Empowering Backbone Models for Visual Text Generation with Input Granularity Control and Glyph-Aware Training

Open in new window