How about this (I know nothing of Bot programming, so pardon me if what I'm saying is already the norm):
Hit Chances (for both human and bot playes):
1) Base accuracy 70% (+- depending on level of playing and target body location)
2) Modify % based on distance
3) Modify % based on lighting
4) Modify % based on target exposure (cover, kneeling etc.)
5) Modify % based on whether target is strafing/jumping etc
6) Modify % based on whether you are moving
7) Add % the longer you shoot continuously (resets if target or you move, or if you reload)
[Note: the above % applies in ADDITION to whether the reticule is within the target location - thus simulating hits that *look* like they're going to hit, but don't actually]
Notice the point 7. In UT, I notice that Bots with lousy accuracy tend to shoot with a wide arc, and seem to continue to shoot like that throughout (I could be wrong). This arc of fire ought to narrow with time. I mean the longer you shoot at someone, the better your aim is going to be - but games seem to miss this vital part of Bot AI. I mean, a firefight is far more interesting if the first few bullets are missing, and you are crossing your fingers to see who fires the bullet that hits - rather than what currently happens when you play CS: you can shoot wildly and still hit your target, but you have to hit constantly to frag him. I think an implementation of the point 7 sort would greatly enhance a real-life element I find lacking in almost every game I play.
What do you think? I guess it looks elementary when you look at it, but I don't see how such a simple system can be difficult to implement - and it would seem to eliminate "sniper pistols" of lousy bot AI games.
I'm trying to understand the basics of how Bot shooting AI is implemented. I don't understand how major titles such as Rainbow Six can botch what I see as something simple - or is it difficult to implement after all??
hmmm..
Don't frag me, love me
[This message was edited by ajy on Jun 09, 2000 at 13:59.]
Hit Chances (for both human and bot playes):
1) Base accuracy 70% (+- depending on level of playing and target body location)
2) Modify % based on distance
3) Modify % based on lighting
4) Modify % based on target exposure (cover, kneeling etc.)
5) Modify % based on whether target is strafing/jumping etc
6) Modify % based on whether you are moving
7) Add % the longer you shoot continuously (resets if target or you move, or if you reload)
[Note: the above % applies in ADDITION to whether the reticule is within the target location - thus simulating hits that *look* like they're going to hit, but don't actually]
Notice the point 7. In UT, I notice that Bots with lousy accuracy tend to shoot with a wide arc, and seem to continue to shoot like that throughout (I could be wrong). This arc of fire ought to narrow with time. I mean the longer you shoot at someone, the better your aim is going to be - but games seem to miss this vital part of Bot AI. I mean, a firefight is far more interesting if the first few bullets are missing, and you are crossing your fingers to see who fires the bullet that hits - rather than what currently happens when you play CS: you can shoot wildly and still hit your target, but you have to hit constantly to frag him. I think an implementation of the point 7 sort would greatly enhance a real-life element I find lacking in almost every game I play.
What do you think? I guess it looks elementary when you look at it, but I don't see how such a simple system can be difficult to implement - and it would seem to eliminate "sniper pistols" of lousy bot AI games.
I'm trying to understand the basics of how Bot shooting AI is implemented. I don't understand how major titles such as Rainbow Six can botch what I see as something simple - or is it difficult to implement after all??
hmmm..
Don't frag me, love me
[This message was edited by ajy on Jun 09, 2000 at 13:59.]
