Real-Time ASCII art rendering using a single decision tree

art.pixlab.io

81 points·symisc_devel··15 comments

Open articleView on HN

Project homepage: https://github.com/symisc/ascii_art

I would have no idea what I was looking at if the input images weren't provided.

Defocusing/squinting helps to make the ASCII versions look more like the original image.

see also:

https://en.wikipedia.org/wiki/Libcaca

https://en.wikipedia.org/wiki/AAlib

One thing that annoys me with this algorithm is that anything that is saturated white becomes black.

I don't recall AALib doing that.

Really cool!

I've noticed that when pointing it towards an overexposed light source (like my phone screen or a lamp) it's "overflowing" and rendering pure white regions as black.

Are there any "Unicode art" edition?

Way more code points to render.

If it's only about shading, maybe. Although I found this style of ASCII art to be the least interesting. The better variants also do edge detection and consider the character shape. If you're going Unicode, I think one of the prettier options is Braille, although that's little more than a B&W subpixel rendering then.

> The better variants also do edge detection and consider the character shape

Sounds neat. Got one in mind?

JavE's (http://jave.de) converter was quite good with a bunch of options. But I haven't used any of those things in ... ages by now, so no idea what's the current state of the art.

Doesn't seem to function on FF64

Works for me.

I want this as a Instagram story filter. Does snap have this?

Way cool!

Nice

Reply on news.ycombinator.com