next »

[Morroque]

I second the idea of asymmetrical multiplayer stealth. I've always wondered what it would be like to have a game where a player operates a large set of security cameras and merely viewed them all through the monitors. As long as the guardsmen players could only remain on a central location of the map, I'd imagine a good part of the game's tension would be just having the assailant wait safely on the far edges of the map, as if being picky about opportunity. (Then again, asymmetrical multiplayer is a genre I've always been interested in.)

When it comes to stealth, I have a rather strange history with it. A while back I was invited to join in the games on a Halo: Custom Edition server with some friends. At that point I hadn't even played an FPS in a few years, and I had never played Halo before. So not only was I incompetent, I was also on a bad connection: where my bullets only hit the other players two to three seconds after I thought they should. This led to a strange phenomenon where I could be packing six frags, ten rockets, invisibility, and fully powered shields, yet still be somehow a sitting duck.

So I ran away like a coward. A lot. I tried being a camper with a sniper rifle, but my aim was too off, and I usually always gave away my location just by even existing on the server anyway. So one day, on during CTF on a relatively complex map, I just decided to try and take the most roundabout, obtuse way to the enemy camp and just swipe the flag. Thankfully, everyone else was too busy shooting at eachother to mind lil' ol' me, and to my surprise, they remained that way even while frantically on the hunt for me. This pattern repeated, and I soon earned my title as "Solid MW" for any CTF match.

However, it took a bit of getting used to, especially since the game of Halo really isn't adjusted well for stealth, especially since Master Chief's feet must remain firmly planted on the ground at all times. A lot of my practice came from planning out my routes on the same maps in solo far ahead of time. It was only through the process of general memorization and effective planning that I got a hodgepodge version of stealth to work, and even then it was only good on very complex custom maps. Continual use of Blood Gultch was always the most difficult to win with.

Stealth games are kind of like really actionful puzzle games in a sense -- just like the Castle Courtyard in Zelda 64, only instead of just being thrown out to try again, failure gets you savaged by a pack of mad hounds. Maps and levels usually have a very singular way of beating them, and the real experience of the game only comes from the first time playing it, and after that their continued success at the game is only bolstered by the fact that they are too familiar with the enemy environment already. They see the puzzle already knowing its most optimal solution. This makes me wonder if a procedurally generated stealth game, or even a stealth rougelike, would even be possible for a player to complete.

You know, I actually kind of figured a way to use bad guard AI to my advantage. Once I made a stealth joke game which was a continuation of my Halo CE history. While it never got past alpha with one complete level, it had a lot of stuff of the standard stealth genre. Yet, the most fun and enjoyable part of the game were definitely the banana peels.

The banana peels actually took the guard's easily exploitable patterns and actually made it a strangely satisfying gameplay goal. For some deep, intellectually primal reason, seeing the dumb guards literally fall for it just never got old. I'd love to make an entire game out of just stealth and bananas. Seriously.

[Malice]

I was wondering while playing it why do all these stealth games have fixed patterns that the guards always follow.  It makes the game just about patience and timing and observation, but if the AI created its own routes as you played, you would really have to be much more alert and aware of what is going on around you.  You couldn't just hide in that one area you know no one ever walks to. It would make a very different style of stealth game, but the current idea in the market seems to be stealth game = fixed patrols. Bleh =/

In general stealth games are about having a large number of resources and systems that can be directly manipulated by the player, and the interesting part is choosing which resources to employ to affect these systems in a desirable way; patrol routes are the system by which guards move around and guards are generally the obstacle to completing the level, meaning patrol routes are one of the systems the player is struggling to affect.

Static patrol routes are generally designed with a few intentional holes in place - ways to spend different kinds of resources, performing different actions, in order to circumvent them. The challenge is in recognizing and exploiting these holes in a way that is most favorable to you. In a Splinter Cell game, you might employ a sticky camera to distract a guard in order to modify the patrol route to how you want it, which loses you one camera in exchange for creating a hole. In MGS you might leave a claymore in the path of a guard, losing it in the process but also killing him and creating a hole to exploit.

