Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Normandy]

Spoiler (click to show/hide)

Well, a little story first. I had been planning to try and have a roguelike demo up and running by now, and to have finished my first complete game by the time school rolled around in the Autumn again, but an unexpected exception was raised: Programming professionally (well, technically, I'm doing serious academic research in artificial intelligence, which translates to programming professionally) makes you really jaded about programming. Now I'm aimlessly hacking away at physics/math problems for fun instead of aimlessly hacking away at code for fun. Anyways. Let's backtrack a few weeks.

I had finally had enough to do with C++ (what, you can instantiate a broken reference and then make that broken reference call non-static non-virtual functions that break in ways even the debugger doesn't understand?) and libtcod (I don't get why everyone likes it so much. It's nothing special. Jice is a good programmer and artist but a bad designer). Moving over to C# and Sharplike, I realized that trying to use Sharplike without a guiding hand required at least a year or two of context that I did not have in order to make something legitimately passable. I decided "hey, you know what might be good learning? If I were to write my own roguelike library in Java, then try and compare my efforts to that of Sharplike and other modern RL libraries". The efforts of this were stymied by my research, as programming for that long just isn't healthy for anybody.

If you're interested in what I did manage to pump out, you can download my source code here: http://www.mediafire.com/?9myad5t8pj8vn7o. It's an Eclipse project, so you can import the project directly into Eclipse JDT. You'll need javax.vecmath (from Java3D), a very good library and one of the reasons I chose the Java platform. SimpleGame.main() is my current entry point; there are a few scattered Javadoc comments here and there, and a few commented out main methods doing unit testing (especially check out MultiConsoleView.main(), you may wish to uncomment it and run it yourself), so you may or may not be able to do anything with it. My main achievement was fast-scrolling/rendering and being able to layer multiple consoles on top of each other easily without any hassle whatsoever. Secondary achievements include a basic map/entity system; but I never quite got around to making the 'game' part yet, so you'll have to make changes to these classes themselves if you want to actually make anything with the library.

Anyways, thanks for reading my digression. Now on with the post.

I'm interested in a few questions regarding game design.

1) Do you think ASCII tiles can repeatedly give a feeling of grandness and magnitude? When you opened your first cavern in DF, or saw your first multi z-level chasm, how do you think the ASCII tiles affected your experience? Of course, a fortress lovingly rendered and painted in 3d with artists packing every nook and cranny with detail is ideal; but what is its equivalent in ASCII? How would you make an ASCII fortress or landscape beautiful? Is it an element of photorealism like Jice uses with The Cave? Or more colorful/detailed tiles like TileMagic? Or is the austerity of plain tiles enough to convey a sense of grandness?

2) How visceral do you prefer your combat? If I had to shoehorn the various combat schemes into nice categories for you to choose: Would you prefer real-time RPG (broad category encompassing Diablo, WoW, Fable, Morrowind, I would even venture to say Nethack), where combat is based on HP, separate attack and special abilities, and combat generally tends to be more predictable? Turn-based RPG (FF, KotOR, etc...) which is similar to real-time combat, except it's turn based? Tabletop (specifically something like the d20 system, since it's more ubiquitous), where combat is still based on HP, but there's much less of it, and combat tends to be less predictable, and there are fewer 'special abilities'? DF's combat system (mostly referring to limbs; the realistic force / material value aspects can be combined into other systems)? Or fighting game systems (think Street Fighter, Super Smash Brothers, etc...)?

What about in the context of a game like Shadow of the Colossus? Does the emphasis on exploration change your tastes in combat?

3) What is your take on item systems? Do you prefer material-item sort of generic items like that used in RuneScape or DF? Or are unique items more your style, as found in Diablo and just about every other game? I'm referring more about the specific item types, not necessarily any special attributes that individual items may carry.

4) Simply put: High magic or low magic? More classical firebolt or less classical naturish magic? Not which one do you think is more novel, I'm asking which one would you rather play.

