2004/10/21 08:44:34.877 GMT+1000

Ludum Dare 5th 48 hour Game Development Competition
Post Mortem

Theme: "Random"
Entry name: "Blight Warrior"

Copyright 2004 Jetro Lauha
http://jet.ro

This post mortem is written a few days after the competition.

This was the third time I entered the 48 hour game dev compo. As before, I took it somewhat lightly, not taking too much stress about the whole thing. This time I also voted for the themes both first and second voting round. For the past two compos I didn't take any part in voting and didn't even check out the candidates for second round.

I feel that not contributing in voting made it easier to think up an idea for the final theme to work with. This time I felt it considerably harder to decide what idea I would concentrate on. I guess this is because I had already thought up more about ideas for a different theme from the second voting phase, which wasn't the final chosen one.

Well, after a few hours I finally settled on something I found interesting to try my hands on. It was an algorithm called "diffusion limited aggregation", heavily relying on randomness. In fact it is just bunch of dots moving through random path until they hit something. The algorithm can be used for generating certain kind of plant looking images. I read about this from the book "The Computational Beauty of Nature" (http://mitpress.mit.edu/books/FLAOH/cbnhtml/) and from a page about the algorithm in Paul Bourke's web site (http://astronomy.swin.edu.au/~pbourke/fractals/dla/]). As you can see, since I didn't have an idea which immediately felt good enough, I got a sufficient idea from spending a little bit of time for research.

Still I started by just implementing the DLA first and spending time optimizing it and adding graphical effects to it to make it look better. It took until about half of work for whole entry until I started concentrating more on game play. I borrowed the excellent Mersenne Twister pseudo random number generator algorithm from the net, which I think should be acceptable with the given theme. I could have used normal rand() as well and you wouldn't see any difference, but I thought it's better to make a bit of extra effort to find and use some better algorithm.

I didn't have practically any really bad bugs or problems hindering development. I just noticed a bit too late how much the game play would need tuning and trying out different things to make the game more enjoyable. I think at the end I got it much better than what it was at some point, but still I know there's room for improvement.

Rough estimate of total amount of actual development time is about 24 hours. Sleeping, eating, etc. are not counted in that.

What went right:

• + Familiar set of libraries (SDL etc).
• + Game design not needing a lot of assets.
• + Picked an idea which was interesting to try how the DLA could be applied to a game.

What went wrong:

• - Too much thinking before compo about other possible themes made it harder to think ideas for final theme.
• - Didn't spend enough time for game play - should have cut some time from graphical tuning for that.

Some random stats:

• 1 PC (3 GHz P4)
• ~24 hours of development
• about 77 KB of source code (3073 lines, 21 files, including the borrowed Mersenne Twister pseudo random number generator source)
• 85 KB of graphics (9 png files)
• 236 KB of sound and music (5 ogg & it files)