algorithms

I’ve been working on a small interactive iPad experience for my daughter Ada. I wanted to create a world where objects would respond to touch, break apart, and evolve into different forms. She’s limited to a slappy/flailing motion and I wanted the game playable with only these rough moves. Gestures such as pinch-to-zoom or even panning the camera purposefully would not be appropriate. My v1 build therefore added a camera which zoomed to focus the most recently touched item. This worked well with my (more purposeful) testing, but zoomed around crazily once Ada got her hands on it.

Oh boy, SUSPIRIA. A prime example of the “cheesy horror movies so well crafted that they can be considered art films” genre. I can’t even remember what first added it to my queue, but I do remember seeing references to it in a crazy 2010 playthough of the SNES game Clock Tower which led me to the following reasoning:

1. Clock Tower is an insanely weird, horrific, and beautiful game.
2. Clock Tower cribs scenes directly from Suspiria.
3. Therefore, Suspiria must be even more weird, horrific, and beautiful than Clock Tower.

I’ve recently been working on an update to the twodee library we use for Ludum Dare games. One (of many) areas I’ll be focusing on is speeding up text rendering.

Text is currently very slow because we have to create and bind a new texture, render glyphs to it, then draw geometry for each piece of text in a scene. One simple optimization is to pack frequently-used text into a single texture which will remove many (expensive) texture binds.

Packing a bunch of rectangles into a texture isn’t the easiest thing to do well. There’s a whole class of algorithms dealing with this “bin packing” problem, each with various tradeoffs. Luckily, I found a very useful paper which covers many of these algorithms (thanks Jukka Jylänki!). To get a feel for how well each of them perform, I decided to implement a few in Javascript (you can see the source here).