5) Finally, do you prefer playing mods or competitive gameplay? I know it's possible to have both (see massively successful Warcraft and Starcraft franchises), but I'm just curious which is more important to you.

[3]

1) Use of ASCII has nothing to do with grandness. It is to do with perfection. Graphical representation will always be imperfect, usually because the representation itself is imperfect, but in the case of poor art direction, it may be that the interpretation of the concept is flawed. The use of (what are generally known as, even if they really aren't) non-graphical interfaces has none of these drawbacks. The supposed advantage of the use of more detailed graphics is that little imagination is necessary, and the viewer is capable of fully understanding the artist's original vision. However, I've never found that an artist's vision is beyond my scope of imagination - and once graphical representation is used, it's impossible to get the artist's rendition out of one's head. It's almost as if the use of graphics, in many cases, is simply to make up for a lack of good writing.

When it comes to good nongraphical visuals I prefer detailed ones. Incursion is a decent example.

2) I prefer fighting game mechanics. But many won't. You'll also have a hard time integrating them with roguelike mechanics.

3) Overemphasis on luck cheapens gameplay. This is possibly the greatest flaw in traditional roguelike design. Ideally, randomisation - and, indeed, procedural generation - should allow the player to repeatedly experience new challenges while still having a chance. Unique items, if included, should not be crucial to success. If they are, it'll just come down to players looking for them. Forever. This is most evident in Diablo and such, but it's also notable in games with limited predefined "artifacts", such as Nethack. The difference between the two variations is that Nethack's results are easily reproducible. Diablo's are not (I haven't seen someone make a Diablo II TAS yet. It'd be interesting to see, but it'd be hard - Diablo gets its numbers directly from the CPU, if I remember correctly).

Hack's interpretation is good. It was designed to limit the use of certain artifacts to certain characters. But it's not perfect by any means, since the items are still important enough to horribly bias gameplay. Dungeon Crawl handles the "artifact" idea better, since there's no way of directly getting randards/unrandarts other than finding them, but that still puts the emphasis on what is essentially luck and randomness. Hence, people will startscum. Forever.

4) Depends entirely on how it integrates with the mechanics, and secondarily on the setting. I'm more concerned about the implementation than anything else.

5) Mods. Competitive gameplay in anything with an emphasis on random chance is stupid. Stupid stupid stupid. Stupid.

[Supermikhail]

1) I like the default ascii. They can give whatever you want but they require some brain adaptation.

2) Tabletop, but more predictable. Because less hacking.

3) DF items. Hate collecting items, and always full inventory.

4) Low magic. In most high magic games I've played there was too much to learn and too little documentation. For me low fantasy is easier to wrap my mind around.

5) I'd go for competitive gameplay. For me mods are usually too much effort to make them work. I generally like when you can get into the game right away and see what it has to offer, instead of fiddling with config files. Also, in response to 3 - life is competitive RNG based and you can make it as fun or as sad as you like, depends on the attitude. And in games you can restart.

[3]

life is competitive RNG based

That's due to necessity, not design.

and you can make it as fun or as sad as you like, depends on the attitude.

Ignoring that this seems to be a terrible analogy, potentially, yes.

And in games you can restart.

You shouldn't have to restart. If you find that it's necessary for you to restart because things aren't going well, then the designer has failed. It is the designer's job to make something entertaining without unnecessary player effort.

[Ari Rahikkala]

1) 3d structures when the screen displays only a single horizontal slice of the world at a time, no. Scrolling up or down a mountain or chasm in DF tends to feel discontinuous, like it's just so many cliffs. Big two-dimensional structures (bigger than the 78x20 that something like ADOM is restricted to) can work really well in ASCII though, IMO. Don't forget that nothing's stopping you from augmenting an ASCII view with a bit of descriptive prose.

