Taking faster product photos with artificial intelligence

Repetitive process

Product photos are actually taken continuously in the same way, so this is a repetitive task. Product placement, taking the photo, post-processing. Can you automate steps in that? Initially, Diego had the idea that post-processing could be automated, but soon came to the conclusion that the time gained was in fact in minimising post-processing by a smarter photography process.

More photos, less work

Normally you take one product photo, which you will edit. What Diego does in his process is take 3 photos. The product photo, a silhouette photo and a highlights/shadow photo. That silhouette is taken with a Fresnel lens, and used to immediately create the crop. AI helps with this when it comes to larger objects that are not quite in focus. Then the highlight/shadow image is used to overlay over the photo to finish it. How exactly this works, Diego explains in the video.

Result

With this technique, each photo was taken in 5 seconds, instead of 15 minutes. To perfect this technique, Studio Litstill applied for an HBO start-up grant, together with Fontys and Gerard Schouten's AI & Big Data lectorate and SIA. With that budget, the studio plans to develop a photography robot, which will also further automate the photography itself.

This project was carried out by Diego at his own studio Litstill, as part of his studies in ICT & Media Design. The grant came about in collaboration with Gerard Schouten, Lecturer in AI & Big Data and leading lecturer of the Applied AI For Society Knowledge Centre.