Rosettatype’s Twitter Archive—№ 2,884

@shancarter @michael_nielsen @typostammtisch I would not object to changing the whole paradigm of a font from a collection of glyphs to a “shape generator” (e.g. some neural network?).

@shancarter @michael_nielsen @typostammtisch The biggest challenges are, imho: use of vectors (i.e. precision and independence of scale) and finding the right interfaces for designers/users to interact with NNs in a meaningful ways. (and probably something else I cannot foresee atm.)

@shancarter @michael_nielsen @typostammtisch Although, I usually leave oracular activities to others, my guess is that we will see ML-augmented type design within the next 5–7 years (kerning, boldening, widening, …). In 10–12 y we will see assistive design tools that will generate new characters with/for you.