IS IT POSSIBLE? I can't control units thinking they shoot enemy. Looks ugly from spectator point of view and also from probability theory because they can miss forever while i think position should be randomized not only around a target, but above the target itself. As a brute forced decision I can give 4x target facilities in the beginning. But they should be invisible and without hitbox.
contra: in addition to the annoyance that the AI knows about everything, even when not having seen it, it will be able to shoot at everything without missing it, even when not seeing it. it will make the AI only more annoying, plus it allows for tons of new cheats. .. no way!
I'm asking about a TOOL to control this. AI will have difficulty levels. Also i need this for new group behaviour. Finally, who's AI developer? :) Also I would like to understand the nature of this behaviour. I feel this happens not when i assign ATTACK command, but when unit is in idle mode and automatically searches for target.
edit: there's nothing stopping an AI from sending a command to a gadget to disable the 'floatyness' of radar blips and the units they represent for that team under uber hard AIs. I would say though that such a gadget disable it for all AIs as well as the AI that issued the command, in the name of fairness.
Exact position position is returned (you can watch this by entering "!e323ai mil", line is drawn directly to target). Attacking position instead of target ID is valid for static units only. Though can be used in some cases.
I'm not sure but i think shifted target pos it returned only when unit found a target by itself within its weapon range (result of Spring logic in maneuvering state), not under direct AI control.
Last edited by slogic on 19 May 2010, 21:39, edited 1 time in total.
Just tested. Rocko was exactly at weapon range towards the target. AI due to coded algorithm issued attack command on target. Rocko was shooting NEAR the target. Shooting & shooting, missing & missing. Looks really ugly. He was shooting until Spring algo decided to move it further a bit. Then again push it, until target appeared in his LOS and Rocko could hit target directly. This is abnormal behaviour. Cheating AI has no LOS! And those blips are NEVER pointing at REAL target position.
Looks like zwzsg is right, i need to shoot at geometry, not target ID. It is more accurate, LOL.
look ... the reason that most AIs use LOS cheat is, cause scouting is hard to code. That is at least the case for E323AI (Error said that), and it is the only acceptable reason. if this would also make units aim perfectly, even doh they have only radar coverage, it would be bad, cause it would make the AI just more annoying/appearing to cheat, with the advantage that "it looks not stupid from a spectators POV". IF you care so much for it to look good for you as spectator while you are debugging your AI, do what Beherith said: let the AI cheat give its self 4 targeting facilities and the soars it needs to run them.
Why are you binded with spectators? Leave them alone. It was minor point. AI with LOS cheating is coded in a way it shoots at its weapon range, not LOS range. AI coders needs a tool (API method to control this cheat). That is all. There is some switch for passive cheats in API. Could be one more method. I'm thinking you're trying to control AI development from your own point of view how AI should behave. Are you fear AI will be too overpowered? Then why does cheating interface exist? It is a point of allowing AI coder to fully control AI instead of dancing around by thinking how can i overcome this problem without unofficially modifying AI interface.
As we concerned cheats, my point of view is: GiveMeMetal(), SetMyHandicap(), GiveMeEnergy() and CreateUnit() functions are the most ugliest cheats AI can have. But they do exist! While we're arguing about adding control to much more moderate cheat.
Rocko was shooting NEAR the target. Shooting & shooting, missing & missing. Looks really ugly. He was shooting until Spring algo decided to move it further a bit. Then again push it, until target appeared in his LOS and Rocko could hit target directly. This is abnormal behaviour.
So, can i consider this behaviour as Spring bug? Each shot occurs on _static_ random position which is calculated once. But this position should be random each shot because blip is not accurate. Or may be my thoughts about random radar blips are out of Spring conventions?
> if each shot was at a random position, you would know where exactly the unit is after 10 shots
No until unit got los on him.
Re-read what? I already understood that AI devs won't get control on this behaviour. I just would like to understand if current "shooting at blips" behaviour is normal for Spring devs. For me it is ugly. It would be normal if unit was shooting at random position near the target (when target is out of los). Even if unit shot the target this is considered as occasional. Next shot can miss easily.
Users browsing this forum: No registered users and 0 guests
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot post attachments in this forum