Things that could be improved

[Fidcal]

"avoid light" could be useful in another idea I have: stealth mode for AI. It sometimes happens that the player has an accomplice who in the story is also an outlaw or victim to be rescued etc. So if we had a stealth mode as well as an alert mode it could add gameplay.

 

Suppose you set an accomplice on team 0. He can interact with the player without alerting. Say he has to follow you. You keep to the shadows and he stays with you. You see some guards you crouch low in the shadows. Your friend runs off to attack the guards or runs away to a flee point depending if he is armed.

 

Alternatively, you set him neutral on team 5. This is OK unless the guards spot you and attack. You fight for your life while your neutral friend stands by and watches. He won't help.

 

So it would be nice to have an AI stealth mode where an AI avoids light, crouches if possible, does NOT attack guards unless they first attack him or his friends. He should then resume stealth mode.

[Springheel]

It just executes a smoothed-out random walk, with constant collision checking against its surroundings.

 

Collision detection essentially IS pathfinding. You can easily make AI move around if you don't care about them moving through geometry. If you want them to actually move around geometry, they need pathfinding information.

 

I'm not just talking out of my ass here; I'm the one who tried to get the flying behaviour working on the fire elementals, and I've spent a fair amount of time trying to make animations that involve AI hanging in the air. If there's some easy way to do this that I'm missing, I'd *love* to see what it is.

[pusianka]

Ehm this post is suppose to be a little digression*

I thought of it earlier but forgot to add it. Spiders seem to me to be ehm... they are not scary! Is it hard to make them:

1. Walk on ceilling, walls.

2. Jump

3. Use web like a rope to get down or get to some higher place or simply swing over the holes?

[grayman]

Collision detection essentially IS pathfinding. You can easily make AI move around if you don't care about them moving through geometry. If you want them to actually move around geometry, they need pathfinding information.

 

I'm not just talking out of my ass here; I'm the one who tried to get the flying behaviour working on the fire elementals, and I've spent a fair amount of time trying to make animations that involve AI hanging in the air. If there's some easy way to do this that I'm missing, I'd *love* to see what it is.

 

 

I think of pathfinding as getting from A to B, with precalculated help identifying the best ways to move. What I was referring to was setting a body in motion and letting it move freely while it checks its surroundings. It's not trying to figure out how to get to B. Nothing needs to be precalculated. The difficult part is making sure it doesn't start down a course that would end up in its being stuck. That involves rules that define what the object's allowed to do and not do so it looks as realistic as possible.

 

 

 

 

 

 

 

 

 

 

[Tels]

I think of pathfinding as getting from A to B, with precalculated help identifying the best ways to move. What I was referring to was setting a body in motion and letting it move freely while it checks its surroundings. It's not trying to figure out how to get to B. Nothing needs to be precalculated. The difficult part is making sure it doesn't start down a course that would end up in its being stuck. That involves rules that define what the object's allowed to do and not do so it looks as realistic as possible.

 

Yeah, if you think of a bird in the terms of "Doom3 AI", it needs pathfinding. But if you think of a bird in terms of a "rigid moveable body" it just needs collision detection, and some flight planning that adds forces/changes direction/speed so it won't collide with anything. And if it collides, it might just fold up and die.

[Mr Lemony Fresh]

What about... guard dogs lol... I'm serious, dues ex did it so why not.

 

Also, I think the variety in how you can defeat and enemy could do with improvement:

 

Jump on your enemy: depending whether you land on their heads, you should be able to break their necks. Their necks would surely buckle under the supposed 80kg/176lbs of a thief from a height. Of course you'd make a great noise doing it, and you'd need to be really precise when you jumped down.

 

Dropping objects on AI: Not quite sure if this has been implemented, but you should be able to drop things on an enemy AI which should either alert them, knock them out or kill them based on the weight of the object. IE if it's a pebble, they're alert. If it's an empty barrel (that you can lift) they should be knocked out, it should alert guards within a 30 metre radius, minus the noise reduction of various walls. If it's something that can only be dragged it's so heavy, then it should kill them it should alert guards within a 50 metre radius minus the walls again.

 

This adds strategy and intuition to the game so that players can have a sense of self achievement when they spot an opportunity to cleverly deviate from linearity. I'm sure It will not interfere with any past missions in a negative way.