2) The most interesting fighting mechanics I've played in a roguelike was that of PrincessRL. Your fighter skills are discrete special abilities that are triggered by certain movement patterns or conditions. For instance, the "close combat" ability gives you a damage bonus when surrounded by walls, whereas "wall leap" allows you to jump off walls at monsters that are several tiles away in the opposite direction - great for kicking ranged attacker ass! There aren't any combos per se, but there's an item that gives you much-needed health for using special moves, so you end up using them a *lot* anyway

Otherwise, well... it depends. But IMO the roguelike genre is best suited for combat that's reasonably predictable so that the player can benefit from tactical play. Limb systems, locational damage, etc. are fun to code and read about, but I don't really know how good they are for gameplay in the end. Also, never forget, special abilities are for special situations: It's a rare roguelike that doesn't have hordes of low-level monsters for you to slaughter without a thought.

3) I haven't yet played a game where it made much of a difference whether items were explicitly made of some specific material or not. Though, if you do make items have materials, make sure that the player can know what the material is anytime it's pertinent. For instance, in ADOM, you're likely to find out early on that iron items can rust - but you aren't told that, for instance, swords of sharpness are made of iron, despite the fact that they have a better damage code than even eternium swords (which are most definitely rust-free).

4) I like to play magic characters in high-magic worlds, unless fighter characters have more interesting mechanics (case in point, PrincessRL)

5) Roguelikes are single-player games for me, so no interest in competitive gameplay here. I don't even participate in tournaments, share scores, etc.. Moddability in a roguelike isn't that important to me either (people seem to be so eager to get into modding to add random pointless stuff to what was a good, coherent concept originally), but a lot of it arises naturally from good engineering, so I guess I'll just say you should concentrate on that.

[Bricks]

1) In my experience, and looking at a few tech demos, you can create a pretty impressive scene with ASCII.  Most roguelikes don't attempt this, however, as they are often restricted to small maps, limited viewports, and few colors.  You have to play to the strengths of ASCII - it's very poor at showing more than one layer, so the "vastness" of something like the caves in DF is lost.  You also have to nurse the player into the display style.  An incredible DF map is so much text garbage to the inexperienced, and even old pros will have to carefully watch scenes to understand exactly what the hell is going on.  I've been thinking about my own little roguelike, and I think the best way to make it impressive is through full-color and lighting.  "Brogue" is a wonderful example of this.

2) In realistic art, especially portraits, there is something known as the "uncanny valley."  Basically, it means that an image can get close enough to realism that starts to be disturbing in its minute flaws.  In roguelikes, I like to think of the "unplayable valley."  I've tried to play some of the really deep roguelikes out there, and I quickly get mired in all the races, classes, spells, items, traps, monsters, professions, etc. that can screw me over or actually allow me to get into the game.  My absolute favorite roguelike is Doom RL because it uses a small subset of features that I can hope to understand and overcome in a relatively short time.  On the other hand, I love the combat in DF's adventure mode because it is largely intuitive, despite its complexity.  Ideally, I'd like to play a game like the latter, but a simpler combat system (perhaps even a rudimentary one) is probably the most feasible and most fun.  Doom RL works well because it has a few key strategic elements; cover and weapon choice.  I'm not sure what works best in a more traditional roguelike, but I've been thinking about a system with relatively few commands, like the Assassin's Creed games or Arkham Asylum.  If I have to start setting up hotkeys and macros just to use a key spell, I'm not having fun.

To answer your question more precisely, I like my combat fairly predictable (ambushes are fine, but game-ending critical hits are frustrating), turn-based, and with a limited number of commands that vary in efficacy according to your circumstances.

3) I prefer unique items over stepwise improvements.  Replacing a wood sword with a stone sword with an iron sword with a steel sword is boring.  If there are only a few such advancements, or the game is very long, I can see the purpose.  Otherwise, I'd like to see nearly every item remain relevant for the entire game, and focus on either character improvement or item diversification as the measure of progress.

