메뉴 건너뛰기

A study on Synthesis of High-Quality Korean Fonts based on Deep Learning considering the Structural Characteristics of Hangeul 

 

Goal:

  • Having only few human designed characters in a specific style we would like an AI based machine that generates the rest of the font characters in that particular font style
  • “More Specifically”
    • Given 114 font characters in a specific Korean hangul font style, Deep Learning model generates the rest of the 11k characters in that particular style

 

 

 


 

Problem Statement:

  • Font designer takes weeks to months to create a single font style for alphabets which are only 26 (a, b, c, d, …, z) ☹
  • Imagine the time and effort required for CJK (Chinese-Japanese-Korean) which are 10k-50k? ☹
  • Font design requires detail font knowledge

"How about using an AI to solve this font design problem?