The simple trick to making good game AI

2011/03/06

Contention: Stardock’s Brad Wardell (Or is it Brad Wardell’s Stardock?) is often upheld as a good AI programmer, and Stardock games as containing good AI.

But from all I could ever see, neither was shockingly exceptional (and I’m not just saying this to be catty, though the Galactic Civilizations series, from the beta of the first game I played on OS/2 back in the day, was always something I wanted to like more than I ever actually did).

Good AI is like good Art Direction. You choose your battles to fight on your own terms so you can easily make yourself look good. Art design sets player expectations: VVVVVV isn’t criticized for its poor character modeling or voice acting because the game very intentionally sets its aesthetic limits at about the level of Atari games. Wii Sports avoids the expectations of any sort of detailed character animations or normal-mapped textures or environmental destruction and all such noise by utilizing a simple stylized aesthetic.

The visual experience of these games is designed so that asset creation is not an overwhelmingly complex and expensive job. And they manage to meet the requirements of player expectations without too much trouble. It’s good design.

Likewise, the gameplay of various Galactic Civilizations games happens to be designed in such a way that it’s easy for the AI to act meaningfully and responsively within its framework. (I don’t particularly enjoy this game design as a player, but that’s in the details.) Now: if a realtime chat interface was added to Galciv so that you could do diplomatic negotiations with AI opponents in real-time, as you would would a human, the game would fall on its face because it’s really, really hard to fight that particular battle.

The trick is to set expectations that you can fulfill, and fulfill well. This is why you won’t see me making an AAA-style 3d game.

Update:

There’s a great thread on Quarter to Three on the very subject of game AI and mechanics design scope with figures like Paradox‘s Johan Andersson, Vic Davis, and Civ V’s (now Stardock’s) John Shafer weighing in. Go there. The topic is discussed with far more depth.

Starfarer: Celestial Spheres

2011/03/02

The world of Starfarer has lovely 3d planets which roll beneath whatever chaos you are sure to wreak out in space. It is only right, for what good is space without planets to fight over?

Below: Flying an absurdly over-armed frigate past a jungle-covered planet.

Read the rest of this post on the Starfarer development blog

Gaslamp Games on the Immortal Machines podcast

2011/02/19

[re-posted from the Gaslamp Games blog.]

Nicholas and I were interviewed on the Immortal Machines podcast (an affiliate of Colony of Gamers) last week by the good bunch of guys there who were far too patient with certain Gaslampy … excesses. Now you can listen in to learn dark secrets about the Dungeons of Dredmor — about the unnatural habits of Thrusties, why Nicholas should not drink a pot of coffee on an empty stomach, and how much Dredmor isn’t going to cost. Here’s the link:

Immortal Machines episode 47: Gaslamp Games and Dungeons of Dredmor

Much thanks to the Immortal Machines guys for letting us on and to Colony of Gamers and their wonderful community for all their support!

The real question now is if they will ever make the mistake of letting us back on.

Dredmor Skill Icons

2011/01/28

I’ve just finished the latest round of revisions to the entire pile of spell icons. This is just one task which is part of the massive spell overhaul we’re doing for Dredmor’s beta 0.92 (when I’m not getting distracted drawing the disembodied heads of founding members of Gaslamp Games).

Man, there are a lot of these buggers, but they do get easier (and better) every time I redraw them. Telling you anything about them would ruin the fun*, so I’ve just thrown together a collection of some of my favorite spell and skill icons for your enjoyment:

dungeons of dredmor skill and spell icons

Still have to draw animated effects for most of these. Urrgh.

* whereas “the fun” refers to how much fun I have as people try to guess what the hell some of these skills do.

[Originally written for the Gaslamp Games blog]

Starfarer: Modern Ship Designs

2011/01/27

(This post is written to promote the game Starfarer by Fractal Softworks.)

