Perceptual Font Manifold from Generative Model

Though in recent times, various fonts are available online for public usage, it is difficult and challenging to generate, explore, and edit the fonts to meet the preferences of all users. To address these issues, we propose in this paper, a font manifold interface to visualize the perceptual adjustment in the latent space of a generative model of fonts. In this paper, we adopt the variational autoencoder network for font generation. Then, we conducted a perceptual study on the generated fonts from the multi-dimensional latent space of the generative model. After we obtained the distribution data of specific preferences, we utilized a manifold learning approach to visualize the font distribution. As a case study of our proposed method, we developed a user interface for generated font exploration in the designated user preference using a heat map representation.