Contents

Aoki and Aizawa (2022) | SVG vector font generation for Chinese Characters with Transformer

Aoki and Aizawa propose a novel architecture to generate SVG images, in particular Chinese characters. The architecture is based on Transformers and, unlike DeepVecFont, does not use a differentiable renderer. Instead, they generate SVG characters from a style reference and content reference in a style transfer manner.

Available resources at a glance

Aoki and Aizawa paper

Fig. 57 Screenshot of the Aoki and Aizawa (2022) paper

Data representation

Aoki and Aizawa adopt the embedding method from DeepSVG.