But since you can't always predict what resources a player will have at a certain stage - maybe I'm out of sticky cameras, or claymores, etc. - static routes also generally have predesignated holes that you can slip through with a bit of skill with timing or dexterity. Such as slipping between two guards when they turn at just the right moment.

Dynamic patrol routes make the game less about manipulating a set system and more about reacting to random elements. There might be little or no way to ensure that the player can beat a certain stage because as a designer you don't have a clue what the conditions of that stage will be.

This isn't necessarily worse. Loads of roguelikes do the same thing with procedurally generated levels. You just need to understand that it changes the focus of the game from resource management to risk management. Many of the most popular stealth games are firmly in the former camp, and it's probably worth considering why - risk management can be very frustrating in a type of game where even the slightest realization of risk (getting noticed) may mean failure. In a game about risk management, you need to ensure that most risks are not game-ending, or can be mitigated in a huge number of ways. So you have to design a bit differently.

[Tifu]

Dynamic patrol routes make the game less about manipulating a set system and more about reacting to random elements. There might be little or no way to ensure that the player can beat a certain stage because as a designer you don't have a clue what the conditions of that stage will be.

But the designer does have a clue, if the dynamic paths are well designed. You can make the AI choose paths that none of the other guards are also wandering down, so you never get three guards in a corridor all watching each other's back or such. You can give the player ways to subtly modify guard's behaviour (MGS: tap on walls to create a distracting noise). There are ways this is entirely possible.

Also, completely forgot this when posting before: in Batman: Arkham Whichever, there is this kind of dynamic patrol AI - though it only really happens once you begin to terrify the guards.  Once they are sufficiently scared (by finding KO'd bodies) they'll start teaming up, looking over their shoulder, deviating from their normal routes, randomly investigating hidey holes you may or may not be in, shooting at the ceiling... and for me, that bit was the most fun :D

[stevesan]

Dynamic patrol routes make the game less about manipulating a set system and more about reacting to random elements. There might be little or no way to ensure that the player can beat a certain stage because as a designer you don't have a clue what the conditions of that stage will be.

But the designer does have a clue, if the dynamic paths are well designed. You can make the AI choose paths that none of the other guards are also wandering down, so you never get three guards in a corridor all watching each other's back or such. You can give the player ways to subtly modify guard's behaviour (MGS: tap on walls to create a distracting noise). There are ways this is entirely possible.

Also, completely forgot this when posting before: in Batman: Arkham Whichever, there is this kind of dynamic patrol AI - though it only really happens once you begin to terrify the guards.  Once they are sufficiently scared (by finding KO'd bodies) they'll start teaming up, looking over their shoulder, deviating from their normal routes, randomly investigating hidey holes you may or may not be in, shooting at the ceiling... and for me, that bit was the most fun :D

Keep in mind though that Arkham Asylum was a very different stealth game from most. I loved it, but the challenge (and fun for me) came more from being efficient rather than getting past the level at all (which in itself, wasn't ever too difficult).

Just as a general rule of game design: You can do ANYTHING you want, you just have to make sure that everything else in the game supports it properly.

[EdgeOfProphecy]

Thief 2 remains my favorite stealth game of all time.  It's a little weird how it was able to perfectly nail first person stealth, and so many games that came after botched it.

I'm not sure there's anything particularly revolutionary by today's standards in the mechanics, but the game did a lot of things right.  They set up a very rich world for the player

It had a good light and sound engine.  Different materials you walked on would make different noises, and lights were dynamic.  Guards had very good eyesight if you were in broad daylight, but if you were in total darkness they could walk within a foot of you and not notice.

