Patrol Routes in FMs

[Springheel]

I barely have time to play missions, but I've already played a few of my favourites twice. I actually prefer playing a really great mission twice then a below average one once. And I did notice the guards behaved differently in Mandrosola the second time around.

 

It's certainly possible to go overboard, but I don't think random patrols are that much extra work, if you're willing to invest in the extra beta-testing.

[Sotha]

umm yeah that's basically what I said, lulz

 

Where? Here? See below:

 

If it's a long patrol it's almost a waste to set up random stuff because..

 

Then your claim is incorrect as I'm doing exactly the opposite. Long distance patrols are totally messed up by the random deviation matrix. Short range local patrols are static with dynamic twists, like I described earlier. Like I said, the idea is to have the static patrols make it certain that the critical areas are manned. The random patrol's idea is to increase tension as there are few always safe areas.

 

Any area could be potentially dangerous if the player hangs out there too long. A matrix-walking guard could approach from any direction. If the player has immensely bad luck, he could be sandwiched by three incoming matrix-walking guards, each approaching from different directions. Or then the player might get lucky and all the matrix-guys are on the opposite side of the map and he only has to deal with the local guards. Or then the average tension situation where he occasionally has to dodge one matrix-guard.

 

 

 

I actually prefer playing a really great mission twice then a below average one once. And I did notice the guards behaved differently in Mandrosola the second time around.

 

Wow.. That was a nice implication to make! Thanks and much appreciated!

 

 

At any rate, a progress report on the matrix thingy:

 

I've implemented the long-distance-AI-patrol-mess-up-matrix in my WIP and I have to say it works quite nicely. The matrix walking AI walk very unpredictably now, but do not make any bad looking kinks.

 

Setting this up involved no scripting or anything, just placing one path_corner per spot where the map spreads off in different directions. Then when the AI reach that point, he randomly chooses one of the directions in the crossroads. Then he chooses again in the next crossroad, and again and again until he reaches the starting point.

 

I've set up two matrixes which direct the torch wielding AI's to patrol the whole map street areas in a random fashion. They have a rare chance to deviate from the big streets and inspect dark alleys here and there. Setting this up did not consume much more time than setting up approx 4 ordinary AI static patrol routes (per matrix). Timewise this means a few hours, which I find acceptable and normal.

 

extra beta-testing

 

This is probably The Biggest Issue with the matrix approach. We will have to see. My expectations would be:

If the matrix-walking guards do not have torches, their impact is not that large.

 

But torch wielding matrix-walking guards certainly require me to place some crates behind which the player can duck&cover. This will certainly gonna be a mission difficulty level dependant feature: easy will have 2-4 torchless matrix-guards. Medium has the same amount with torches. Hard would probably be okay with six guards with torches, some of which would be blackjack-proof.

[Radiatoryang]

But don't you think this will be too hard for new (or average) players to perceive and understand? Or would you have a static fallback? And if so, isn't that a lot of content and balancing to do, considering most players play missions just once?

 

It seems like dynamic patrols assume the player has habits they may not have yet. (And for the record, Randy Smith's GDC 2006 presentation comes out specifically against long static patrols and branching patrols, so there's one privileged perspective.)

Edited July 16, 2011 by Radiatoryang

[Sotha]

But don't you think this will be too hard for new (or average) players to perceive and understand?

 

They do not necessarily need to perceive nor understand it. The main idea is to enjoy the game.

 

I personally would be enticed to replay missions if they would not be totally the same what they were the first time around.

 

And if so, isn't that a lot of content and balancing to do, considering most players play missions just once?

 

Is there a point in mapping if the end product is a single-use entertainment that the user deletes after playing only once? This is the reason I want to try and make a mission that changes: replayability would make the mission a challenge that the player could undertake a few times.

 

Also, if it is successful, those who do replay the missions are getting more than meets the eye from the mission.

 

I'll see this trouble through with this mission. Once I get it out, player feedback will tell whether it is worthwhile to do this ever again. As a mapper one needs to try and learn new things to keep the building interesting. Without this pet project, I'd probably would have run out of motivation already.

