What about the reaction to females compared to males. I wouldn't think of a woman with miniskirt as much of a treath, but of course I didn't know she was a badass ninjachic.

Quote:

Original post by Torfas What about the reaction to females compared to males. I wouldn't think of a woman with miniskirt as much of a treath, but of course I didn't know she was a badass ninjachic.

Do you have her number? ;)

Quote:

Original post by RDragon1 Do you have her number? ;)

Did you think I would give it to you?

Quote:

Original post by Kest To paint a picture, I'll use Half Life. Freeman walks into a Combine facility completely unarmed, and is quickly spotted by several soldiers. What should their reaction be? That sounded like it should end with a punchline [smile]

Is Gordon Freeman very suited for that example? I'm imagining him getting shot after soldiers shouting some questions and him being awfully silent for a while :)


Eh. Carry on. Miniskirt ninjas.

Quote:

Original post by Gnarf Quote: Original post by Kest To paint a picture, I'll use Half Life. Freeman walks into a Combine facility completely unarmed, and is quickly spotted by several soldiers. What should their reaction be? That sounded like it should end with a punchline [smile] Is Gordon Freeman very suited for that example? I'm imagining him getting shot after soldiers shouting some questions and him being awfully silent for a while :)

I can understand his reasoning. The guns make talking feel unnecessary. But in a world where Freeman is purposely unarmed, he can talk.

Quote:

Eh. Carry on. Miniskirt ninjas.

For the record, the player character is specifically male. However, three fourths of the game world's population is born female, due to radiation effects. There will be a few cyborg mutant miniskirt gangsta ninjas here and there.

Hawt. I like those odds.

One idea is to have a "Suspicion" factor that the AI has about the character. Certain actions raise the suspicion factor (like running, not following the orders given, etc). If the suspicion factor is high enough, the AI then tries to "control" the player (tell them to stop, raise their hands, move them towards the holding areas, etc).

There could also be a threat factor, and this would effect the reactions of the AI. If the threat level is high enough then the AI "feels" that their "life" is being threatened, so they will respond with deadly force (shoot first and sort it out later behaviour). If, however, the threat level is a little lower, then they might respond with a threat of force (draw weapons, etc). If it is lower still, they might only respond with verbal threats.

Most actions that raise the threat level will also raise the suspicion level too.

For instance, actions that might effect the suspicion level could be:
Clothing: If the player has appropriate clothing for the area (lab coats for a lab for instance), then this might lower the suspicion level. Alternatively, incorrect clothing would raise the suspicion level.

Eyeballing: If the player keeps looking at the AI character, this might raise the suspicion level. Likewise, if the player avoids looking at any AI especially if the AI starts to look at the player, it might also raise the suspicion level. So if a player occasionally looks at the AIs but not for too long, then this would not raise the suspicion level.

Loitering: If the player spends too long in one area (where they can be seen, even if they are allowed in that area), it might raise the suspicion level.

Avoidance: If the player character moves out of sight of an AI, this might raise the suspicion level. Or if the character moves away when approached by the AI.

Area: If the player is in an area that they are not supposed to be or that it is restricted, this would also raise the suspicion level.

Now, each factor might not be enough for an AI to necessarily respond, but with several factors in play (say avoidance in a restricted area, or eyeballing while loitering) might raise the suspicion level high enough for the AI to respond.

This initial response might just be to approach the player or take some other action that will create a test of suspicion (moving towards them, looking at them, starting a conversation, asking for ID, etc). The idea is that the AI, attempts to increase the suspicion level with their actions towards the player. Once the threat level is high enough the AI switches from actively attempting to raise suspicion levels to actively raising threat levels (however, threat levels can be raised by player actions without the AI activly attempting to raise them).

It turns the suspicion into a minigame. Actions that the AI takes attempts to increase the suspicion, whereas the actions the player takes will attempt to reduce the suspicion level.

The threat level would also create a minigame where the player can manipulate the AI's perception of them to control their reactions to them.

Going in unarmed will create a low suspicion level and a low threat level. This might be enough to get them past the guards without having to resort to combat. Or the player might accidentally take an action that causes the suspicion level to rise enough that the AIs see them as suspicious and attempts to take them in. If the threat level rises high enough, then the AI will respond in kind.

If for example, the player raised the suspicion level high enough for the AI to take them into custody, and as part of this they "search" (look in the inventory) of the player and find a weapon, then the threat level is raised high and the player might have to enter combat, if they didn't have a weapon and intend to attack using unarmed combat, then the AI's might not see them as a threat, but due to the suspicion still attempt to take them in.

As the AI's get in close, the player then attacks with an unarmed attack taking the AI's by surprise (the rate that the threat level rises might not be instant to reflect the reaction times of a real person).

