Sign in to follow this  
u235

Thief players

Recommended Posts

Greetings, Some of you may have seen my post in the Help Wanted forum concerning the need for a 3D modeler and texture artist for my game that is currently in the works. Well it seems I am having some design issues concerning game play. Anyone who is familiar with the Thief series, by which my game is largely inspired, is encouraged to offer up what they think Eidos/Looking Glass/Ion Storm could have done to make the Thief series better. I have my own list, but I would like your input. Obviously I can't please everyone, but I will try my best to incorporate the improvements into the game. Thank you to everyone for any input. -AJ EDIT: I meant to put this in there, but somehow it didn't make it in: Even if you are not familiar with the Thief series, input as to features for this type of game will also be greatly appreciated.

Share this post


Link to post
Share on other sites
The various Thief games had strong and negative points for each. They also had a fairly devoted following that demanded a particular type of game - a challenging game of sneaking, one that didn't have easy ways out or ways that the player could "cheat" the system (i.e. the 3rd person camera in Thief 3 that let you look around corners).

If you want to get the thief players you'll need to make a game with variable levels of skill, and variable levels of completion - it will need to be easy enough for a "normal" player to beat, but with insanely hard enough optional elements that the thief players will attempt to complete in order to score perfect. In past thief games players often made up special modes of play to increase the challenge by forcing perfect stealth - you couldn't knock out guards, except for loot everything had to be put back as it was found (even keys), it had to be like you where never there (except for the fact that you stole 100% of the loot and completed the primary and all secondary objectives). The annoying robot enemies where also a low point of Thief 2.

As far as the actual game mechanics go, you can learn from what mistakes where made in each version.

On Thief 1 Looking Glass didn't know if the stealth gameplay would catch on, so they also made the horror levels - levels where the enemies where zombies, monsters and such that where not always fooled by stealth and had to be killed.

On Thief 2, they went full into the stealth mode of gameplay, and in many ways Thief 2 was the best of the series. The only bad part was that floors where often overboard - many levels where all metals and other materials over which the player had to move at a snails pace. The type of levels players prefered where often the mansions with patches of soft carpet, where guards would follow large overlapping patrol patterns. These created the best thief experience, as the player would hide in the shadows and actually come up with their own unique plans for how to sneak past the guards and get the loot, often involving precise timing to spring out from the shadows, run along the soft carpet, and duck into another room just before the other guard came around the corner.

Thief 3 finally fixed the graphics in the series, however the gameplay was destroyed to make a console player pleasing game - even the hardest level was easy, and it was even easier if you just went about the game either running past enemies without any stealth, or just attacked and killed everyone outright. Thief 3 also seemed to lose the bow handling of the first 2 games - in Thief 1 and 2, there was a skill to using the bow (mainly in estimating the travel time of the arrow and how far it would dip based on gravity). This made it extremely rewarding when you pulled off an amazing trick shot of shooting from one raised level to another, hitting a guard just as he appeared from behind a piller, killing him with a headshot from a distance of 60 feet.

Share this post


Link to post
Share on other sites
Thanks Michalson,
A lot of good suggesstions. Everything is going into a list for the moment and then the culling process will begin. I guess that will be when this thread has worn itself out. Thanks again.

-AJ

Share this post


Link to post
Share on other sites
Quote:
Original post by Michalson
..stuff..

Thats a rather good summary of the Theif games. [grin]

IMHO some of the important aspects of the Theif games (ignoring 3, which is a pale imitation)

- Suspension of belief
AI always does sensible things, never seems to have magical knowledge of the player's position, no holes in the AI which can be exploited.

- Consistant
Items interact well, AIs interact well, AIs notice objects presence/absense, AI hears all sounds, respond to all sounds, respond to other AIs.

- Open ended
Not just multiple routes, but basically big (realistic!) buildings and areas where you can truely go anywhere. Exploration is a large factor - there's rarely someone/thing to show you where to go, you have to think on your feet and pick up on the slight clues left lying around. Rewards players who are observant with hints/clues during the level (eg. overheard conversations etc.). Add to this the large range of equipment, weaponry and objects (see 'consistent').

- Rewards genuine skill/stealth
If you ended up exposed, and going toe-to-toe with a Theif guard, you'd better be careful. Opponents were generally pretty equal to the player in terms of health and strength, and would do intelligent things like run and scream for help when wounded. Going against multiple enemies was usually suicide (compare with any FPS, where you're massivly more powerful than even the most intimidating opponent).

