Photo credit: Benjamin Bezine via Peta Pixel
Designer Benjamin Bezine wanted to create an automated film scanner, but in an unconventional way. So, he decided to use a Raspberry Pi mini computer, LEGO pieces, and a mirrorless camera that uses machine learning to scan entire rolls of film. The version you’re about to see is the fourth-generation Roboscan model, with the biggest upgrade being the Raspberry Pi and using machine learning to detect when a photo is properly framed before activating the camera to fire.
Benzine didn’t just stop at the device itself, as he even developed a web interface for controlling the scanning process, complete with full control of the camera settings that can be saved and restored at the end of each session. This version can scan a full 36-frame roll of film in approximately 6-minutes, including transferring all images to a PC. More information here.
The first version was a simple proof of concept, but was too sensitive to variations in the film, such as rigidity of the film, imprecision in the photos spacing. There were also many manual operations such as camera settings you always forget to change. So, I made a new, more robust version that tries to make this faster and fool-proof,” said Bezine.