4) If the game is about magic, and I am playing a wizard, I want throw a goddamn fireball at someone's face.  Otherwise, "lower" magic is better, as it keeps magic from becoming mundane or underpowered.

Spoiler: Longer rambly answer (click to show/hide)

I'm not sure there are any games out there that really capture "low" magic.  I just don't see how it would be playable.  It's more of a storytelling element at that point, and the only way you could rationally express it is if its woven into the narrative of the game.

High magic in the sense of Diablo is cool, since it allows for specific class development and prevents the player from becoming an all-out powerhouse.  High magic in the sense of Oblivion sucks majorly, since it allows anyone to be a master of everything purely through effort, and ultimately you are just reaching for stronger fireballs.

It goes back to my comment about items - it's not about whether the magic is "high" or "low."  It's about the magic's uniqueness and purpose in the game.  If you are going for a "classless" system where magic is available for everyone, you'll have to find some way to encourage the player to be a specialist.

Mana bars are OK, as long as I don't have to mash the "wait" key for it to recover.

5) Mods.  Competitive gameplay just isn't my style, and I really don't think it's appropriate for a roguelike (except in the context of final score and other achievements).  That said, I'd rather play a fun, hard-coded game than a bare-bones engine that is barely more than a level generator and a damage calculator.

Final Note:  For the love of god, no player needs like hunger or exhaustion, unless you are going the super-realism route.  The same kind of applies to things like torches.

In reviewing the other posts, I see 3 and Ari said a lot of the same things.  Cool!

[Normandy]

Ah, thank you guys for your responses. I've been considering it very carefully, and I think I'm dropping any 'true' concept of a z-level. Diablo did it just fine - going deeper really just meant going to an entirely new board. I think that descriptive prose actually is a fairly good idea - I hadn't quite thought of that before.

I'm actually rather a newbie when it comes to actual roguelikes (really, I've only played Diablo extensively; I have touched other roguelikes for perhaps a day or two, but playing them felt so... anachronistic), so I'm hoping to delve into them with a fresh perspective.

I'm getting a lot of stuff about keeping the universe minimalistic - should I interpret this as say, keeping the variety but the differences more defined? Even if that means sacrificing a certain amount of replayability (i.e. if you play DF with one of the zoo mods that add hundreds of creatures, you can play multiple fortresses in a row without encountering the same animal twice)? If you read my rambly spoiler, I've been working on AI systems and I'm interested also in changing up the AI between different creatures, so I'll see how that pans out.

I've been considering a real-time combat system which is segmented into turns depending on your character's speed, reaction, abilities, etc... rather than a more genuinely turn-based system. I'm considering fighting game combos; but I don't really know how to translate the idea of a 'tilt' into a RL. Tilt-punch shooting a fireball is slightly less immersion breaking than move forward one step-punch shooting a fireball, IMO.

Anyways, it's kinda late over here, I'm out. But thanks for the feedback - hopefully some more people will join the discussion.

[Supermikhail]

Hey, have you been here. I found it quite fascinating, on topic of rl AI.

[3]

I'm getting a lot of stuff about keeping the universe minimalistic - should I interpret this as say, keeping the variety but the differences more defined?

Essentially. The main danger is overloading on secondary mechanics and apparently minor details which are actually vital to gameplay.

[Muz]

1) No. But I love detail and by having no graphics, you can describe more detail. I'd rather see a game where you can see what weapon the character is holding. In the case of ASCII Dungeon Crawl and Dungeon Crawl with tiles, ASCII was inferior to the graphical version in feel.

2) I love detailed combat too. Fallout is pretty much my favored balance between too much effort and too little detail - it provides a good, realistic description of how combat happens. I think it matters a bit more in any hack and slash games, which roguelikes usually are.

3) Roguelikes shine in that they're replayable and give different experiences every time. I like my items, even artifacts randomly generated. But I love "set items", in that you have special artifacts that become stronger together. I actually hate DF's material item systems... gabbro chair vs orthoclase chair doesn't mean anything to me. They're just excessive info.