- Multiple objectives == replay value
...and not just "replay the level with harder AI". Harder levels had additional enemies, less equipment, more places to go and objects to retrieve, harsher winning and loss conditions

- Fantastic atmosphere
Nuff said. [grin]

So basically, highly believable AI and environment, coupled with lots of quality level design.

Good luck with your project, you've got a lot of work ahead...

Share this post


Link to post
Share on other sites
I agree with Michalson

zombies = bad

I hated the zombie levels in Thief 1, the groaning/sucking sound they made freaked me out. I was glad they got rid of them for Thief 2 although the robots werent much better.

I particularly like the monologues and visuals between levels, I thought they did an excellent job of advancing the plot.

Share this post


Link to post
Share on other sites
Quote:
Good luck with your project, you've got a lot of work ahead...



Indeed, I do, hehe, but I think I am up for the challenge...and by think I mean hope, lol. Thanks for the suggesstions. AI is going to my main focus as it's one of the multitude of things I haven't really programmed much. I don't think it will be too big of a hurdle, though...I mean how hard can AI be? Guard sees you, guard attacks you, guard hears you, guard searches for you in direction of sound....sound pretty simple to me. Lol, just kidding...I know it's more involved than that, but from a logical perspective, it pretty much makes sense. Anyway, thanks again.

Quote:
I hated the zombie levels in Thief 1, the groaning/sucking sound they made freaked me out. I was glad they got rid of them for Thief 2...



And then they brought them back for Deadly Shadows.....I freakin' hate those things. And the Shalebridge Cradle mission, even though it's not that hard or dangerous still freaked me out crazy like. I'll never look at another puppet the same way again, lol.

-AJ

Share this post


Link to post
Share on other sites
I had a long reply of comments, but my taffing Firefox browser crashed on me.

I'm a huge fan of the Thief series, and in my opinion Michalson was spot on with the flaws of the individual games in the series. The first had those annoying zombie levels, the second (although my favourite) was very hardcore (especially in the higher difficulty levels), and the third suffered greatly from consolitis.

The list of good points in Thief outweighs the bad in my mind, but since you are asking for problems I'll summarise what I had before Firefox crashed.

My main problem was the A.I. was a bit too formulaic for my tastes. I think it would be a bit more exciting if each guard reacted a little bit more randomly to various stimuli. In particular, I didn't like how the guards would overract to a single loud noise. I also thought it was a little bit odd how guards would react to any small noises that Garrett would make, but wouldn't react to the absence of noise. The interior of most levels would sound like a cacophony of footsteps, which are gradually silenced by skillful use of the blackjack. Frankly, if I were the last guard standing I would be seriously spooked out by the eerie silence of the now empty building.

Actually, that's really my only real problem, as I liked (or tolerated) pretty much everything else. I agree with the positives that everyone else listed.

Share this post


Link to post
Share on other sites
Yes, the Thief series tended to combine both sneaking and a horror element - the non-sneak levels in Thief 1 could well be described as parts of a survival game (which go well with good horror theme). To be honest, I think keeping a bit of horror, in small doses, is a benefit to a thief like game. The darkness and tension created by creaping around, trying to be as alert as possible to the smallest sound, is the perfect setup to properly scare the user (for example the Cradle level in Thief 3 could scare you just from the sounds), compared to say a regular FPS, which isn't very good for horror (i.e. Doom 3 billed itself as a horror survival game, but frankly how scared can you be when your mission and "god given" abilities are to blow apart everything else in the game world).

As for AI, yes, it's not going to be an impossible challenge. The way to emulate/continue the Thief type AI is to create a simple state machine with a strong attention to the details and some descent scripting and voice acting for the various AI reactions. Essentially the AI guards should have a variable that controls awareness (Thief 1/2 had 4 levels, unaware, suspicious, searching, found), and a few other values to control things like combat - i.e. have a frustration/fear level, if the player keeps going somewhere they can't pathfind to within a reasonable time (ie they jump into a tree to escape the swordsman) or if they take too much damage, they should retreat and look for help.

