Ericisonit 2023-09-01 19:54
Timo 2023-09-01 21:46
Timo 2023-09-02 07:21
If you used different shapes instead of ASCII for the characters, you could probably get quite accurate video. But it wouldn’t look that geeky anymore ;)
Ericisonit 2023-09-02 09:12
Yeah the thought definitely crossed my mind, but I was on an ascii art generation kick, but I am thinking about having two layers of text to help fill in the gaps.
Now that I think of it, it would be kinda neat to have it take every piece of every frame that is represented by the same character, compute some sort of average using the original image/video data captured, and see how much like that character the result actually appears. Like if I took every 8x8 cluster of pixels from the original video that it had decided looked most similar to an ‘M’, and averaged the colors of the pixels amongst them all, the result might look enough like the letter ‘M’ that it could be used for a new font, which could be used in this program to draw the text, and not only would it be a lot more accurate, but the font would probably look pretty crazy (and work for other videos as well)