I'm gathering resources on a feature that I dream of implementing in my real-time strategy simulation which is: A model of an ecology with vegetation which can support animal life, but I'm having a hard time finding the resources on this topic. These are the things I have in mind for the simulation: Plant growth based on temperature, sunshine and water and nutrients in the soil Cyclic flow of nutrients from the soil to plant life and back Once this is up and running, I'd very much like to add large herbivores and predators to the system. These will be modelled as AI agents. In the end, I hope that I can model agriculture as well. Of course I realize that any model I create would be at risk of becoming unstable, so I am prepared to use plenty of cheats and hacks to make it work, but my goal is still to start with a clean-looking and fairly plausible model. So, are there any resources out there that might help me in this quest? My library had this book which appears useful, but isn't directed at games of course.

Quote:Original post by wodinoneeye
You may have to compromise and do much of it as generalization because real ecosystems are very complicated and you probably dont want your simulation
to spend all its time doing interactions that wont be detectable by the player.

I'm going after emergent gameplay here, and I think there are plenty of rewards for the player with such a system. It isn't a first person game, but a strategy game. By modelling nutrients, for instance, I can have gameplay that facilitates burning grasslands in the spring to get a higher yield on them, or burning down forests resulting in fertile farmland. I can simulate the effect of overgrazing and the competition for land between farmers and cattle ranchers. Fertilizer becomes an important resource too.
Modelling large predators and prey gives rise to behaviour such as bears seeking into towns because their natural prey has been driven out or hunted. In many other ways the player's actions on the environment will have real consequences.

Quote:Original post by wodinoneeye
Will your map be large enough (in macro size and in small detail grain to fit the continuums and extents of realistic systems (ex- many real large predators travel 20 miles in a night hunting and have territories in that scale...)

I might tweak the consumption needs of the animals in order to get a more compressed, interesting world.

Quote:Original post by wodinoneeye
Start from the most general simulation first (actions of predatoprs on prey and the consumption of vegetation preplaced in appropriate terrain spots.
Statistical models/patterns for 'encounters' may be needed instead of running discrete objects to carry out the sim.

Yes, this was also my idea for modelling the lower rungs of the food chain. The populations of small mammals would be abstracted and perhaps only show up as a brief animation when they are being hunted by the larger, individually modelled predators.
And you're right, to get things going I can start off with static or simple vegetation models, then flesh them out later.

Quote:Original post by wodinoneeye
Add more detail if needed and the seasonal variations. A Cellular Automaton mechanism can be used to have general trends (groupings of object types) interact in an abstract fashion. The player then encounters actual objects that are a sample representation controlled by those higher level 'group' objects.

Yes, cellular automatons would be a great way of simulating the spread of vegetation and smaller creatures across the map.

There was a study done tracing a simple ecosystem, a pond i believe. The scientist initially though their were different tiers within the ecosystem where each member of the tier interacted with the others in very limited fashion, instead they were surprise to find that it was more like a net, where each member interacted freely with all other members in some fashion or form ( ie being eaten or eating ). This makes things much more complicated for ecology modeling.

Hope this helps! :D

-ddn