ScanType: Evolutionary Letter Finding

João Cristóvão, Alexandra Oliveira, Tiago Martins, Sérgio M. Rebelo, Pedro Martins

Abstract: In this paper, we propose the evolutionary system ScanType for type design which enables the automatic creation of visual compositions of a set of input images to resemble letter forms. We developed a user interface to enable users to interactively experiment with this approach by defining input figures, as well as to guide the evolution of the visual compositions through the expression and specification of visual features of the outputs and evolutionary parameters. Input images can be scanned by users in real-time or loaded from a file. We conducted experiments with the proposed tool to evaluate its performance and capability by generating uppercase letter shapes with varied input images. Furthermore, we conducted evaluation sessions with graphic designers to assess the legibility, visual quality, and potential real-world applicability of the outputs obtained through this approach. The evaluation demonstrated the approach potential for co-creating visuals that are perceived as letter shapes.