Posts mit dem Label jbox2d werden angezeigt. Alle Posts anzeigen
Posts mit dem Label jbox2d werden angezeigt. Alle Posts anzeigen

Mittwoch, 17. März 2010

On my way to TweetMountain (Part 2)

Almost two months ago I wrote a first post about TweetMountain. Back then I only had a faint idea of what I wanted to do and what it could look like. Now - several weeks later - the idea still is the same but the look and a lot of technical details have changed. This post will summarize what has changed since then.


Before

The biggest step was for sure the change over from single-letter-rendering to the usage of whole words. This step changed the mountain from a sheer mess to a mess that can be read and interpreted as something that was created by humans ;-)
In addition to using whole words as building blocks I also introduced colors to the mountain. The color of the words now represents the account that has sent the according tweet.
Both ideas base on discussions with @prokrastinator whom I'd like to thank for his great input!

These content based changes were accompanied by one other change: Whenever an image gets posted in a tweet it will not be rendered as the URL that was shown in the tweet but as the actual image. This brings the visual content of a tweet into the mountain in its natural form.

After

So this is what TweetMountain currently looks like in full action (Unfortunately the people it currently follows do not post many pictures).

Next to the changes to the look and algorithms of TweetMountain, it also got its own twitter account: @zwitscherhuegel. It is used to follow different people and institutions that this way produce the mountain itself. But it is also used to output screen shots of the mountain and to get in touch for discussion about TweetMountain or other twitter-art-projects. So, the account is not (only) a bot, it also gets used manually. So if you have some questions or input, go ahead and tweet ;-)

So much for the current status of the project. Some further steps I may/will take include:
  • Support of more image services. Currently I only support Twitpic and yfrog, there are several other platforms that shold be used.
  • Find a place and a computer to host the Mountain 24/7.
  • Use a "dynamic camera" that zooms in or out depending on what is currently going on on TweetMountain.
  • Translate the twitter output the mountain generates to several languages so it gets more diverse.
  • Change letter rendering from a hard coded image map to dynamic font based rendering.
  • ...
Okay, that's it for today from TweetMountain. I will continue this project and keep you informed. If you want to, just follow @zwitscherhuegel and get in touch with TweetMountain or me on twitter.

Other posts that may be of interest:

Mittwoch, 27. Januar 2010

On my way to TweetMountain

I am using twitter now for quite a while. It has been a never ending source of inspiration and a great tool to implement and communicate some of my ideas. In the last few days I somehow realized that there are times in which the output of my twitter timeline just gets "somehow" overwhelming.

In the last view months I also started to dig a little into a programming language called processing (see processing.org). It is basically Java tweaked in a way that allows easy access to all graphic routines. Surfing thru the examples on OpenProcessing.org (a site built to share and exchange processing projects) I stumbled upon this cool sketch by antiplastik. I shows how simple physics and the use of OpenGL are in processing. This is a screen shot of the sketch in action:
Both inputs (the twitter overflow and the stacked boxes by antiplastik) fermented in my backbrain for a while till they came together in a small processing project I call TwitterMountain:
It uses little boxes that fall from the ceiling to create a pile of letters. The input is taken from my twitter timeline tweet by tweet, split into little boxes/letters and dropped onto the pile. Every now and the letters are lost and their boxes get deleted, but as twitter keeps on generating output the pile never vanishes.
My first implementation looked like this image above. The physics where there and I got the letters in the image. Twitter was connected and kept sending input.
Today I reworked it a little. First of all I added some limitations to box creation so the pile never fills my sketch completely. Secondly I created a nice sprite map that I can use as textures for the blocks. This way now the letters rotate correctly with their boxes.
So for now the TweetMountain is running and running, flooding my screen with an endless stream of tweets and letters.
There is some work left to do:
- clean code, write comments, use logging (gna gna gna... you know the routine)
- there seems to be a newer (and faster) version of jbox2d that I should check out (jbox2d is the library that does all the physics simulation in the background)
- build an Applet that can by used by everybody to watch his own twitter timeline

Maybe I even come up with some more ideas to put into the algorithm as well... I will keep you up to date with new versions.