[fllood]

The main idea is to enjoy the game.

Right. That's this all about .. both for players and mappers.

 

Some more thoughts and considerations added:

 

I think dynamics and randomness will suit classic "empty that place from gold" heists more than tense story-focused FM's. Also likely will add replayablitity in unlinear maps more than linear ones. And in a small rather than in a huge FM. ... Or in other words I think methods like dynamic AI path design, random loot placement, varying spots for objective items, etc... are imo. worth the efforts and would work great for increased replaybility in FMs like "Knighton Manor" or "Beleaguered Fence". But would probably less succeed for "Glaham Tower" or "The Transaction" type maps.

 

And yes, if you implement methods of added replayability do let folks know about!!

( Mission Notes in the downloader readme or where ever seam best... )

 

Whatever each of us feel about that, please keep on whatever you enjoy and feel worth trying to implement Sotha.

[Radiatoryang]

Huh? I don't...

 

Okay, I enjoy Thief when I understand the level layout / patrol paths / obstacles and use them to my advantage. Are you trolling me, or do you really play Thief without thinking?

 

In contrast, something like Serious Sam is a lot about reactions and reflex with not as much deep meditative thought.

 

I want more people to like Thief, so I always err on the easier side of things and make levels more intuitive / accessible. That's my thinking.

Edited July 17, 2011 by Radiatoryang

[Springheel]

Okay, I enjoy Thief when I understand the level layout / patrol paths / obstacles and use them to my advantage.

 

Everyone does. But the easier it is to do that, the less satisfying it is.

[Baddcog]

Where? Here? See below:

 

 

 

Then your claim is incorrect as I'm doing exactly the opposite. Long distance patrols are totally messed up by the random deviation matrix. Short range local patrols are static with dynamic twists, like I described earlier. Like I said, the idea is to have the static patrols make it certain that the critical areas are manned. The random patrol's idea is to increase tension as there are few always safe areas.

 

Any area could be potentially dangerous if the player hangs out there too long. A matrix-walking guard could approach from any direction. If the player has immensely bad luck, he could be sandwiched by three incoming matrix-walking guards, each approaching from different directions. Or then the player might get lucky and all the matrix-guys are on the opposite side of the map and he only has to deal with the local guards. Or then the average tension situation where he occasionally has to dodge one matrix-guard.

 

 

 

 

 

Wow.. That was a nice implication to make! Thanks and much appreciated!

 

 

At any rate, a progress report on the matrix thingy:

 

I've implemented the long-distance-AI-patrol-mess-up-matrix in my WIP and I have to say it works quite nicely. The matrix walking AI walk very unpredictably now, but do not make any bad looking kinks.

 

Setting this up involved no scripting or anything, just placing one path_corner per spot where the map spreads off in different directions. Then when the AI reach that point, he randomly chooses one of the directions in the crossroads. Then he chooses again in the next crossroad, and again and again until he reaches the starting point.

 

I've set up two matrixes which direct the torch wielding AI's to patrol the whole map street areas in a random fashion. They have a rare chance to deviate from the big streets and inspect dark alleys here and there. Setting this up did not consume much more time than setting up approx 4 ordinary AI static patrol routes (per matrix). Timewise this means a few hours, which I find acceptable and normal.

 

 

 

This is probably The Biggest Issue with the matrix approach. We will have to see. My expectations would be:

If the matrix-walking guards do not have torches, their impact is not that large.

 

But torch wielding matrix-walking guards certainly require me to place some crates behind which the player can duck&cover. This will certainly gonna be a mission difficulty level dependant feature: easy will have 2-4 torchless matrix-guards. Medium has the same amount with torches. Hard would probably be okay with six guards with torches, some of which would be blackjack-proof.

 

Then why don't you read back through more than my last post.

 

I didn't give it any fancy name or confusing drawing. It's a very simple technique with a very simple explanation that I already gave in at least 3 posts, no I'm not going to re-read them and quote myself.

