Spring RTS Engine

AF wrote:possibliltys

Just like pre release B&W?

I youtube'd it a while back, and all I saw was a series of fairly unexciting looking mini games strung together.

<rant mode ON>

I hate to put it this way, but Spore is the most completely over-hyped thing since, oh... Daikatana.

The thing about Spore is that I see it more as Will Wright's philosophical statement about the state of the art, more than as a serious game in its own right. However, I think that Mr. Wright is essentially sidestepping a large number of issues with the design.

Ok, you can evolve from an amoeba to a spacecraft-flying race of infinite variety. Okie doke. That's impressive programming. But there are a huge number of problems that result, which I suspect that Wright will also artfully fail to address, by remaining cute enough that nobody calls him on it, aside from grumpy curmudgeons like myself:

1. A game, by definition, has victory conditions. There really aren't any in Spore- you merely pass through stages of vaguely interesting mini-games with no real resolution. Evolution isn't victory.

Even most relatively open-ended games (Escape Velocity being a classic in the genre) have a theoretical goal that could establish an "end"- you can conquer the known universe. So far as I know, nobody has actually *done* that, but meh, it's at least theoretically possible. In Spore, it just appears that you go from level to level to level, and eventually progress to the point where your POV is entirely global and abstract- Will's going from the incredibly finite to the nearly infinite. Again, that's clever... but what's the point?

I mean, why not just make SimUniverse... oh, wait, that's this guy's goal... and he always makes me happy because whenever I think I'm too egotistical to survive, I use him to put myself in context

When I think of game designers of Will's generation, and compare them... well, you've got Sid Meier, who is famous for creating games with fluid gameplay, accessible detail, and multiple winning conditions, Hironobu Sakaguchi, who really defined level-grinding as a pleasurable pastime, Timothy Cain, who gave us Fallout's "any path to victory is acceptable, but there is only one true victory", and good ol' John "Daikatana" Romero, who before he succumbed to his own hubris, gave us Quake (yeah, Carmack made the engine, but Romero made the game), with its tight focus on area interaction and AI.

I guess the thing is, that when I think of what a game is, I see it in terms like these, not just a random collection of vague thinglets that vaguely thinglet about

2. Spore is about fighting against the costs and complexity of content creation, according to Wright.

I've read many, many rants about this topic, and quite honestly I totally agree with most of them. It's getting too darn hard to make a good video game. To make anything bigger than a smallish demo, you need to have a medium-sized team, willing to work for months towards a goal, plus at least one or two very decent designer/coders to bring it all together. I know how hard that is in freeware projects, and I'd imagine it's at least as hard to pull of in for-profit gigs. Sure, you get more work out've people who're being paid, but directing the herd of cats is still probably the biggest issue.

However, the way I see it... we're just about to hit the wall on how much further graphics can go before it's mainly a matter of artistic choice and technical skill, rather than a necessary selling point.

Seriously, people- in another 10 years or so, we're going to have games that are darn-near photoreal, even if they're still using tricks to hide how much is not actually being simulated... and in between then and now... well, we've all seen shots of Krisis, right?

How much further do we have to go, before it's pretty darn unimportant detail, and gamers, even not-so-jaded ones like me, start saying, "hey, the graphics really aren't that important". That's not the case right now- the dividing line between something that looks crude and unprofessional and something that's first-rate are still pretty clear, and first-rate is defined mainly by professionally-developed titles that can spend huge sums getting artists in China to produce 1000 different trees or whatever... but with the huge and ever-expanding toolsets for making art affordably getting ever better (trust me, guys, Wings may not be Rhino or Maya or 3DS, but it beats the hell out've Infini-D, which is what I started with) I suspect that el cheapo projects will eventually catch up.

For me, Freelancer was one of the big points, and I think that when we look back someday, it and games of its era will be seen as the watershed moment, when graphics started towards a null point where it just didn't matter very much any more. Not Oblivion. Freelancer. Which, with some modding and gentle love, can still look fairly good. Oh, it's not X3, in terms of sheer detail. But it doesn't look like crap, either. It's kind've like the era where we (very briefly) went from 8-bit tileset games (Diablo I, Fallout) to 16-bit tileset games (Diablo II). Sure, 16-bit tileset games could be more beautiful, but meh... it wasn't that massive of an improvement. Oblivion impressed me more by the sheer about of human labor expended than it did by any massive jump forward in beauty- Doom 3 looks almost as good, if you turn it all the way up, it just doesn't have the million man-hours invested in the artwork.

I mean, there's just so far you can push things, before it's kind've pointless excess- you will never see a Freelancer-style game where the interiors and bulkheads and mechanical sections are all created, down to the last bit of Ion Hyperdrive, Mk. 3, because well... for the time-cost of having somebody create that, we could just build 5 more really cool-looking spacecraft hulls, and have our procedural FX coder and physics wizard hide the fact that our ships don't completely accurately disintegrate.

