Affective Color Theme Generation System for Visual-textual Design: A Case Study of Banner Design

Abstract

Color design is a crucial component in creating an appealing media presentation. Designers always prepare many color themes in their design work, while it is not an easy work for non-designers to obtain suitable colors. In this paper, we propose an approach for automatically creating affective color themes for visual-textual design. Banner design acts as an initial application. First, we create 2 estimation models, one is for affective color theme generation and the other is for legibility of text color on a background color. Then, a recommender system is developed to create banners with different colors corresponding to specified impressions. The mechanism of color unification with input images and text color legibility checker are implemented in the design system. Moreover, we conduct verification experiments to explore the effect of system output.