Impressions2Font: Generating Fonts by Specifying Impressions

TL;DR

This paper introduces Imp2Font, a GAN-based model that generates font images aligned with specified impressions, effectively translating impression words into visual font styles with high quality.

Contribution

It extends conditional GANs to accept multiple impression words, enabling flexible and impression-specific font generation using a novel impression embedding module.

Findings

Imp2Font outperforms comparative methods in quality of generated fonts.

It successfully generates fonts from unlearned impression words.

The model supports multiple impression words simultaneously.

Abstract

Various fonts give us various impressions, which are often represented by words. This paper proposes Impressions2Font (Imp2Font) that generates font images with specific impressions. Imp2Font is an extended version of conditional generative adversarial networks (GANs). More precisely, Imp2Font accepts an arbitrary number of impression words as the condition to generate the font images. These impression words are converted into a soft-constraint vector by an impression embedding module built on a word embedding technique. Qualitative and quantitative evaluations prove that Imp2Font generates font images with higher quality than comparative methods by providing multiple impression words or even unlearned words.