 

I said that you make one main patrol, then you can have any random splits in that patrol, that can be loops of their own at most, or even just stopping to blow their nose at least. And you can do this anywhere in the path, and have any number of random splits, yet maintain the integrity of the main loop as a whole so the player has some kind of idea where the guard is going, but also there is element of surprise.

 

I also mentioned having some ai on statics to give the player a good sense of action, while having others on random to mix it up.

 

===============

 

I think the thing you don't understand is that a bit of variation in patrol doesn't really make that big of a difference in play to play.

 

Sure the timing might be different to slip through this corridor, or that...

 

But most of the time players don't memorize the map, THEN play it. That's selling them short. It's like they can only be in the moment if you force them by making patrols random???

 

No, when I play I am in the moment, I am studying that guard to see when I can get by, then I make my move. I don't wait 5 minutes for him to do a full loop, then wait 2 1/2 more minutes for that sweet spot to come around again. I see him, I hide, I wait til I can slip by and I go...

(And I certainly don't put such small details into memory for the next time I MAY replay the mission)

 

So if I start a mission and go down the left hand street and see a guard it might have taken me 30 seconds to get there. So he'll be under a lamp.

If I then reload, and it takes my 45 seconds to get there, he won't be under the lamp.

 

So the players actions themselves lend a great deal of randomness without the author doing anything BUT static patrols. Maybe on the second load I decide to go down the right street instead, and there's a whole new set of consequences I deal with. Again with nothing but preset static patrols.

 

Maybe I reload 3-4 times... Every time things change due to my own timing. And each time they do I won't notice an ai's randomness much, because I have my own randomness.

 

And in none of the discussion has anyone even mentioned this, the fact that the player themselves have as big of an impact on the randomness as the author has with random patrols.

 

Then between ai there is randomness. One patrol does his loop ion 5 seconds, the next in 11. The LCD is 55, so the player has to wait for both of them to finish their loops 55 times before the timing lines up again. That in itself can cause large changes in what occurs.

 

So if the player can memorize all that, for each ai, for each reload. THEN play a mission again and notice a difference, well I think they should go to Vegas and count cards for a living.

 

The point is, with all that changing, to expect a player to realize somethings changed between run throughs, I think is still pretty far fetched, whether they claim to or not. When I played Madro. The lady outside always walked the same, and I climbed around inside and out several times and all I can say is guards were walking around, and I avoided them.

 

Sure, If a guard walks in a small square pattern for 30 minutes on one load, and he walks in a zig-zag for 30 minutes next load it would be easy for anyone to notice the pattern. But once he leaves the room, and you don't see him for 5 minutes, if ever. Will you REALLY notice the pattern change in one play or many?

REALLY?

-------------

Further more, imo you need more than some random patrols to make a mission noticably different play through to play through.

 

Take Witcher 2 for example.

 

If you make friends with so and so then you align with the good guys. Or if you kill so and so you align more with the bad guys.

 

Then the game CHANGES noticabley, because different options become available. But you only notice it IF you play through twice AND make those keys decisions differently.

 

But if you play the same way twice and one ai is walking in circles instead of straight back and forth, does it really change the game play enough to be noticeable?

 

Again I'll make this point and you can quote me.

 

'It's easy for authors to have allusions of granduer on how a random patrol will effect the player, but 99% of the time the player won't even be aware of the effort'

they won't know if the guard is in a different spot because of their timing, the guards timing, or random patrols. They will only know they have to be in the moment and deal with that guard IN THE MOMENT.

[Baddcog]

Everyone does. But the easier it is to do that, the less satisfying it is.

 

Exactly, but the 'static' paths are only noticeable if they are in very small areas and happen continually. Or unless the player 'stalks' the ai to memorize a long patrol.

 

In which case the player is no longer 'playing' the mission. They are goofing off, they might as well stack crates.

[Sotha]

@BC:

that was almost tl;dr. And yes, I am aware most of the things you mention there.

 

I will not go back to read your posts again, but if you first said in this post that there is no point in putting random stuff in long-range patrols and then you say you did actually recommend it and then you say there is no point since the player does not spot it, then you are contradicting your own posts.

 

But whatever. This is not the main topic and I will not discuss about it anymore.