The attention to detail is making the guards simple AI state affected by as many elements in the game as possible - light is out that shouldn't be out, object is out of place (weighted toward large and important items), guard not at his post - in addition to just see and hear the player. To really make the AI seem realistic, I'd give each guard a little table describing their "memory" of the world (the state of all the objects from the last time they encountered them, some weight values for how they should react). Then make it so that guards or other NPC (including non-player creatures) can "accidently" affect the world state (once you start the world going you'll need to do some major tweaking to make it not overreact and start huge chain reactions). So basically the player is hiding in the shadows a few feet from an unsuspecting guard - another guard suddenly decides he wants a snack, and changes his route to pass by the first guard on his way to the kitchen. The other guard hears this, and goes to investigate (upon finding the other guard, his "memory" of that guards position is updated, and the script "oh it was nothing" event is played back). In addition to level specific events/conversations, this makes the player feel like they are in a living, breathing world, which was a big part of the Thief series.

Append:

In light of Trapper Zoid's comments, I'd reiterate my guard "memory" device. Guards have expectations about the world (place of objects, place of guards, place of light source, place of sound effects). When that expectation is broken, it uses the weights to determine how much it has deviated to either change the guards awareness state, or simply fire an event and go on ("hmm, I could have sworn that was on, oh well"). Then once the guards awareness goes back down, the "memory" is updated - this way the guard only finds the open door odd the first time around.

Combined with the semi-random "accidental" events created by in game NPCs (guards can randomly decide to do different activies - use "The Sims" model where objects in the world advertise different types of fullfillment, so that a hungry guard will path find to a fridge, and a sleepy guard will path find to a bed) and you've created an amazingly realistic world through emergent behavoir.

Share this post


Link to post
Share on other sites
I must admit I have not played alot of the thief series, but the one element I did like was haw important and imersive sound was.. it was one of the few games where I would tweek my surround system, crank the volume, and insist anyone nearby be quiet.... listening for clues, gauging how close and in what direction the gaurds were, overhearing conversations.. was a great experience...

Share this post


Link to post
Share on other sites
I think "accidental" events would really go a long way to expanding on the existing prescripted conversations in Thief, plus they'd be a lot of fun - created events that can randomly occur when a certain combination of NPCs are in the same area - ie if there are at least 2 guards in proximity, there is an argument event, and a lower chance event of an actual fight (which would draw the attention of other guards). Or a guard and a dog in the same area - guard plays with dog, or dog annoys guard, or dog bites guard, or all kinds of other humorous situations (and many other combinations - guard and servant, 3 guards, peasant and town guard, etc). With only a few underlying variables and some lists you could create a very lively world (much like how The Sims works) - players might even just stand back in the shadows having fun watching the world go on around them.

Share this post


Link to post
Share on other sites
Very helpful. I was trying to come up with some sort of AI scheme that would be as realistic as possible, but that was pretty much a perfect example of what will happen. However, being that it was just an example I will I have to expand that many times over to create as much realism as possible. Also, level ideas. I've come up with a few, but I am no designer, not by a long shot. I don't know how to get a level just right so that it has the proper difficulty for each diffuculty setting. Perhaps a document someone could point me to that has some info about level designing? This is great though, I love these ideas y'all are putting out, which spawns more of my own ideas. It's just a shame that I won't be able to implement them all [sad] Thanks alot guys.

-AJ

Share this post


Link to post
Share on other sites
Typically a fun Thief map works like this:

Entrance: This is usually a safe area (no guards in the immediate area and no patrol paths, so if you stay in one place you are safe), like out on the street or beyond the wall of the grounds. From here there are multiple entances (2-5) to the "level" itself - one should be fairly obvious (like an open window to a well lit, empty room) while the others should be obscured and have varying amounts of time to enter - thief players like to explore, and they like their exploration to pay off. For example if our level had 3 ways in, and we already covered the obvious one (window), the next entrance might be a cell door that's hidden by overgrwoth (those double door deals that a low to the ground at an angle), which takes the player into the basement of the level. The last entrance would require more then just observation while taking the "move in the direction of the goal" path - it would require that the player go off the path to explore the world, perhaps finding that the servants house can be broken into (where the player will find a small piece of loot), and by following the stairs they can get onto the roof, from where they can jump onto the roof of the main building, which the player walks across and finds a skylight they can open and drop down into.