4) Both are fun. As long as they're consistent and work into the story.

5) Depends. For most games, mods don't mean anything. For an epic engine like DF or Civilization, they have room to shine. If you're going to put a lot of work into the engine, moddability may be worth more. Otherwise, any other feature is better, IMO.


EDIT: Fell asleep at the keyboard, apparently.

[Lap]

1) ASCII is shit. Switching between playing so many different roguelikes is a huge pain when 'D' in one means a huge demon and it means doppleganger, dog, or dagger in another. All it does is waste my time and cause needless confusion. I don't mind low res graphics or tiles and I could care less if I have to play with an anime tileset if that's all that is available. The purpose of the art is easy recognition of what is going on; my mind will fill in the rest. My recommendation is to use the simplest graphics you can get away with that aren't ascii and then make it mod friendly so once your game becomes popular you can attract a good artist.

Seeing in 3d with 2-planes is a huge pain for me too and it really doesn't do a lot of the DF forts justice. Only when finally seeing the DF visualizer shots of some of the epic fortresses did I have any idea how epic they actually were.

2. I'd prefer either very simple like Dungeon Crawl. The amount of time that went into DF's material system doesn't seem like a good payoff. Though the logs are fun to read the interface nightmares that come with trying to display combat information, creature status, and character control seem to be not worth the trouble. Not to mention it's an endlessly complex source of bugs.

3. I am a strong supporter of random items and don't understand how coders can't be either. It's easy to code and provides a massive amount of replayability for the time you spend coding it. Unique items can always be added so that's not a problem at all. I also think that most all descriptors on items that are purely cosmetic be buried deep inside an extended information panel and not shown up front.

The one thing you need to be careful with random items is their rarity. If they are so common that I'm throwing out 20 magic items every time I go back to town (a la Diablo) they really lose their special charm.

4. I prefer magic to always be powerful. This doesn't mean rare, but it usually correlates. What I mean is that I don't want wizards to just be 'fighters that attack with a magic bow'. As with the items, too much magic and it loses it's awe. The game Dominions had a nice option. You could play the game in three eras. Each era had different strengths and rarities of magic. Having a simple option for magic item/monster rarity might be nice, but it does add some extra work.

5. Mods. There is no room for competition in roguelikes, especially with how random they are.

[Normandy]

Hrm. Although I appreciate the responses about ASCII vs. Graphics, my original question is if a symbolic mode of display really could lend itself to epicness. It's a question of whether or not it's moot to try and create breathtaking ASCII (or otherwise symbolic) landscapes. It's a subjective thing, indeed, which is why I'm asking people about it.

[Lap]

Hrm. Although I appreciate the responses about ASCII vs. Graphics, my original question is if a symbolic mode of display really could lend itself to epicness. It's a question of whether or not it's moot to try and create breathtaking ASCII (or otherwise symbolic) landscapes. It's a subjective thing, indeed, which is why I'm asking people about it.

I never have nor do I think ever going to find an ascii landscape 'epic'. Your time would better be spent on making other things epic.

[Normandy]

Not even: http://www.youtube.com/watch?v=TRwRudGpnR8?

[Lap]

Not even: http://www.youtube.com/watch?v=TRwRudGpnR8?

It looks pretty and flows nice, but 'epic' doesn't really come to mind. It's just too hard to get such a grand sense of scale without 3D. I do like the style and I prefer color blocks like this video and the landscapes of Cult more than I like using normal ascii characters for them.

Also, at this point you're just using color blocks as big pixels and I really don't even think of that when I think of ascii. It looks good but with the amount of animation and effort I'm seeing it's almost to the point where I start to wonder 'why not just go all the way and use normal image based animation?'

If you're making the graphics similar to this video because of technical limitation or that you don't want to find a separate artist I totally understand though.