SHAZAM! Now there is a topic where anyone can post secret-lab-worthy things without having to make their own thread.
May the posting candles be lit and may the posting robes be worn! And may there be posting!
Now to launch immediately into some tripe. ^_^
Often, attempting to screenshot a game will give you a corrupted image.
You can use these images for neat purposes. This one uses 'Color Dodge' blending mode to make a cityscape-thing:
I dig it.
Now, this glitched data is not only useful for making fucked-up textures and tilesets for games. Its structured-yet-non-linear nature makes it ideal fodder for databending. So, let's turn these harsh, unforgiving pixels into "music"!
I'm using Audiopaint (an image-to-sound program), and the waveform used as the basis is this one I generated:
https://dl.dropboxusercontent.com/u/1566676/sfx30.wavThe tuning is harmonic and all are 180 seconds long. Nothing changes between iterations except the pic I feed the program.
--------------------------------------------------------------------------------
ITERATION #1 Here's a 320x240 section from a failed attempt to screenshot Heavy Bullets.
Audiopaint interprets the lines as oscillators. I tell it to use brightness and saturation for left/right channels respectively.
It sounds like this:
https://dl.dropboxusercontent.com/u/1566676/apex0.mp3 Yowza! Let's see if we can refine it any.
--------------------------------------------------------------------------------
ITERATION #2 I've set the original image's blending mode to 'Color Dodge' on top of a black background again, then messed with color balance. This one also looks a bit like a cityscape. Sounds like this:
https://dl.dropboxusercontent.com/u/1566676/apex1.mp3 Much more sparse and pleasing. Different textures shine through with it ALMOST sounding like music at points!
--------------------------------------------------------------------------------
ITERATION #3 I've done gone crazy with gradients'n'shit. The square forms have audible effects on properties of the sound themselves. Let's hear it:
https://dl.dropboxusercontent.com/u/1566676/apex2.mp3 Huzzah! Weirder, but somehow more structured. The gradient bands play into it as well.
There's thousands of ways to have it reinterpret a single image. The main two factors in anything I'd call "success" are having a good Scala .scl file and having a good original waveform to use as the basis. AudioPaint only does sines, so if you wanted squares or ramps or whatever, you'd have to sample a wave yourself. But it's worth it, if only for texture pieces to go in your more ambient songstuff.
--------------------------------------------------------------------------------
Now for some shape studies:
This's part an experiment to see how the sizing and spacing of basic shapes alone can be manipulated to make different audio/visual aesthetics. It's also useful for messing with image-to-audio stuff like the above. Even small differences like those spacings can totally alter the look of a tileset or the sound of some databent stuff.
Well, this has been amusing, I hope.
Feel free to post away! This is like dumbtopic, only SECRET.