main projects  side projects  about

Scribe Development Process

The Internet Archives, was interested in working with AIP Engineering to design & manufacture a new book scanning system.  They had some experience with an automated page-flipping book scanner, and wanted this new system to be manually operated to avoid the high per-unit costs ($120,000), as well as the high maintainance costs and downtime associated with a robotic system.

They also wanted the new system to shoot through glass, which is used to keep the pages flat.  Image quality suffers in systems that do not use glass to flatten the pages of the book, as the imaged pages have a visible waviness.  This can theoretically be fixed in software, but any software fix involves trade-offs, and any software engineer knows that a solution will not be 100%, especially when dealing with materials as varied as books.  With Scribe, the goal was to design a system that will take a near-perfect image without having to rely upon post-processing to finish the image, thereby avoiding quality-degrading post-processing steps. 

A week after reaching an agreement, the first working mock-up had been built from wood, glass, and steel:



This demonstrated the feasibility of shooting through glass, but highlighted that avoiding reflections with glass, while maintaining nice, bright lighting, was going to be a challenge on this project.

About 6 weeks later, a compete prototype had been made out of aluminum, this time with Canon EOS 1ds Mark II, and a wide-spectrum flourescent light:




About 5 weeks later, yet another completely new system had been made, which now used 6 multi-phosphor "daylight" cold cathode light tubes, as well as an adjustable light aperature for different sizes of books:




And then, another 6 weeks later, came the beta unit, now with overhead lighting using daylight-equivalent, specially coated halogen bulbs, as well as all new platen, cradle, lifter arms, and corner bracing:





We worked with the beta unit for about 8 weeks, redesigning the lights and many other subsystems while working with a local production machine shop, Swerve, to make parts for the first 10 units in the summer of 2005.  Below is the first scribe unit used in prodution, which includes many extra holes and un-anodized parts.  The client didn't care about looks, they wanted something that worked, and fast!




A number of relatively small changes have been made since that time, and pictures of the most recent Scribes are available in press images online.