Difficultopia! (GameDev Question)

[Oskuro]

Hello, and welcome to another thread where I ask for help in my quest to develop games.

Now that I'm done with the basics (Pong, Tetris, Block-Out, Pac-Man) I'm thinking of moving up to something a bit more complex. I'd say a platformer, or a top-view shooter (Both of wich I have somewhat outlined already).

Regarding the top-view shooter, my intention is to <naïveté>achieve the impossible</naïveté> by striving to do the following:

• a)Avoid design mistakes that would make the game suck

• b)Make the game extremely difficult, to the point of players swearing in swahili

• c)Despite the difficulty, still be fun to replay

My general idea regarding how to achieve this is by means of progressive difficulty increase, by deployment of increasingly bullshitty mooks of over 9000 power levels the longer the player stays alive, or by simply increasing mook numbers (Zombie-apocalyse style)

Note that I'm not asking "make a game for me", but rather, "give me your opinion on how to tweak the gameplay". The game is meant to be a fast-paced top-down shooter, meaning you see your character from above, with posible multiplayer, simplistic controls (as in shoot/don't shoot), lighthearted tone and general wackyness. Oh, and although I will use 3D graphics, the gameplay will remain purely 2D.

Any ideas/opinions on how to make such a thing while keeping the game interesting/rewarding/fun?
Wouldn't mind advice on things to avoid in order to have it not suck donkey balls too.

[Ford Prefect]

If you want a game to be extremely difficult, but you also want people to play it, the game must be an absolute blast. IE. I kept playing the most bullshit sections of the already tremendously difficult Ninja Gaiden II simply because in other areas of the game it was a wild ride. Because the game could be a serious gaming rush, I was willing to endure its stupendous cruelty.

[Stark]

Or there are games like Audiosurf where people go out of there way to make it harder, because playing it is so much fun.

[Covenant]

You're going to need to approach it like a schmup. Use what works--and what works for what you're talking about it a bullet hell game. Just make it manz or tankz instead of fightarz and you should be fine--and it's multiplayer combatable from the get-go. The form of increasingly ridiculous difficulty comes from the low tolerance for failure combined with a high degree of success if you manage to do a few basic things right--such as not die.

Remember, it's easy to kill the player. What you want them to do is fail in such ways that they feel personally responsible for it. Enemies with massive screen-clearing attack sequences that include windows of survivability play well with the goofballs who like that sort of thing.

[Feil]

If you want an extremely difficult game that isn't a pain in the ass, make degrees of winning and don't force the players to replay certain parts unless they actually die - or want to win harder. In stead of making the objective "get to the end of the level", make it "collect the five golden sandwiches to win". Then if you die or fail to collect any sandwiches the game says "lol you suck" and punts you back to the beginning of the level, but if you survive to the end and collect at least one golden sandwich, you go on to the next level, maybe at a lower starting health or something, or maybe just a lower score.

[Vendetta]

• a)Avoid design mistakes that would make the game suck

Make sure that the player's avatar is responsive and reliable to control. The rest takes care of itself.

• b)Make the game extremely difficult, to the point of players swearing in swahili

Lots And Lots Of Bullets. Also, make the barrages imaginative and pretty.

[Dooey Jo]

a)Avoid design mistakes that would make the game suck

That depends on how the actual design of your game would look. "Top-down shooter" is more of a format than any specific design.

b)Make the game extremely difficult, to the point of players swearing in swahili

What should be difficult? To survive? To get a high score? To get to the end? To save the children?

c)Despite the difficulty, still be fun to replay

There are a many things you can do. For example, you can make things random, so there will always be a somewhat new challenge. Or you can do the opposite, and have things deterministic so that every time the first parts of the game are played through, it gets easier due to the player's knowledge. That way the player can feel a bit like an ace, while still getting his ass kicked by the giant enemy crab in stage 10. Eventually it will get old, though.

[Starglider]

For example, you can make things random, so there will always be a somewhat new challenge. Or you can do the opposite, and have things deterministic so that every time the first parts of the game are played through, it gets easier due to the player's knowledge. That way the player can feel a bit like an ace, while still getting his ass kicked by the giant enemy crab in stage 10. Eventually it will get old, though.

There are a few people who actually like rote memorisation, but generally if you can only beat a level by dying 100 times in different unpredictable ways, that is bad design. For example good boss fight designs include some sort of warning sign about where the next attack will strike, so that a skilled player always has a reasonable chance to avoid them.

The most sophisticated kind of difficulty adjustment would be to do some kind of time/motion or AI-Monte-Carlo analysis of randomly generated attack patterns, such that the estimated reaction time, accuracy and (where applicable) short term memory required fall just within the limits of player capability. That's the makings of a PhD research project though, not a casual game project.

