Contemplating: what contains less style information, the written word "word" or the spoken word "word"?
-
Show this thread
-
Replying to @quasimondo
In other words: how many dimensions of the latent space of visual information are clearly discernible yet don't contain core semantic information, compared to the same for the auditory modality? (My guess is that the visual domain has much more room for stylistic variance.)
1 reply 1 retweet 6 likes -
Replying to @Plinz @quasimondo
you can mathematically define an epicircle for any glyph/text, 2d fourier analysis, with all possible variations, and compare precision and length in bits of usefull closed curves, that are distinguished easily and associative. use step functions separate 2 closed curves.
1 reply 0 retweets 0 likes -
Replying to @ollj @quasimondo
I can see how that could work as an estimate, but I don't yet understand how close this particular representation is to the Kolmogoroff complexity of the minimal automata required to do the same thing. Any intutions?
1 reply 0 retweets 2 likes -
Replying to @Plinz @quasimondo
circles have best derivatives/curvature/continuity. so you epicircle all the shapes.
1 reply 0 retweets 0 likes
For glyphs, I would prefer Beziers. But we want not optimal glyph matching, style variance depends on the artist's representation model and rendering method. Why not use a meta function that searches for a compression model over possible support vectors and projection functions?
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.