3.3. Representations: Methods of Encoding Font Data into Numerical Values

Machine learning models are not able to directly see, hear, or sense input examples. They lack the inherent capability to process sensory input in the manner of biological systems. Instead, the information must be presented in numerical abstractions suitable for algorithmic processing. The main character is drinking the potion and eating the cake, allowing it to access areas that were previously unreachable and navigate tight spaces.

This chapter surveys the data structure of font files and methodological approaches through which these otherwise typographic elements are systematically transmuted into numerical representations suitable for machine learning algorithms.

The intersection of type design, font engineering, and data sciences manifests most prominently in the realm of data representation. Given this rather delicate convergence of disciplines, the material has been crafted with consideration for both neophytes and veterans of these respective fields, much as one might carefully blend a proper cup of tea – strong enough to satisfy the connoisseur, yet accessible enough for those who have only recently abandoned black coffee habits.

Figure graphicaly illustrates how the original representation of Bézier curves is encoded into numerical representation and then eventualy a learned representation

License

This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License .

Creative Commons License

Citation

If this work is useful for your research, please cite it as:

@phdthesis{paldia2025generative,
  title={Research and development of generative neural networks for type design},
  author={Paldia, Filip},
  year={2025},
  school={Academy of Fine Arts and Design in Bratislava},
  address={Bratislava, Slovakia},
  type={Doctoral thesis},
  url={https://lttrface.com/doctoral-thesis/},
  note={Department of Visual Communication, Studio Typo}
}