Not right at the moment, but I could probably put a few together without too much difficulty. I'll try to find a spare moment to do that at some point.
eta: Done; here you go.
It's a bit blocky, because it's a low-res font designed for a game played in 320×240 VGA. (And it's almost embarrassingly '90s-gamer-bling in its aesthetic, but then I originally dreamed up the font design in the '90s for a game.) You can see the multiple styles of character in a few places, notably the N and the H. The line at the bottom with the CHEEESE illustrates the global effect: add another E at the end of the string, and the connections between Es flip round and end up with the H going the other way.
Looking at this demo text, actually, I'm not so impressed with the optimisation; QUICK looks particularly nasty to my eye. Partly that's because the Q just isn't very good (I don't think I've had occasion to use it before now!) but also I think the I would have been better extended downwards rather than upwards; I should have had my value function reward a good balance of upward and downward extensions. Still, that's just a small tweak in the optimisation goal and doesn't invalidate the basic approach.
Cool! Oooh, I thought of a use for it! On a modern computer, you could have it in real time, dynamically morphing the characters, having constantly slowly shifting but constantly well designed text...
![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
Do you have screen caps of the output, out of interest?
eta: Done; here you go.
It's a bit blocky, because it's a low-res font designed for a game played in 320×240 VGA. (And it's almost embarrassingly '90s-gamer-bling in its aesthetic, but then I originally dreamed up the font design in the '90s for a game.) You can see the multiple styles of character in a few places, notably the N and the H. The line at the bottom with the CHEEESE illustrates the global effect: add another E at the end of the string, and the connections between Es flip round and end up with the H going the other way.
Looking at this demo text, actually, I'm not so impressed with the optimisation; QUICK looks particularly nasty to my eye. Partly that's because the Q just isn't very good (I don't think I've had occasion to use it before now!) but also I think the I would have been better extended downwards rather than upwards; I should have had my value function reward a good balance of upward and downward extensions. Still, that's just a small tweak in the optimisation goal and doesn't invalidate the basic approach.