I'm surprised at how good it is to detect the characters. The algorithm is like a 100 bytes long. This is madness.
Top-level
I'm surprised at how good it is to detect the characters. The algorithm is like a 100 bytes long. This is madness. 11 comments
@tendigits yeah maybe :D I'll finish implementing some characters, and then the goal would be to make it so I can enter the Uxntal alphabet. @neauoire that's cool. those directional artefact glyphs are pretty interesting. how does line length factor in? by the looks of that translation only direction counts? @glitchpop yeah the size of the glyph doesn't matter it's only encoding the changes in direction. See this: https://git.sr.ht/~rabbits/grail/tree/main/item/src/grail.assets.tal#L11 @neauoire oh wow, doing it with directions like this is something I’ve never considered, how interesting @neauoire this is very neat! it makes me wonder how different people's writing styles could be catalogued based on their stroke direction / order for different letters ... and i wonder if a similar algorithm could be used in reverse to create generative handwriting-style text @neauoire The ‘B’ in the first line and the ‘13’ in the second are identical. It is only through context that we are able to read them correctly. |
@neauoire daaaang that’s impressive. Could open up some trackpad-only device options like Uxn on old PDAs maybe?