Drawing lessons on complex writing are too sensitive

[Wilker-uwu]

the more complex the letters are, the shorter the strokes often are. but because the strokes are shorter, the precision required for these does not make much sense. a few examples includes

• the last stroke of 先 can be drawn like 失;
• セ can be drawn with a straight stroke like in 七;
• 食 allows for the second-to-last stroke to be a long straight vertical line as if you were drawing 月;
• kanjis involving 言 can be completed by straight-up tapping around the screen;

one solution could be to have the tolerance be dynamic in proportion to the length of the strokes, but i don't have experience in dealing with drawings in programming to tell yet :(

[blastrock]

It's quite hard to make a good balance there. It's very frustrating when you draw correctly but the app rejects your answer, so I preferred making the algorithm very tolerant.

I don't expect kakugo to ever differentiate between 先 and 失, they are just too close. Maybe we can do something about short strokes to avoid that last case about 言, but it seems complex.

[Wilker-uwu]

in a side note (aka i forgot to tell), recently, i had tested with another device, which had a significantly larger screen resolution, and what i noticed is that both the stroke lines are a lot thinner, and the drawings are just a bit less sensitive. could that imply that the stroke widths can have an impact on this?

btw, that gives me the idea for a feature of selecting stroke widths for better accessibility. i'll open an issue on that :3