There’s more to space combat in Starfarer than nearly-obsolete front-line sluggers like the Onslaught. Meet the Astral class capital-ship carrier: a modern, refined platform for supporting squadrons of small fighters and bombers such as the Dagger torpedo-bomber you can see below. Escorting the Astral is a nimble Wolf class frigate.

This Astral carrier has some top-of-the-line point defense systems on its starboard side while the port has some nasty repeating torpedo launchers. Once you get past the escorts and fighter-bomber cover, you’ll have to be sure to choose your approach carefully — and that’s before dealing with the full-coverage shield. Astral-class carriers should not be taken lightly.

What follows is a bit about how I went about designing the modern ship classes of Starfarer along with some concept sketches for the Astral and Wolf.

Read the rest on the Starfarer blog

“Binb is not Bomberman”

2011/01/19

Here’s a trailer for “Binb”, a simple but compelling game by Maxim Karpenko (posted on IndieGames at the end of December) for which I drew some cute animated pixel-art sprites and terrain:

Very bomberman, but it of course isn’t, as it’s given a somewhat different game format and goal, plus special items. And look at those cute little guys!

I’ll be sure to post the link to the playable game when it’s released somewhere.

Dungeons of Dredmor Trailer #4 & Release date announcement

2011/01/19

[Incredibly, this sat in my draft pile for ten days before I noticed that I forgot to post it. Oh well. All the action is going on at the Gaslamp Games blog anyway.]

Alright, I’ve been slaving in the pixel mines on this one for years and now the end is truly in sight. The release date for Dungeons of Dredmor is set for April of 2011!

Here’s a lovely trailer that Nicholas put together:

Be sure to view it in high quality so that you can appreciate our awesome resolution jump. Yes, 800×600 just isn’t good enough for the year 2011.

Also: Gaslamp Games has also finally picked up on using that Twitter thing, so check out our twitter page here.

The last … long time, but the last week in particular, has been exhausting as we’ve just pushed out beta 0.90 which is our first beta build with completely revised game mechanics and increased resolution.
Yeah, this is happening for real.

Starfarer: Pew Pew!

2010/12/23

Or: Weapon design & graphics modularity in Starfarer

A game which revolves around combat in space naturally places great importance on the weapons uses in said space. In short: They must look really cool. Here’s a picture to show how I’ve been going about this:

Click to view full size. You see here the process of taking a weapon design from concept art to pixel-art sprite to in-game screenshot. The barrels of the Heavy Autocannon — a nice standard warship cannon — recoil individually upon firing.

And how about some background on what influenced this outlook on displaying weapons?

Read the rest of this entry on the Starfarer blog

Bard’s Apprentice

2010/12/21

Bard’s Apprentice,  a cute little Flash combat rpg, was released earlier this month on Age of Games. It was developed by Cun Sun for whom I drew the overworld terrain tilesets, some UI backgrounds, and some items.

You can play Bard’s Apprentice for free online here.

Some screenshots here! :



Commitment Anxiety in Skill Selection

2010/12/21

In the current revision pass on Dungeons of Dredmor we’ve finally had to make some hard choices about what skills mean to a player’s character. Thus far, all skills have been more or less freely available to select from any point for testing purposes. But if every skill is always available then by the time a player earns a few levels they shall have had the chance to buy a completely new set of skills which would render the importance of their initial choices mostly meaningless. We want every playthrough of Dredmor to be about an experience which is meaningfully different from a playthrough with different starting selections — so far as we are able to make it so.

Which will you choose?

To restate our assumptions: At the start of a game of Dredmor, you must select seven skills to create your character. Yep, just seven. True, some skills are probably more useful than others, for how can ‘mushroom farming’ compare to ‘fire magic’? – Ah, but appearances may be deceptive, and I hope to make mushroom farming a skill to be feared; The fungi from Yuggoth compel me. (But that madness shall come in the crafting skills iteration…)

Why seven skills? I don’t know. Maybe it felt like a good number. It could reflect influence from Dwarf Fortress (whose use of seven dwarves has an obvious folklore connection), except that the foundation of skill selection was implemented before DF was released, if I recall correctly. I’ll have to ask Nicholas about it … and he says: “Oh, I just picked it at random”.