 

Main thought:

Regardless of whether the player spots the randomness or not, I find the matrix-thingy interesting and that is why I want to implement it. Something new. If it rewards players who play missions more than once, then great.

[Springheel]

No, when I play I am in the moment, I am studying that guard to see when I can get by, then I make my move. I don't wait 5 minutes for him to do a full loop, then wait 2 1/2 more minutes for that sweet spot to come around again. I see him, I hide, I wait til I can slip by and I go...

 

I frequently wait until I understand when guards will pass by certain areas. When I'm in a room that AI come into regularly, I very quickly learn how much time I have in between to search. Just last week in Alberic's curse I had to search a room for something small, and an AI came into the room as part of his patrol. By the time I was done searching I knew exactly how much time I had in between to take my lantern out and search, as well as knowing approximately how long the AI stood with his back to me (in case I wanted to KO him). Guess what happened to the tension once I figured it out?

 

Most AI patrol relatively small areas in maps that are only medium-sized to begin with. It is a rare patrol that takes longer than two to three minutes to loop at least once.

 

In which case the player is no longer 'playing' the mission. They are goofing off, they might as well stack crates.

 

Observing the AI to learn their routes is "goofing off"? We obviously have very different play styles. I would call that one of the main points of the game.

[SeriousToni]

I am following this discussions with interest and I am also glad, that I have asked this question as there's so much to talk about

[Professor Paul1290]

Okay, I enjoy Thief when I understand the level layout / patrol paths / obstacles and use them to my advantage.

 

I do too, to a point.

 

I also enjoy Thief when things go wrong for me. My favorite memories from a lot of games tend to be when something doesn't go according to plan, when something shows me my understanding of the situation is flawed, and when I have to adapt and work around it.

 

If I go through a whole mission and everything goes as I expect it to then I get bored. I like having a plan, but I also like having one that I can quickly revise and adapt as the situation changes and I like it when a game challenges me to do so. If everything always goes my way and I already know the layout, patrol paths, and obstacles then there's not much more for me to say about it.

Edited July 18, 2011 by Professor Paul1290

[New Horizon]

Observing the AI to learn their routes is "goofing off"? We obviously have very different play styles. I would call that one of the main points of the game.

 

As would I. It's one of the tactics I use, and certainly one of the things the developers of the original games desired the players to do.

[fllood]

I want more people to like Thief, so I always err on the easier side of things and make levels more intuitive / accessible. That's my thinking.

Surprising opinion Radiatoryang

 

That had been the PR Mantra Ion Storm told us all the time during development of TDS as excuse just to dumb down down Thief making it console audience "compatible". The result speak for itself I think!

After many disappointments in game sequels with "improved, easier accessibility" I meantime avoid every game which has the words "broadening the audience" or "more accessible" in the PR & marketing bla bla... and welcome any bold game and mod that is courageous not to succumb to the mindset that accessibility stand for bringing gameplay to the lowest common denominator (hence my sig. quote). I think a lot more players are into quick-witted or unconventional game design as commercial publishers do embrace these days. One reason some great mods and an increasing share of indi games fill the gap.

 

I do belong to the group who enjoy Thief most "when I understand the level layout / patrol paths / obstacles and use them to my advantage" and voted static (but would have 'static with dynamic variations' if choice given). But the same time I think that "dynamic" map routes could make sense and will be fun if the the TDM map is suited and build focused for the kind of gameplay. Increased diversity of map styles is my favorite kind of "broadening the audiance".

Edited July 18, 2011 by fllood

[Baddcog]

 

 

Observing the AI to learn their routes is "goofing off"? We obviously have very different play styles. I would call that one of the main points of the game.

 

I love how people on this forum tend to reword and put everything out of context.

 

Like I already said, I observe them too, long enough to get by/get done what I have to.

 

What I said due to 'goofing off' is following an ai on a ten minute patrol to learn the whole patrol. It's one thing to wait for 1 minute to learn where your safe spot is and to follow the ai on an entire patrol so you know where that one little sweet spot is.