[Wing Commander MAD]

First off I'll state that I have just recently graduated with BS in CompSci, have taken three semesters of game design courses where I worked with windows GDI, DirectX, Zune/Windows/Xbox 360 XNA, Nintendo GameCube (not as fun at first as you might think), and am looking to get into the games industry. With my experience out of the way, I'll offer what help and advice I can as someone who was doing this kinda thing not too long ago. Seriously listen to anything that comes out of Starglider's mouth as well.

Can you be anymore specific in what you'll be doing (ie what language/rendering API you are using)? I can tell you right now that whatever you're using, you should avoid the urge to go crazy on FX and concentrate on gameplay. Get the core gameplay down and refined somewhat before you even think of getting fancy. Depending on where your models come from (are you making them or are they premade) I would really recommend desiging the game using basic geometric models before worrying about animations, fancy lighting effects, etc.. Since you're doing 2D gameplay, I would recommend trying to keep collision detection as simple as possible to speed up calculations (this is an area where the graphics API your using comes into play as I know the model formats DirectX based systems can have built in bounding which can make it much easier than trying to do collision detection yourself, though it is probably also less efficient depending on your mathematical skills, (I for one know anything done by an actual company or published in a book on game design are likely to work better than something I would code from scratch as far as accuracy and speed are concerned). In short resist the urge to make what ends up amounting to a tech demo and don't obsess if a particular technical effect can't be implemented the way you want, as that way leads to not ending up with good gameplay, to which I can attest from experience.

One gameplay aspects to consider seriously is randomization, which generally helps replayability more than rote memorization (though it does have its ups as has been pointed out). Personally I would put some decent work into your enemy spawning algorithm if you go the randomization route as that can make or break the enjoyability.

You should avoid omnipotent enemies as much as possible, especially avoid perfect accuracy if you have enemies that shoot. Likewise, enemies that track perfectly all the time are generally not fun either. It definitely makes them more difficult, but tends to make the game suck more. In the end its your call, though I have to question the Ninja Gaiden insane difficulty approach (maybe I'm just not enough of a masichist) unless you know the players like that kinda thing.

Varied power ups are a good idea, as are special abilities. People love screen clearing nukes as well in topdown shooters, the trick is to balance their availability of which there are a couple of schools of thought as to how to approach it.

As far as difficulty goes, insane numbers of enemies can be extremely fun and add to a hectic sense of survival, though performance issues can arise quickly depending on complexity of what your doing.

Controls are important as well for the game to be fun. You need them to be responsive, without being way over responsive such that minor actions by the player don't result in overexagerated actions by their avatar. This actually brings up another point, what kind of control scheme are you planning: keyboard, mouse, keyboard mouse combo, joystic/gamepad, a combination of all of the above? All the bitching about Resident Evil 5, is quite apt, the controls should not be what makes a game difficult. If the player feels that their losing not because of mistakes they made, but becuase the game isn't doing what they tell it to, you have a bad design.

Some kinda of story or something to keep it somewhat cohesive is also important. If the player has no sense of purpose or goal (highest score is even good enough) that engages them they probably won't come back to it.

Finally, you should have fun playing it even though you know it in and out and are fully aware of any exploits. If you don't want to sit there and play it yourself again and again after all is said and done, then most likely others won't ethier.

If you would like some more advice or clarification feel free to ask, and remember you probably aren't gonna make the next killer ap during learning and experimentation. Above all have fun, and enjoy what your doing.

[Oskuro]

Ok, some general info on my project structure:

• Graphics: Ogre3D (OpenGL)
• Language: C++
• OS: Ubuntu (Windows too eventually)
• IDE: Code::Blocks (MinGW)

I'm particularly interested in gameplay design as opposed to graphics, as well as simulation engine /game logic programming. I also have the design philosophy of "less is better", trying to make my designs shine by how they play rather than how they look, thus forcing my hand by setting pretty severe limits on the hardware-specs expected from my software. In other words, it must be able to run smoothly on my old Geforce2-powered system.

Regarding this particular development, tentatively named "Shooting Up the Crowd", I've set the following constraints:

• Minimalistic HUD: As in no ingame HUD, save for maybe highscores and info text.
• Simplistic Controls: As in it should be playable with an old NES controller.
• Pick'n'Play: It should be a game easy to pick up for a short burst of gameplay, wich affects not only gameplay design, but the implementation of elements such as savegames or checkponits.
• Moddable: Or to make as much of the game elements scriptable as possible, to be able to try different gameplay implementations on the same engine.

Now, regarding the issue at hand, I've come up with a rather gimmicky gameplay element that forces players to keep shooting stuff to stay alive (combining health meters with a time limit), wich is nothing new, but I'm afraid might be too frustrating or stressful. Also, relying on gimmicks tends to be bad.