There's an upper limit, frankly. It's just about been reached, for RTS games, imo- with stuff like the latest Medieval Total War and Warhammer games out, we're seeing armies of somewhat-individual looking dudes with LODs and nice use of displacement maps... if it gets much more detailed, it's almost entirely a waste of people's processing power, and could be better-spent on real innovation, like making each member of a unit a truely independent agent, etc., etc.

The only real limiting factor keeping us away from Nerdvana, where we're back to practical garage-team games and guys like me can actually build nice projects without having to recruit and herd cats for months or years... is that the toolsets are missing.

Even that situation is improving somewhat- we have stuff like OGRE and other engines for physics, sound, netcode, etc., making things a lot more practical. I'd love to have enough coders (and free time) to kibbitz, cajole and generally beg my way towards having multiple game engines that are as powerful and flexible as Spring, but for different genres of game. Unfortunately, I can't do that, and that's just tough cookies. Maybe I should apply for a grant, or build a game that was actually (gasp) profitable (meh, Silent Dark was "profitable", but barely) so that I can afford to take a year or two herding even more cats...

</rant mode>

ahaha let me write 1200 words about the flaws with a game ive never played

<shrugs> I haven't ranted in awhile, guess it was building up

I would have thought you were going to say Fable...

And I disagree, I still want to play Spore. We're just going to have to wait and see.

You're misunderstanding me. Like all of Mr. Wright's games, I am sure that lots of people will enjoy it. Don't get me wrong- while I personally don't get into his game designs, I'd be an idiot if I didn't see that they have vast audience appeal, and sell heartily.

It's the supposedly-profound philosophical statements that he's weighted the game down with that get my back up:

1. "It's a completely new approach to game design". No, it isn't! Mr. Smart, who I linked to for the benefit of those who aren't familiar with his work, is probably the foremost creative genius behind, "yes, let's really try to simulate it ALL" in game design. It's the mad-programmer's dream! I dunno how many game-designer wanna-bes I've dealt with over the years, who when you get them talking about what they wanna do, get into these crazy, insane, just plain silly idea sessions, about building gigantic universes of Fun.

Guess what? Rage Racer, which is one of my favorite games ever, had all of 13 cars in it. It was still one heck of a nice, tight, fun game.

Twisted Metal 2, which is on my short list of, "best games ever designed, period", had ... hmm... 13 cars.

Not 50 billion, of which maybe 10 are optimaxed and nobody in their right mind will care to find them (we'll leave that up to the utterly-crazy dorkwads who write uber-walkthroughs as a substitute for meeting girls). 13!

And, to further my point... there are plenty of games, from the Leisure Suit Larry (and pr0n clones) series to many adventure and arcade games, that feature multiple twists on a theme (aka, mini-games or puzzles) to keep gamers happy as they trudge through them. However! They have a story to tell that goes beyond, "yup, evolution is kewl". And they have definite victory conditions.

2. "This game goes around the limitations of Content by making Content easy make and trade with others". Um... yeah... look, we're in the DSL age, here. 1KB procedural stuff that still looks weird, cartoony and lumpy, if freakishly well-animated, or 3000KB full-mesh models with mipmapped, gorgeous detail, animations hand-tailored for the task at hand, and LODs that give it performance benefits?

The real problem with Content is that nobody has (yet) designed an underlying rendering / physics / etc. engine that is stable, mature, and completely plug n' play, for people who want to add new capabilities. If we ever get there, and enough people sign on to the idea that it's worth keeping bleeding-edge development going on it... then, with good tools to build and work with content, the sky is really the limit.

I suspect that, some day in the not-very-distant future, the game coders of this world will realize that if they just build modules that can adhere to a common, flexible backbone (how many iterations get called per second, how do we control what gets sent to what module when, etc.) then we'd be 100% closer to building whole game designs by writing a fair chunk of high-level code to actually do stuff that's important to the game design (like, I dunno, something like BOS for speed-critical bits, and LUA for slower but more complicated things like economic simulation and GUI feedback) and plugging in everything else. For example, I'd make a game about fighting in giant submarines with lots of players:

Core

Graphics (basic drawing of meshes, specific modules that can handle specialized cases like voxelmaps, plug-in shader libraries for water FX, particle events, etc., that can be programmed within parameters through the LUA high-end code)

Sound (this is getting nearly optimaxed at this point in history, anyhow- it won't be long before surround-sound using the fastest possible codecs and compression is just about all there is, and it becomes a niche science, imo)

Physics (multiple plug-ins to handle collision events, underwater movement, and specific widgets that do special stuff, like simulate the dynamics of whirlpools and other things)

UI (really, this is one of the hardest things to separate from the core, and Trepan's work is really quite amazing and exciting stuff, because it's just a few steps away from being non-dependent on the rest of the modules)

Netcode (server/client relationships, sync'd or interpolated, security)

Gameplay modules (on-load event handling, "realtime" events, AI, database creation and management for creating large-scale data collections and making them meaningful)



Whereas, a single-player driving sim might skip the netcode module entirely, add entirely new shader libraries and drop many others, and have an entirely different approach to the gameplay module, as well as a faster cycle through the core code.

