Open hugolpz opened 6 years ago
There is a problem when a stroke overlaps on itself, such as for manuscript "a", "b", "e", "f", "g", "h", "k", "l" ... because the "median" colorizes also other parts of the stroke at the cross point.
This is never the case when one draws a hanzi (no stroke overlaps on itself). However, there is the same kind of problem with some Japanese hiragana such as the 3rd stroke of "あ" . I didn't find a simple solution until now to solve this issue (at the moment, I cut the stroke in 2 and use some special medians to render the two parts of the stroke as a single stroke, but this is really tricky).
Oh I see. Due to the svg masks used, true.
Note : My professional partners @Inalco asked me if data was available for other alphabet. They are a French language university so they will not bother helping, but as I expected they seemed interested to use webapps if existing for arabic and co.
As for my conflict of interest disclosure, I work for education (paid employee), and I push for opensource and open data (as volunteer, wikipedia editor).
Are there Hangul glyph done?
It may sounds silly...
But your technology could be used on simplier alphabets such as latin, greek, arabic, thai, etc. These alphabets may themselves have variants: lower case, upper cases, cursive.
Down stream usages such as Hanzi-writer could be helpful for these languages as well. (cc @Chanind )
As for doing it, most of them would require modest workload, to define further (find a font, vectorise, break in stroke, define the path), but I wanted raise the idea.