[Mortem Desino]

Not a bad idea. This would have to be done in the code. Many, many, many things to consider: Angle of approach, What AF object (head, hand, arm, leg, torso, foot) the object hits, velocity of the object, mass/density of the object, impulse strength applied from the object, alert level of the AI, any possible bugs that might happen (maybe the large object clips into the AF figure of the AI, and rigidobject physics starts going wonky?)

 

Add a bug report, and call it a feature request for Dark Mod. Then start praying we find someone with time to burn.

[Mr Lemony Fresh]

Isn't there already a measurement of velocity for objects? when you drop a cup, it makes more sound if you drop it from a height than an inch off the ground. However the sound threshold is calculated from the impact on a surface, perhaps that same result could be timed by the value of weight of the object.

 

As for the angle and AF object stuff, i haven't a clue.

[Baddcog]

They should be alerted by anything that drops on them simply because of the noise. Doesn't matter if a cup hit ground or AI, all impacts have sound.

 

The KO by object is more complicated. Would probably need set mass, ie: anything over 25 kg would be enough to knock them out. And probably just set to the head body, but then you could probably throw a barrell into their face and ko them too.

 

Alternately I guess an extra body could be added to the top of the head so there's just a thin little square up top and most likely only a direct hit from the top would connect. Would be cool to drop a flower pot on a guards head and ko him (ala Kung Fu Hustle)

[Sotha]

The KO by object is more complicated. Would probably need set mass, ie: anything over 25 kg would be enough to knock them out. And probably just set to the head body, but then you could probably throw a barrell into their face and ko them too.

 

Depending how fast 25 kg's hits the victim, I'd say it will cause a lot of damage. Meaning no KO's, rather than kills. Try a hard 5kg weight object, a dumbbell, for example. Imagine that hitting someone on the head at decent velocity. Dead meat or skull fracture (resulting in death anyways).

 

Also keep in mind that thrown-object-KO's would probably change gameplay a lot...

[Tels]

Depending how fast 25 kg's hits the victim, I'd say it will cause a lot of damage. Meaning no KO's, rather than kills. Try a hard 5kg weight object, a dumbbell, for example. Imagine that hitting someone on the head at decent velocity. Dead meat or skull fracture (resulting in death anyways).).

 

Yeah, anything over 1kg will likely result in fracture and thus death unless modern medicine is around, even tho the death might come a day later...

 

Also keep in mind that thrown-object-KO's would probably change gameplay a lot...

 

I don't think so, throwing things make for rather noisy bounces from the helmet, floor and whereever it lands, plus we could easily add a "argl" sound to the AI for this special case. In addition, trying to hit a guard with a pot from far away isn't easy. Even when you are above him, because it is hard to judge whether you will hit.

 

The "get close and blackjack him" approach will always be more stealthy. So I think having a head-impact doing something special on the AI (either damage or KO) would be a nice addition, even tho it will get used seldom. But you could for instance then make traps for zombies (let a crate on a rope swing into them and that style of things).

[Baddcog]

Well, keep in mind the base value for TDM is 1 kg.

So we don't want things like forks and carrots to knock out AI, even small clay pots might break before they KO'ed you.

[Mr Lemony Fresh]

Also, if anyone is worried about picking up something and relying on it to knock out a guard, perhaps the thief should have the ability to guesstimate the weight, since he's holding it in his hands, or even better get a round number, so that a player can pick up and object, a little number appears down the bottom, and you can then think about the weight in relation to how high you are.

Edited October 8, 2010 by Mr Lemony Fresh

[lost_soul]

I've seen AIs get killed by the environment while flopping around as a ragdoll. The other day, I was playing The Alchemist. I knocked out the guy upstairs sleeping as he was standing up and his body started going crazy. It was flopping around like a fish out of water. Then it slammed into the treasure next to him and/or the wall, and I heard his death cry. I reloaded and was able to do this several times, laughing histarically the whole time.

 

So, apparently movable objects *can* kill??? or was it the wall that killed him? These are the things that would make great youtube vids.

[Fidcal]

An idea I just had: Run and walk speed reduced in proportion to loot carried. But the player can stash loot anywhere and pick it up later. When he stashes loot he clicks the 'use' button while loot is showing in the inventory and a loot bag goes into the carry position which he can then drop. The loot bag might carry a suspicious marker so the player can't just place it anywhere. The inventory loot is reset to zero.

 

So you could have several loot bags stashed around. Obviously you'd judge your exit route and put them where you can pick them up on your way out. Each one is created with the loot value at the time it was dropped so your inventory loot is incremented by the same amount when you pick it up. You might choose to take the slight loss of speed and not stash any especially in a small mission where the total loot is not high, or maybe just stash one.

