Understanding Peanuts and Schulzian Symmetry: Panel Detection, Caption Detection, and Gag Panels in 17,897 Comic Strips Through Distant Viewing.

Abstract

In this article, we applied distant viewing to a corpus of 17,897 comic strips from Charles Schulz’s _Peanuts_ as a primary case study. Distant viewing uses computational techniques to study large-scale visual media, and draws upon interdisciplinary areas including visual media studies, cultural studies, data science, and semiotics. We focus on comic strips, particularly _Peanuts_, due to their widespread readership, historical and cultural cache, and complexity as a medium built on the interplay between text, image, and meaning. First, we discuss previous work done at the intersections of comics studies and computer vision. Next, we establish the processes for applying computer vision to comic strips. After that, we provide several examples, including: panel detection (variations in panel length over a cartoonist’s career); caption detection (identification and location of captions in panels); and comics paratext (computer vision analyses/exclusions of copyright text, signatures, dates, etc.). Combined studies of panel detection, caption detection, and comics paratext reveals new insights into the success, longevity, and influence of one of the world’s most famous newspaper comic strips. Ultimately, computer vision reveals a subtle stability and symmetry to Schulz’s artistry that played an understudied but significant role in the comic strip’s popularity.