The AI is pretty good.  I mean, this game is old but they chose the right parts of the AI to focus on.  Guards in the game are pretty aware of their surroundings.  They'll notice bloodstains, or even if the lights in a room were changed.  Sometimes they'll flip lights back on or relight torches.  They'll enter a search mode if they get suspicious, and will fight if they catch you.  The guards are stronger than Garret in direct combat, and very tenacious in their pursuit.  However, they'll run away if they're wounded (screaming the whole time), and Garret's more agile (none of that Assassin's Creed parkour bullshit, but the man can do some pullups), so it's possible to escape.  The one thing guards don't notice is stuff that's stolen, or things that have been moved around in rooms.  That would have been cool.

What was also interesting is they had non-combatants, civilians and the like, with a different AI that focused on fleeing and seeking help.  While a civilian wouldn't kill you, if one got away they'd probably bring back a whole slew of guards.

I think the real success of Thief (beyond the core sneaking mechanic being well executed), is how totally committed they were to supporting their core mechanics, and meeting the expectations of the players.  This ended up making the world highly interactive for the time, and gave me lots of great moments while playing it.

For instance, a big focus in the game was augmenting your environment to make it sneak-friendly.  Your bow had a slew of magical arrows available for it.

Water arrows could extinguish torches and wash out bloodstains.  What's really cool here is that they establish early on, "Water arrows can extinguish fire" and all throughout the game that remains true.  Later on you encounter these terrifying steam-powered sentry robots.  At first they seem neigh unkillable, but on their back is a small open vent to their furnace.  They never explicitly tell you this, but if you shoot a water arrow into the vent it will extinguish the furnace and shut them down.

Moss arrows could grow soft patches of moss on the floor, deadening the sound of your footsteps.
Rope arrows would embed in anything wooden and deploy a rope that could be climbed.
Noise arrows will start making suspicious sounding noise wherever they land for a few seconds.

Flash bombs were another great item that worked as a sort of "get out of combat free" card.  Using them would blind guards for a few seconds and leave them open to being knocked out (usually only something you can do when you're unnoticed).  Against a lone combatant, the flash bomb can let you easily win the fight, and against a group of guards it can distract them long enough for you to escape.  The flash bombs were great, because it allowed the designers to ramp up the deadliness of direct combat without making the game impossible.

Since the equipment was really powerful, they were able to design some pretty tough levels with few holes in the guard patrol paths.  They also had an awesome equipment system.  You got a base equipment loadout per-level, but you could use the money you stole from your previous mission to buy additional equipment.  That way the designers gave you a bare minimum needed to finish the level, but you could still customize your loadout.

Level design was well done in the game, too.  You had to sneak to get anything done, but they placed lots of alternate paths and shortcuts throughout the levels.  They also packed the levels filled with optional areas that had valuable things to steal.  This tied in very well with the whole "how much you steal is how much you have to spend on equipment for the next level" dynamic.

tl;dr:  Thief 2 is an amazing game and did basically everything I'd ever want in a stealth game.  It would be a good place to look for inspiration, I think, and if I had to make a stealth game I'd just make something like Thief 2.  I know that's not really "breaking the mould", but stealth is so consistently bad in games I'd consider making a really good stealth game quite the achievement.

[PlasmaMan]

Haven't had time to read the full thread yet, but I want to throw in that one nice possibility about "stealth" games could be how the difficulty works. If the player is spotted once, the guards naturally and probably permanently become more active, nervous, watchful, acting as both a penalty and as a difficulty curve-- say one evil facility or whatever is infiltrated, other evil facilities will naturally beef up defenses by deploying more guards or security systems, even if the player had a flawless run and was never spotted the first time. If the player plays far more action-y, a combined run-and-gun strategy, the enemy strategy would probably be to deploy beefier dudes or more weaponized defenses, increasing the challenge for the gunner player by making fewer strong dudes rather than a horde of weak dudes that could be mowed through effortlessly (though some of that could still be included. Sounds pretty fun to me).

To sum up, stealthy victories or losses means more perceptive defenses, gung-ho victories mean more powerful defenses. Of course, either way the enemy gets a lot more nervous with each sighting, which could be fun to play around with as a "fear" mechanic. I.e. just keep tormenting a group of guards until they lose it, call in for replacements, and leave before they arrive, giving you free reign for a few minutes.