameras simply don't compare to film. This is especially true when you're talking about film in formats larger than 35mm.
One problem that has arisen, though, is how to properly archive or convert to a digital format. Consumer scanners aren't high enough resolution, and commercial scanners are both extremely expensive and often aging.
What you see above is one creative solution. A DSLR is mounted above a platform, which will hold and backlight a negative. The DSLR will take a picture of a small section, then the negative is moved and the process is repeated. Ultimately you have many pictures that are stitched together to get acceptably high-resolution results. Although this does yield good results, the process can be quite tedious. Someone has to sit there and move the tray, check alignment, and snap the picture manually. As noted in the forum thread, a 6x7 mm scan can take 25 shots to complete.
The solution was to motorize the tray and automate the system. The bed was reconstructed like a CNC mill with stepper motors handling the bed movement. An Arduino is harnessed to control the bed as well as trigger the camera. The bed on this unit could hold a variety of sizes, but you would have to change the construction of the frame if you wanted to scan anything too big.
He seems pleased with the quality, stating: "The scan shown in the video from a 4x5" negative would produce a print 22"x28" at 720dpi, the native input resolution of pro Epson printers."
Dynamic range is often just as big an issue, if not bigger, than is resolution. How much control do you have over camera operation? I could see this being used to take multiple exposures of each sub-image to allow for a nice HDR (High Dynamic Range) final image.