Ah. Lovely.

Let me briefly consider some other games’ approaches:

An old favorite of mine, Ultima Online, had a dynamic advance-through-use system of skills rated from 0 to 100 with a total skill point cap of 700. Once your skills added up to 700, you just shuffled that set number of points around. The downside of UO was that the system promoted rampant macroing; Raph Koster explains this (and more!) in his writing on UO’s use-based system. The capped dynamic skill system fits the theme of an open world and I really like how it is a radical difference compared to the lineage of MUDs that revolve around grossly linear advancement, the influence of which we see today is most MMORPGs (read: WoW). In UO, player power relative to one another was kept within a reasonable range – a new player would start with 1/10th to 1/2th the hitpoints of a maxed out player. Five or six fresh noobs could conceivably fight a veteran character and win (though no Red worth their black pearl would be taken by a pack of noobs).

Dredmor, however, is not an open world sandbox game, nor does relative player power matter because it is a single player game. Roguelikes as a genre tend to be about making a few important choices at the start of a game and then exploring how those choices affect a playthrough which requires relatively little investment compared to an MMO character. What I’m getting at is that I think almost the entire point of starting a roguelike character lies in those choices at creation being a meaningful statement of how you intend to explore the rest of that playthrough – or at least a shot in the dark that will give you a unique experience. This suggests to me that we should be unforgiving about changing skills, as in: you can’t.

I know that modern games like Titan Quest and World of Warcraft are rather forgiving about letting you undo skill selection decisions — in TQ, you may change skill choices for increasing gold cost, in WoW, likewise for picks in the ‘talent’ tree. I see these design choices as a result of wanting to play nice with more casual gamers, alleviating the pain of character optimization mistakes in games that both take more time investment and revolve quite centrally around number crunching. Dredmor certainly has numbers and crunching, but I hope that the spirit of the game comes through: that it’s more about exploring interesting choices within given systems than linearly optimizing DPS numbers and threat/tank mechanics.

A rather poorly organized skill design spreadsheet. We try not to pay too much attention to it.

(Can you defeat Lord Dredmor with just crafting skills? A shiny goat figurine to whoever does it first!)

Funny thing, Diablo 2 was not so forgiving with skill re-allocation while I’m certain that Diablo 3 will have some mechanic for it.

To ramble on tangentially, an interesting point from Diablo 1 is that you read spellbooks to learn spells rather than gaining them via experience; And this is more properly Roguelike, if I recall correctly. Remnants of this book-advancement exist in having to purchase spells in something like WoW, though that’s entirely functioning as a money-sink rather than being loot-based. It’s an interesting thought, with books as spell advancement: This means that a mage’s spell power is attached to item acquisition rather than experience advancement. This gives a Wizard goodies to find in all the piles of loot which are generally armour and weapons and item-based character focused.

At that, learning from books was how spells were originally acquired in Dredmor. And there was this awful system where you had to roll to see if you succeeded learning the spell, otherwise the book crumbles – it seems to arbitrary and punitive, so I argued to have it cut. (Some people just like pain, of course, and maybe that’s why they play Roguelikes.)

… What to do with all these old spellbook graphics …

Right, so to bring this back on topic: We’re having player’s choose seven Uberskills at game start. An Uberskill is a skill category (eg. Swordplay, Fire Magic, Fungus Mastery, Veganism) which has between three and eight sub-skills (Unterskills, if you like) which you may advance linearly with skill points earned through leveling. Swordplay sub-skills, for instance, grant bonuses to combat — especially when using swords — then starts giving special attacks that have status effects and more damage or area effects to give some tactical depth to work with.

I’m not really sure what we’re going to do when someone maxes out the paths on all the skills they’ve chosen. We’ll think of something cool.

[Written for the Gaslamp Games blog]