Compartments: Thief levels are usually divided into compartmentalized spaces (this helps the AI pathfinding, makes design easier, and gives the player at least a fighting chance of creating a mental map [Thief has no "exact" map, just hand drawn notes, so sometimes you go in circles, all part of the fun]). Each compartment is composed of interconnecting rooms and passage ways - there is more then one way to get from almost any one point in the compartment to any other point. A typical example is a square looped hallway - you have a hallway that makes 4 90 degree turns, thus folding back on itself. On the perimeter you have rooms and some of those rooms connect to their neighboring rooms or connect to other compartments, while inside the square you have more rooms, some of which have multiple doors to the hallway (i.e. they are in the corner and have a door to two different facing hallways) and some of which are to other rooms. The central idea is that the layout of the compartment is not linear - it feels more like a musuem, with many exibitions that are interconnected to allow freeform exploration.

Plot Points: Usually compartments are connected in a semi-linear fashion. This means there are early encountered compartments (the "entrance" can be thought of as a very open compartment itself), middle compartments, and late compartments that come together to describe the "story" of the level. For example a level might have the entrance compartment, which has connections to the west wing of the buildings (a compartment). The west wing has connections to both the east wing compartment, and the greenhouse compartment, and both of those compartments have connections to the basement compartment, where the vault and ultimate objective is kept (and there is probably one secondary objective in the green house and one in the east wing). The point of this is so that while the player has free exploration and so is basically writing their own story, they still can't jump right to the "end" of the levels story, they have to progress through the beginning and middle first. While thief has locked doors (which can normally be picked or opened with the correct stolen key), there are often "hard doors". Hard doors don't have to actually be doors - they are choke points, single (though sometimes the player may have a choice, each with a different challenge) passageways, often connecting an earlier compartment to a later compartment. To get by these hard doors the player must overcome some task - in games without a lockpicking kit, this is where the player would be expected to go find a key. However since you can pick locks, the challenges are handled differently. A hard door might be a drawbridge you need to lower in order to enter the inner keep (after having passed through the compartments that make up the outer keep), or a tight corridor with some seemingly impossible to pass guard.

Goals: The primary goal is placed in the compartment furthest from the entrance. Secondary objectives are spread around the other compartments, sometimes requiring travel between several. Getting to/completing the final objective or getting through hard doors can also require travel to several different places, to throw switches or collect items. Often the actions the player needs to take to complete the primary or secondary objective (say for example open the draw bridge) are described in clues - perhaps a guard mentions some mechanical fault in a pre-scripted conversation triggered when the player gets near, or the player finds a document giving clues about the location of three levers that must be thrown in the correct sequence.

Share this post


Link to post
Share on other sites
Thanks, that was nice of you to take the time to write all that out, previous replies included [smile]. I will definitely use that as a rough template for my missions. Ok, so back to the AI aspect. The idea about guards doing different amusing things while the player whatches. I was thinking about how I might implement it while I was at work tonight. If I create an event handler that receives an event every time the player comes near && there is more than one guard or a dog, or whatever the case may be, I guess randomly generated and the things that can happen change based on the mission...wouldn't want to see the same scenario every mission. Then the event handler determines which event it is with a switch, case, break thing and causes the associated scenario to play out. I am assuming I would need a different event handler for every NPC in the level, but if I don't, please let me know how I can do that. Also, if the idea is ridiculous, please feel free to let me know, lol. Thanks again for the help I am getting here, I really appreciate it.

-AJ

Share this post


Link to post
Share on other sites
I think the event AI could work in a similar way to the Halo AI. Any event has 'precursors' or triggers that need to be fulfilled in order for that action to take place. Dialogue and other fun and interesting events run off the triggers, a random generator and possibly a repitition timeout (to stop repition or events).

In terms of implementing this for the complex interactions (involving one or more entities) I don't really see any optimisations other than periodically testing a bunch of them (ie two guards, a guard and a dog etc.) and then randomly choosing an action that incorporates the situation. It only really needs to be tested every 5 seconds or so for the generic interactions.

[Edited by - umbrae on November 5, 2005 8:18:46 PM]

Share this post


Link to post
Share on other sites
This message half way just a bump, and half way just to help me get the word out. I bought a domain and put up a site vastly different from the one I had up before. Still not alot to it, but at least it looks pretty good, lol. I will get more content soon, like when the design doc is finished and hopefully I will have a screenshot or two soon. Can't guarantee a time frame, but I will try for next week sometime. Thanks again to all who helped me out, I hope this turns out to be a really good game.

-AJ

If you want to take a look the website is http://www.u-235games.com

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

Sign in to follow this