While playing with 3d mesh of a cube drawn with random triangles I noticed that some of them look similar to the text letters. I started to look for the whole alphabet, and here is what I finally got:
This might look nice but is not exactly the font. So I made another try. This time I made computer to recognize the letters by OCR (optical character recognition). I was drawing the mesh, this time just lines (no triangles), randomly rotated. This random shape was then recognized by OCR engine and this way I looked for every character in English alphabet, lower case and upper case.
I proudly present to you - the ugliest font ever! Computers think we should understand this. Not sure if it is the first truly computer generated font (without any reference to existing fonts) but I'm quite sure it is the ugliest. Some characters look ok, but some of them just have the same number of corners/lines or something... This is probably due to the reason that OCR is returning the letter every time, even if it doesn't look as any letter, just the closest one. And I couldn't find any way to make some sort of "letter quality threshold".
Changing cube rotation step from 45 degrees to 1 didn't make any difference.
I cannot call this experiment a success however i noticed a couple of interesting moments. The search always was done within 200-300 seconds. Capital letters were found much quicker than lower case ones. The "h" and "o" letters were found the last almost every time i ran the software. These were the same two letters i found last when i was looking for them manually.
The custom software is written in XCode using OpenFrameworks. The OCR engine is Tesseract-OCR, and is used thru OpenFrameworks plugin ofxTesseract.
Feedback is welcomed. 01.2016