Real-time high quality font generation with Conditional Font GAN

作者：

Highlights：

• We propose a real-time novel font synthesis network based on GAN.

• Without additional fine-tuning steps, the network generates controllable font sets.

• The AdaIN operation is used to fuse style across all layers of the generator.

• Architecture is intended to be more computationally efficient than SOTA algorithms.

摘要

•We propose a real-time novel font synthesis network based on GAN.•Without additional fine-tuning steps, the network generates controllable font sets.•The AdaIN operation is used to fuse style across all layers of the generator.•Architecture is intended to be more computationally efficient than SOTA algorithms.

论文关键词：Font generation,Generative adversarial networks,Style transfer,Weakly-supervised learning,High resolution font image synthesis

论文评审过程：Received 27 December 2021, Revised 13 September 2022, Accepted 23 September 2022, Available online 29 September 2022, Version of Record 9 October 2022.

论文官网地址：https://doi.org/10.1016/j.eswa.2022.118907