 

Main problem would be FMs where you can no longer return to the place you left loot. This is a real killer because you can never know for sure. It's not just sequential 'progress along a road' FMs but some FMs have something special happen where you find yourself in a new area unexpectedly. That said, the FM author would have to allow you to get loot from that last place anyway, stashed or not.

 

Possibly this idea might be done in script so could be tried out in individual FMs. I'm not sure mysefl whether it's a good or not though. I was just pondering.

[Midnight]

Loot slowing down the player sounds like a mechanic that might work well as described in a multiplayer thief-vs-thief game mode, not sure about single player though. It would probably just get irritating after a while. Cool idea about leaving a loot stash around the place though.

[Mr Lemony Fresh]

Well, I think it's possible that you could design a level that makes such a mechanic work, and makes for interesting gameplay, (like if you had to throw your treasure into hoops or something fun) but then it may become necessary to explain why it is in this particular mission that you are suddenly weighed down by stuff.

But then there may come a need to invent that the thief has had a magic sack all along which temporarily sends his stash to another dimension. I guess at least this weird aptitude for carrying tons and tons of treasure without the clink-clank sounds could be explain, providing you can justify the extent of magical power that would have to lie behind such an oddity.

 

However, I can guarantee you that this is not ideal for everyone, so I doubt it. lol

[Fidcal]

Maybe it could be reserved for big missions with huge amounts of loot and this would make these seem more sensible. We accept the reasonably large amounts we accumulate so far but what if there 20,000 in loot? So suppose move speed is only reduced at 8000 plus and even then only slightly. So you could get to 10000 until it starts to be a minor nuisance. This would not then really affect existing FMs while giving some kind of credibility to hoovering up complete gold dinner services and so on repeatedly through a giant FM.

 

Probably more of a one-off script to be used in such or perhaps a multi-hub FM and not for general use.

[LEGION]

I like that idea. The 'leave-bag-spots' could be defined by mappers, so you can´t leave the bag everywhere you want?

Also it could add a risk in stealing more than the objective resulting in stealing not all possible, or judging "That weighs too much, I leave it", hmm.

[Fidcal]

I think leaving it in defined spots is rather artificial unless there is a special reason like the mapper knows that is the only spot you can get back to. So in that case he might make an objective: "There is a lot of loot in this area. You'd better leave it at x once you drop down to y or you won't be able to get back up to retrieve it." But generally I see no reason why just leave it up to the player to drop any number of bags or none at all wherever he wants bearing in mind that if a guard sees one it will cause an alert and he might take it.

[LEGION]

So then it would need a way for the mapper to define areas where the player cannot drop the bag in cases where the player can´t get to that place later in the mission along with kind of an explanation. Maybe there is already a Playervoice-comment fitting for that.

[Fidcal]

Well my above suggested objective is one way of informing players. Another way would be separate loot objectives: "Get at least 5000 loot from the pirate's cave before stowing away on their ship." Once on their ship a new objective comes up: "Add at least 3000 to your loot while on this trip."

[Baddcog]

I've seen AIs get killed by the environment while flopping around as a ragdoll. The other day, I was playing The Alchemist. I knocked out the guy upstairs sleeping as he was standing up and his body started going crazy. It was flopping around like a fish out of water. Then it slammed into the treasure next to him and/or the wall, and I heard his death cry. I reloaded and was able to do this several times, laughing histarically the whole time.

 

So, apparently movable objects *can* kill??? or was it the wall that killed him? These are the things that would make great youtube vids.

 

While it can be funny to see I don't like the idea of purposely uploading bug vids to utube. it makes it look like TDM is full of bugs when things like this are actually quite rare.

 

Who wants to play a game chaulk full of bugs?

 

Have you ever seen a dev put up videos that show bugs in their games?

 

it probably happens more because he is in a confused anim state between level of alert when you clock him. generaaly walls and moveble objects don't just kill ai.

[Springheel]

It's more likely a physics issue. When AI go into ragdoll state and are clipping into world geometry, all kinds of weird things can happen. I've shot a gas arrow at a group of AI and have seen AI go spinning off into the sky like an explosion went off.

[demagogue]

Practically every game has bug videos, even the best of them, so we're not alone on that.

 

I think what we should be doing is putting up more walkthrough videos of good FM gameplay in action. I don't know about other people, but I watch them to see how other people played the same FM I played.