Research on GAN-based Text Effects Style Transfer
Yinquan Liu, Zhuang Chen · 2021
With the development of neural style transfer and generative adversarial network, the research of text effect style transfer has appeared. The text effect style transfer aims to render text images with style images to produce text effects images. However, for more complex text, the existing methods will generate unrecognizable font images. Therefore, we propose to add morphological methods to the glyph transformation to limit the degree of glyph transformation, and add distance transformation loss when training the texture network to limit the texture transfer, so as to improve the overall transformation effect. Experiments show that, compared with other existing technologies, our proposed method is more suitable for stylizing complex glyph images.