And about the suggestions offered, the Bullet Hell gamestyle is one of the options considered, although I'd maybe prefer to have a Horde of close combat baddies cluttering the screen. Combined with the degrading health concept, these baddies wouldn't even need to hurt the player, just crowd him until it dies.

One early concept is a zombie apocalypse scenario, based on this Zombie Infection Simulator where each execution generates a random city, populates it with civilians, and then drops a single zombie in their midst, wich proceeds to infect the civvies, eventually taking over the whole map. I'm planning on dropping players in the middle of that. With guns. And maybe giant Zombie Crab Boss Monsters (of DOOM).

[Dooey Jo]

One way to come up with interesting ideas is to take a well-known concept and reversing it in some way. Like the zombie scenario; maybe one can play as a zombie who has to eat and infect humans to stay alive, but these then become competitors for the remaining living humans. You'd then get a pretty simple way to control difficulty by just controlling the density of humans and how long zombies can survive without eating. You could also have some humans fight back to make things even harder. The best to get them might be to get a bunch of zombies and let the humans kill the first ones and then get them as they reload...

This wouldn't necessarily be any sort of shooter, though I suppose you could make the zombies extremely disturbing and use chameleon-tongue-like feeding tubes.

Actually, you should make this game. I want to play it, but don't have the time to make it myself for at least a few months.

[Oskuro]

That's an interesting notion, Zombies that need to eat BRAINS to keep their health up, while puny soldiers keep pelting them with ineffectual ammo. And brains in jars as medkits. And a Thriller-Dancing Michael Jackson as a DDR-style boss.

Or maybe a Zombie, an Alien and a Vampire walk into a bar....


Ok, seriously now, that's precisely why I want to work on the modability of the engine, so those wacky ideas can be put on the screen easily.

[Dooey Jo]

I remembered that Gamasutra had an article a while back on a bunch of difficult games that were difficult in interesting ways. Maybe you can get some further ideas there.

[Oskuro]

Thanks for the link, it's a very interesting article. And I do remember playing through many of those and wishing all kinds of hurt upon the designers.

Apparently, someone has taken the concept of a difficult game and given it form HERE.

[Oskuro]

I must reiterate, Dooey Jo, awesome article. Not only is it a wealth of information regarding making challenging games, it's also educative on the shift in gameplay style in recent times. Also, it has some very funny lines, like the following regarding the Deadly Towers game:

How hard is it?

Walk over the wrong unmarked spot on the ground? Oh, what unspeakable sin you have committed! Your penance is to search for a secret exit somewhere within a gigantic death-zone filled with monsters who can just hate you to death.

Talking about hate, the AVGN did a review on it.

[Coalition]

One stunt, if you like Genetic Algorithms, is to have several different types of enemies each with their own attack styles.

Each one has a seed composed of 4-10 'segments' that govern how it will react every second, ranging from move forward, dodge left/right, fire, hold position, etc. The easier critters might only have 4 segments, while the fancier critters have more segments. The game just cycles through each segment. For higher skill levels, give the critters more segments/hit points/firing rate/etc.

For the beginning, make sure all the seeds are randomized. You don't want the expert critters from the last game being restored for the next battle.

You could even have segments for larger critters that tell it to spawn a weaker critter. I'm sure players will love this feature. If it gets destroyed by the player, the next one won't spawn as many critters or will dodge more. The carrier critter could have its basic instructions be to remain on the screen within X distance of the center, spawning smaller critters.

The game might spawn one critter, with the corresponding stats:
1) time started (in seconds)
2) current randomly generated seed
3) default stats for critter (including hit points, base movement, firing rate/type, etc)

When it is destroyed, the time between started and destroyed is compared to the average time for destruction. If it is less, that seed is discarded, and two of the other seeds are worked together (and a random bit is flipped) to create a new pattern. If it last longer, the average is updated, and its genes are added into the base data for the critter. If the critter kills the player, you add additional time to the critter (it really succeeded).

So as the game goes on, depending on how the player kills and is killed by the critters, they will get tougher to deal with.

The key is to make sure that all the options for the segment have results. I.e. a 2-bit segment for a basic attacker would have one of the following results:
00 - move forward
01 - move left
10 - move right
11 - shoot at player

4 random segments might wind up as 10000111, where as soon as it appears, it moves right, then forward, then left, then fires a shot, then it repeats the pattern. If the default movement of the critter is to proceed down the screen you will see it go down to the right, then down at twice the regular speed, then down to the left, then fire while moving down.

Only do one critter type at first, to give yourself practice. After that, you can have fun with different critter types that have different base movement and attack patterns, longer and more complex sequences, etc. You also want to make sure you don't do memory holes like crazy with something like this, when you spawn and delete critters.

It'll be fun for the CPU too.

