RTS-Starter kit info?
by Alex Huck · in Torque Game Engine · 05/20/2005 (1:56 pm) · 2 replies
Hi, all.
I've been planning on getting the RTS-Starter kit, but I had a few questions about it.
1) How many bots can it support? Could I have 4 teams, each with a maximum population of 200?
2) How intelligent are the AI, how many tasks could you manage to program without much lag? For example,
"Build Building", "Gather resouces" to name a few. It would be quite a strain on a computer to have around 200
bots each -intelligently- doing tasks.
3) (Not related to AI) How many buildings could be on the screen at once, since it depends on the amount of polygons in the building shapes, lets say for each building having 300 polys.
4) Are there any game demos made with the kit, to get an idea what it's capable of?
-Alex Huck
I've been planning on getting the RTS-Starter kit, but I had a few questions about it.
1) How many bots can it support? Could I have 4 teams, each with a maximum population of 200?
2) How intelligent are the AI, how many tasks could you manage to program without much lag? For example,
"Build Building", "Gather resouces" to name a few. It would be quite a strain on a computer to have around 200
bots each -intelligently- doing tasks.
3) (Not related to AI) How many buildings could be on the screen at once, since it depends on the amount of polygons in the building shapes, lets say for each building having 300 polys.
4) Are there any game demos made with the kit, to get an idea what it's capable of?
-Alex Huck
About the author
#2
a) target platform(s).
b) Model complexity, LoD usage design, intended field of view restrictions (if any), etc.
c) various optimizations that you intend to use (and of course, implement), as well as expected bot behaviour.
There was an experiment run by one of the RTS-SK owners (Jeff Leigh) which graphed performance vs unit count using some of the example models that are included in the RTS-SK, which gave a very good breakdown of the capabilities. I myself have been able to have 150 units on screen at once using models >750 polys, with no LoD implementation and some very basic auto-attack AI calls all fighting at once, at roughly 30 fps on a top end gaming machine--and that was with zero optimizations other than the RTS-SK itself, across the internet. All tests have indicated that video performance is your limiting factor, not TGE.
2) AI is completely up to you. The RTS-SK does not provide AI implementations because AI is an extremely game-specific design and implementation. There is a script only resource that shows examples about how to build buildings, gather resources, and many other basic semi-genre agnostic RTS tasks, but these are example implementations--how your game works will drive how you implement the functionality.
3) See 1) above.
4) There are not currently any demos (get cracking RTS-SK owners!), but there are several projects in various stages in the process of development. See Andy's post above!
05/20/2005 (2:18 pm)
1) Just like with TGE, this question is extremely difficult to answer, because the kit itself isn't your limiting factor. You need to define your:a) target platform(s).
b) Model complexity, LoD usage design, intended field of view restrictions (if any), etc.
c) various optimizations that you intend to use (and of course, implement), as well as expected bot behaviour.
There was an experiment run by one of the RTS-SK owners (Jeff Leigh) which graphed performance vs unit count using some of the example models that are included in the RTS-SK, which gave a very good breakdown of the capabilities. I myself have been able to have 150 units on screen at once using models >750 polys, with no LoD implementation and some very basic auto-attack AI calls all fighting at once, at roughly 30 fps on a top end gaming machine--and that was with zero optimizations other than the RTS-SK itself, across the internet. All tests have indicated that video performance is your limiting factor, not TGE.
2) AI is completely up to you. The RTS-SK does not provide AI implementations because AI is an extremely game-specific design and implementation. There is a script only resource that shows examples about how to build buildings, gather resources, and many other basic semi-genre agnostic RTS tasks, but these are example implementations--how your game works will drive how you implement the functionality.
3) See 1) above.
4) There are not currently any demos (get cracking RTS-SK owners!), but there are several projects in various stages in the process of development. See Andy's post above!
Torque 3D Owner Andy Schatz
Yes, you can have teams, and there is no strict limit on number of units ather than performance.
The AI are not intelligent at all. That's up to you. The book AI Programming Wisdom, or the Game Gems one on AI are good resources for figuring out RTS pathfinding.
"Tasks" as you put them, are generally not performace-sucks at all. Pathfinding is almost the only thing you will have to worry about performance-wise with the AI. There are ways to amortize this pathfinding as well.
Buildings are the same as units, add their poly count in to your number.
I don't believe anyone has shipped a game using the RTS pack yet, but Wildlife TYcoon: Venture Africa is using it as a foundation with great success. I'm afraid I can't release a demo yet. But the kit itself comes with a basic demo. Just download the demo of the kit and you'll get an idea.