NY Times Word Cloud Portrait?

[Guest]

Looking for a little insight into this app. If you don't know what it is check it out - here's an image it generates:

http://admin.inc.com/2013/03/18/how-to-be-best-in-show-at-sxsw/cloud4975/

I have a customer looking to do the same thing. I could easily fake it - by taking an image with a bunch of words laid out, and then using a hi-con of the face to mask out the words. But that will cut off letters and such... and that's not ideal. So looking for insight into a nesting / fitting algorithm to do this. Anyone know who developed this app for NY Times?

[moccamaximum]

This "classic" example could be somehow bended to your needs.

http://help.adobe.com/en_US/as3/dev/WS5b3ccc516d4fbf351e63e3d118a9b90204-7eee.html

I would look into how this algorithm converts shades of grey to ascii-signs.

[Guest]

Doing ASCII art is easy, and nothing like what I am looking to do. Thanks anyway though.

[dmennenoh]

Wow... full year ago I was initially researching this. Originally this was intended for a client, but it never got traction so I stopped working on it. Have been revisiting it though, as it's an interesting problem and have made good progress. Still not 100% where I want it to be, but here's a sample image from just today.

[dmennenoh]

So have been doing more work on this to make it work as an actual tag cloud portrait generator. Before, as in the image above, I simply found rectangles and then picked words that would fit the space. To make it right I had to find rects that matched the word size. Right now I'm using a service that provides Twitter hashtags along with usage counts and I use the counts to figure out font sizes. Anyway, made this test image just now - only problem I'm having now is that it's taking about 20sec to produce. I'd like to improve that for sure.