 

Honestly Spring, do you really follow ai around a whole patrol to learn that little sweet spot? This is what I said and I guess you do since YOU said that's your playstyle.

 

Did you wait for that guard in that room to see when you had time to find the item, or did you follow him and memorize his every move?

[Baddcog]

As would I. It's one of the tactics I use, and certainly one of the things the developers of the original games desired the players to do.

 

Exactly, I already said I do too. But I don't wait for them to do it ten times. And I certainly don't replay it thinking 'this guard does this everytime'.

 

And just as you said, the original devs desired this too. That's why they used STATIC patrols. So you can watch the ai, learn the timing, and go. No original Thief maps to my knowledge had ANY random patrols at all.

 

------

@ Sotha, no I'm not contradicting myself at all.

 

Once again if you can clear out what you already think and read with an open mind you'll see that I'm not.

 

I said having some randomness in a long loop is fine. But having a ton of random patrols will be lost on the player. Unless you play like Springheel obviously does and you follow an ai on their patrol enough times to learn every little nuance. otherwise the player will only see that ai a little here or there throughout the entire mission.

So on a long patrol the chances of a player noticing any random behavior at all is slim.

[Sotha]

Did you wait for that guard in that room to see when you had time to find the item, or did you follow him and memorize his every move?

 

I'm confused.

 

Why ask something like that since it is obvious no normal sane person would follow an AI guard only to memorize his every S I N G L E move? Making such a radical conclusion over the material posted herein is slightly disturbing. Sounds like a provocation to me. Surely we do not aim to annoy and provoke our beloved forum friend here?

 

I'm pretty sure 'memorizing an AI path' simply means the player gets a simple gut feeling how long is safe. Everyone does that, even if they do not actively think about it.

 

@ Sotha, no I'm not contradicting myself at all.

 

Whatever you say, sir! I will not dwell on this anymore.

 

So on a long patrol the chances of a player noticing any random behavior at all is slim.

 

You are correct, the player might not notice the random behavior. But it still may improve the mission and give it larger replay value!

 

I've learned that you are possibly a player who does his playing IN THE MOMENT, without memorizing every AI move. Here is an example:

 

You meet an AI in the streets he is going from A -> B. After him is his buddy. He's going A->B too. Freeze the moment, and analyze it! In a static setup these guys always walk here like that. When the player is in that area, he meets there guys always moving like that. Soon the player realizes that he can always hide in the dark-spot ALPHA since the AI's never go there. The player could go there, and leave the game running for a day and come back and continue.

 

Using the matrix walking guards, the player can encounter in the same location guards going in any direction, depending on the random events in the matrix and that the mapper placed enough nodes. Agreed, the player might not know this is random and that is great since random movement looks slightly unnatural for a human being.

 

But since these guards can pass through this area in many different ways, which is always changing, the player cannot feel safe. "The guards never come to this dark alley." But bang, they COULD! And the DO! The player runs into an unexpected situation. Even the mapper cannot anticipate it when and how his own creations move! All it takes is for the mapper to arrange a (low-level) possibility for the AI to visit every place. There is always the possibility, there is always the risk! Stuff a corpse in a dark alley? On one game session the AI never visit that. On another, the AI's find the body and raise the alarm.

 

And that is the benefit of the matrix walking guards. Let there be possibility for AI to be everywhere, any time and approach from any possible direction.

[Springheel]

love how people on this forum tend to reword and put everything out of context.

...

What I said due to 'goofing off' is following an ai on a ten minute patrol to learn the whole patrol.

 

Well, since you're concerned about quoting out of context, what you actually said was:

 

Or unless the player 'stalks' the ai to memorize a long patrol.

In which case the player is no longer 'playing' the mission. They are goofing off, they might as well stack crates.

 

You didn't say anything about following an AI for ten minutes, which would obviously be ridiculous (name me one TDM mission that even has a patrol that lasts ten minutes).

 

However, stalking and observing an AI to find out where they go and where safe spots are is a significant part of the game, in my opinion. You don't need to actually follow the AI on his entire route (though I've done that too), but you certainly want to keep track of how long he's gone and when you can to expect him to return, as I explained in my example.