[Alan Bolte]

I just want to say that there are two big mistakes I see in beginners' games:
1. Not enough variety: e.g. it's rare that you can get away with only two or three enemy types. Changing their color and upping their hitpoints isn't always a bad idea, but it doesn't count as a new enemy type unless they also have noticeably different AI. Even Pac-Man had different behavior for each ghost.
2. Quality mismatch: e.g. a high-end graphics engine with amazing lighting effects, bland scenery and awful animation; a mix of great music and really awful music.

Also, please avoid anything genuinely irritating that adds nothing to gameplay: especially awful graphical issues e.g. shimmering textures; a sound effect so horrible that you want to mod the game just to remove it; a difference in volume between one song and the next, or one sound effect and another, that forces you to turn the volume up and down. I see these even in professional games.

I guess that wasn't much about gameplay, but it's what came to mind.

[SAMAS]

When it comes to Platform Hell, I find the "Mess with your perceptions/assumptions" kind of hard more enjoyable than "there is only one exact way to do this and any other way results in death" kind of hard.

IOW, more stuff like suddenly-dropping floors and hidden enemies rather than rigidly-timed jumps over pits/through spike-lined shafts.

Also, at least one of those stages where the game's engine basically runs the course for you. Because those look like they would be stupidly hard if you tried to actually go through them (and thus goes with my whole mind-screw thing).

[Alan Bolte]

Blah, edit timeout. Anyway, I read the article, and I also wanted to mention that I loved Blast Corps! No one ever talks about that game, but it was so much fun. I only rented it a few times but never owned it, because I never had much budget for buying games but my parents or friends' parents would happily pay for rentals. Also, my parents and relatives never understood that people who don't play games shouldn't try to select one as a gift. Anyway, one of my favorite brief memories is of the time I convinced a friend to rent it. When his mother asked what we'd rented, he mispronounced the title. I can't properly translate it to text, but I'll never forget the way she shrieked, BLAST CORPSE?!. I hastily corrected him.

I also enjoyed Mischief Makers. Neither of those games really count as super-hard unless you go for all the collectibles and so forth. Seems like there are lots of games that aren't too difficulty, but with gold medals or whatever that are obscenely hard to get. It's interesting that they chose those two as examples. Maybe they really are above and beyond the normal in difficulty, I wouldn't know.

[Oskuro]

I agree with your opinion on newbie games, Alan, but that's a tough one. No matter how good the engine is, or how well programmed the game logic is, the presentation depends on the artistic style and capabilities of the creator, and that is something difficult to learn, to the point that it is sometimes stated that certain people are not gifted with artistic ability.

I say bollocks to that, to everyone who has asked me "how did you learn to draw like that" my answer has always been "practice". I spent most of my childhood doodling, and even then, my skills suffer if I go for a long while without drawing.

The point of these mini-projects is, precisely, to practice my game creation skills, both in the presentation and in the underlying game logic, so they will probably suck, but hopefully will help the next one suck less.

Now that that mission statement is out of the way...

One of the things that I kept thinking about while reading the Gamasutra article is that, even though letting players figure things out is interesting, leaving just a single option to solve the situation can lead to frustration. And still, although not telling players what all the items in the game will look like could be ok, not telling them what their character can or can't do is not.
Still, most of the articles were about puzzle/exploration games, and I'll leave such a game for a future project.

Also, please avoid anything genuinely irritating that adds nothing to gameplay

You mean, like the Turkeys from the South Park 3d game? For those who never played that (not too good) game, on the first level you were continually assaulted by waves of killer turkeys who kept going "bblblblbblgrrglglglg". It was so annoying, that I had people come over to my room (back in college dorm) to yell at me and tell me to turn it off. I'd have you know one of the enemy types I intend to use are based on those, specifically designed to be annoying as hell. And pink.

When it comes to Platform Hell, I find the "Mess with your perceptions/assumptions" kind of hard more enjoyable than "there is only one exact way to do this and any other way results in death" kind of hard.

I'm guessing you've played the first level of I WAnt To Be The Guy? That game is an exercise on frustrating gameplay. I agree with you that playing with the player's perception is good, but there should be ways to find out how to solve a situation other than standing on the trap and getting killed just to find out where it is.

[SAMAS]

How about building a delay into the trap, something that, if the player is paying attention to, they can avoid before it gets deadly?

Take, for example, the Drop-Away Floor Section bit: Make sure the player has just enough time to jump off before it falls offscreen. If you have a Spike-block trap, maybe make it an enemy, so that the moment before it goes off, you see it's eyes open.

Basically the challenge becomes, instead of repetivous (is that even a word?) difficulty, the difficulty comes from the fact that the game is rather aggressive about killing you, without the feeling of helplessness most Platform Hell games seem to fall back on.

That said, I do love the old "Invisible coin block just over the edge of the bottomless pit" gag.