ascii image thing

Basic image editor
​
I’ve implemented a vector drawing editor with smooth curves and pressure sensitivity, and I created a worker to process the updated canvas after every stroke to convert it into glyphs.
​
I intend to implement color support and layers later.
​
I want the glyph rendering to be real-time, so I have a lot of optimization work to do.
​
The video looks a bit choppy because I’m using a tablet laptop that has a very low detection range for its drawing pen. So the cursor teleports.

Refactor for speed
​
Initially, I attempted using kd-trees like I mentioned before. But that led to what felt like at 50x slower execution. So I just sat for a few minutes, staring at the code.
​
Then I realized the three separate color channels didn’t have to be linked together… so I refactored the logic to compute the channels separately, and now it’s running over 240x faster than the original.
​
For the Stardance logo (with the outline I added), it went from taking 290.9 seconds to only 1.2, with what looks like the same output.
​
I need to stop overthinking things.
​
I also realized I was calculating the width of a character without applying the correct font size. Things don’t warp anymore.

I was inspired by the typewriter’s inability to erase, and subtractive color mixing.
​
Now I have a program that generates three layers of stacked ASCII in cyan, magenta, and yellow, to replicate an image.
​
It generates very slowly at the moment–every character requires checking against 857,375 patterns to find the best fit. My plan is to use a kd-tree to reduce that to about 20 matches instead.
​
I also have to figure out why the images are squished vertically (or stretched horizontally).