The thing is, Spring's about halfway there. We're stuck with a huge number of things in the core loop, though, that are tied into so many, many other things that nobody's had the time or mental energy to untie it all and create a straightforward core. I'm very interested in seeing if JC actually finishes that part of Command Engine, because if he does... well, that'll make a lot of things possible, honestly. I have no idea what will actually happen, though.

I buy games mostly on how much you can mess with them. I bought City of Heros simply because it was incredibly customizable. I'm lucky it was fun, too.

That's mainly why I want Spore. I can make stuff.

I understand that completely. When I was a kid, I used to make stuff with Adventure Construction Set, because it was easy enough to program and work with. If Spore has ways to build anything that isn't just procedural- if, for example, you can build a scripted story with a definite outcome... that'd be a much stronger design experience. I just don't see a lot of evidence that that's going to happen, though.

I hope that does happen, because I get the feeling that he'll alienate a lot of people if he makes something so focused on creating so hard to mod.

I just want spore to finally build my galatic empire. My only complaint is..it seems really un exciting to play, not enough 'BOOM'.

Argh: while I agree with you for the most part... you're not really ranting about spore, nor are you really ranting about game design... You're ranting about the state of the gaming market.

I recently read an interesting article you might like. Basically I'd say the case you're describing is the current problem in gaming, more technical crap being thrown around all the time and less actual game design. That being said, Will Write is working down one of the possible fix paths. In the days of old when we wanted content for a game, models and what not, we'd have to work them up in a modeling suite, we'd have to manually build animations, we'd have to texture and rig them ourselfs. Spore does this all procedurally... Think of the man hours saved on art teams doing that... That in itself is admirable, weather or not the final game is really going to be a universally fun experience is sort of second stone... I would agree however, I don't think spore will be the savior of gaming. It's probably too late at this point in time.

What we face right now is somewhat of a dilemma, how do we make gaming smarter without exerting more development effort with each new generation? The current development firms in my opinion are generally doing a poor job of keeping gaming fresh, keeping gaming new, and keeping gaming profitable. There are some cute examples to go off of, but nothing other then spore really stands out right now, and I really don't think spore alone is much of a contributor, because it's all procedural it's going to get stuck in the bubble of it's own engine, it won't be modular enough to be used en-mass elsewhere.

There's some cool possibilities for the future to do with procedural code work... especially procedural animation is quite exciting, but like you say, it doesn't really solve the problem of the tools being too weak and taking too long to produce great content.

Yeah, I read that awhile ago. Personally, I prefer Greg Costikyan's articles about all of this stuff- I think that games are actually eternal in appeal and nature (see, Chess) but that the video game industry, on the other hand, is quite trapped. See, also, the way that chips and video card hardware for anything but a basic DX10 system are still fairly expensive- a sign that the manufacturers are still waiting to recoup tooling costs. I still haven't upgraded my now 2-year-old PC in any serious way, aside from upgrading a harddrive, since I bought my 7800GT, because the price of switching to a 64-bit rig is still too darn high for me to justify it. I'll wait until I can do it for less than $500, but right now, a system that's a real improvement on what I already have is about $850... not really worth it to me.

Spore does this all procedurally...

Not really. Each model segment must still get defined with some hand-tweaked data that the procedure then reads and uses. And I doubt very much if it's just, "meh, this part is a flipper", it's probably more like:

IF (creatureState = water && bodypart == "flipperLike001" && positionPart-->(positionFront = 1)
{
flipperLike001 (partValue, partDuty) = (1, 3)
}

Where partDuty = some ginormous chunk of procedural code used to handle the weighted, boned and constrained IKs. <shrugs> maybe there's a more elegant way around all this that doesn't routinely bork... but I kind've doubt it, tbh. It's basically turning giant man-hours of artist-work into giant man-hours of coder work... and last time I checked, artists are cheaper. Meh, you can get away with more outright copy/paste with code, I guess, but I suspect that from a purely economic standpoint, it's not a lot cheaper to make a game like this. However, we'll just have to see.

Is there going to be some story mode?? Kind of like the feel of Homeworld..

No, it's definitely not cheaper to make a game like that... However, once a game like that is in place all future games like that that are allowed to borrow from it are able to save alot of time on art work...

Especially procedural animation and procedural texturing. We're coming to a point where it's feasible to make one model, and then all the animations for that model, and nearly all the texturing can be generated by the engine during play with very simplistic developer input in retrospective. This is an especially cool concept for RPG titles which hope to have infinite weapon sets. Using procedural texturing, procedural animation, and procedural particle generation we can take 400-500 weapon models and have a billion different iterations of them for the game.