Re: 'not an artist' article

[Erik <br0ke@math.smsu.edu>]

On 01-Jul-2000 Joel Utting wrote:
> Steve Baker wrote:
>> 
>> The knack would be in picking the right controls for the 'genes' to act on
>> in the first place...and generalizing it to 3D with proper colour genes...
>> but I feel that this is something I could actually do sucessfully.
>> 
>> One cool part about doing this right would be that you could evolve in
>> the 'bones' and other widgets that you need for animation right as you
>> evolve the 'skin'.  You could probably also evolve low polygon count
>> versions for lesser levels of detail too.
> 
> A project I've been considering for a while is an RPG where the graphics
> are created based on a similar idea. The game "evolves" the land and
> several species over a number of years, also tracking the movements of
> intelligent colonies, etc, creating a history in the game, into which
> the player is plunged. The graphics were to be similarly evolved based
> on changing environmental/social factors, so each species would look
> different each time a new game is played.
> 
> Doesn't sound too hard, really...
> 
> Bye - Joel.

I, too, have had thoughts like this :) Except my design was primarily evolving
the cultural, economic, and capability aspects, not so much on the visual
imagery. I have a space sim on a back burner that I hope to have gameplay like
a cross between ('xwing' or 'wing commander') and (pirates!), so in addition to
a tight sim engine, there would have to be a large political and economic
playfield that is beleivable. I was actually hoping to write a paper on that
subject, but I never got the time :) Here I thought I was a supergenious having
this idea, and it turns out everyone and their brother is thinking along the
same lines.

One problem with evolving systems in software is that it uses a lot of
technique that are primarily in the field of AI, and most programmers don't
have a strong AI background. I'd imagine most people on this list have heard of
'genetic algorithms', but don't know their strengths, weaknesses, or the 4
steps involved. I do think the genetic algorithm is going to be about the
strongest candidate for this kind of work. Rule sets, 'expert systems', markov
chains, neural networks, etc will produce results that are either not
acceptable/beleivable or not different enough.

Another problem is knowing how to convert that information into a model. I have
training and a tiny bit of experience with AI in general and genetic algorithms
in specific, and I like to think I have a fair understanding of things like
economics, anthropology, cultural evolution, and some of the more general
biological issues, but I still consider myself fairly weak in 3d imagery. I
know most of the math (even have a fair handle on quaternions from reading math
books and harrassing the math professors at the local U), and I feel pretty
confident working with meshes and static models in opengl, but I've never dealt
with issues like inverse kinematics and animated models. The stuff on the web
about that seems to be highly windows concentric, as well.


When I was mulling over the concept of evolving a universe instead of designing
one, my conclusion was that it could produce rich and beleivable worlds if done
right. The time to program it would be fairly static, so for a very large game
world, it would be a much better solution than trying to design that by hand.
For a small world, however, it would take the same amount of time, so it would
*NOT* be a good choice. How big/complex of a world that change happens at
depends on how realistic and detailed you want the world to be.

        -Erik <erik@smluc.org> [http://math.smsu.edu/~br0ke]

The opinions expressed by me are not necessarily opinions. In all
probability, they are random rambling, and to be ignored. Failure to ignore
may result in severe boredom or confusion. Shake well before opening. Keep
Refrigerated.
        

---------------------------------------------------------------------
To unsubscribe, e-mail: linuxgames-unsubscribe@sunsite.auc.dk
For additional commands, e-mail: linuxgames-help@sunsite.auc.dk