If you can add in a slight randomisation. Different AI' agents might have different:
- Threshold level for specific reactions
- Increase/Decrease the amount that suspicion/threat rises or falls for a specific behaviour
- Preferred actions to use when trying to force suspicion/threat on the player
- Preferred responses as reactions to certain thresholds

This should give AI's quite complex seeming behaviours, and allows the player to interact with these behaviours in a non combat way.

Edtharan: Those are some excellent ideas. I'll definitely be saving your post for future reference.

Just one question. When you were thinking up and writing these behaviors, were you picturing the player having a visual clue for the AI's suspicion? And would that clue be a bar, body language, AI behavior, etc?

Whatever the feedback choice, one would need to choose between realism/immersion/emotion and clarity. Seeing exact values for suspicion would be the most useful to the player, but not having exact values might increase their apprehension, which is really fun to experience in these types of situations. But then, if its too iffy, the player may not be able to interact with it.

Quote:

Just one question. When you were thinking up and writing these behaviors, were you picturing the player having a visual clue for the AI's suspicion? And would that clue be a bar, body language, AI behavior, etc?

I didn't have any specific feedback system in mind as I was thinking more abstractly. But designing a feedback system shouldn't be too difficult (labour intensive it might be).

Generally I would expect it through animations for a more immersion game (like a 3D first person game). These would not just be avatar (the AI) body language, but also in their movements (although body language would be included).

The actual behaviour would not specifically reflect their suspicion level, but would be geared more towards reflecting the action they are doing to test the player for suspicion (or threat).

For instance, if they get slightly curious about the player and seek to test the Eyeballing criteria, they might stare at the play (essentially eyeball the player). If they decide to test the player's avoidance, they might move towards the player.

The change of behaviour indicates that the AI has moved from one suspicion threshold level to another. Other cues like the intensity/speed with which they do these actions (and other subtle body language or even verbal behaviours) would give further indication of the state.

As an example, with the avoidance testing, the AI might vary the speed with which they approach the player. So if they were not too suspicious, then they might casually walk towards the player (delay a while before they do, or even have pauses in the motion). If the suspicion level is high, then they might use a fast walking pace. If they have an extreme level of suspicion, they might run towards the player.

It is more work (much more), but it would add a lot of immersion to the game world.

Of course, if you are working with a more abstract implementation (say a top down tile based JRPG style game), then you might use a thought balloon above the NPC's head if the player mouse overs the NPC, or have a set of simple facial expressions to convey the state of suspicion of the NPC, or just use a bar graph over the head to indicate the suspicion level for high levels of abstraction.

Quote:

The most realistic answer is probably to yell at the player, telling them to drop on the ground and surrender. But that's an odd action to give the player - to lay down, and it would be complicated to give the player options from that position.

I would stick to actions the player could do, like dropping their weapon, standing still, following them, etc. If the players can lie prone, then by all means order them lie down as an action.

Of course, not all reactions that an AI would give would be at that level. they might just investigate (try to increase the suspicion level) first and not "blow their top" at the first sign of suspicious behaviour (unless the player is in a restricted area, in the wrong uniform, without ID and carrying a weapon :D ).

I suppose you could craft the threat responses of an AI to attempt to increase their threat perception of the player, but also to give the player orders that will eventually reduce the threat level that they present.

Essentially the AI would be trying to determine how high someone (the player) is willing to escalate a situation in as short a time as possible (the attempting to raise the threat level behaviour) while simultaneously trying to reduce the actual threat presented (the orders to drop weapons, lie down, etc).

Quote:

Original post by Edtharan It is more work (much more), but it would add a lot of immersion to the game world.

In this case, I think the result is worth it. Dishing out any resource assets will be pretty easy. The real challenge will be writing AI to handle all of the variations in player appearance, gear, clothing, and behavior. Figuring out how to mix these elements together with certain weights to generate a generic suspicion level. That sort of thing.

Quote:

Of course, if you are working with a more abstract implementation..

I have full body motion and facial expressions to play with, but limited vocals. Dialog is text based, but vocals are still used for action-based situations, like injury yelps, grunting to lift something, or being suprised.

Quote:

..unless the player is in a restricted area, in the wrong uniform, without ID and carrying a weapon :D

For the record, that was actually the original scenario for the thread. The player would already be under extreme suspicion, possibly even known as an enemy, but he's unarmed and doesn't look to be threatening. I was looking for ways to react to it.

Your suggestions may not be directly connected to that situation, but they're still very helpful. I haven't fully thought out the whole slinking around, trying to blend in mechanics, and this stuff is very useful to that end.