Why Ultima sucked (was Re: Tabula Rasa back from the dead)

[schild]

Yes, because that's what I was saying. Don't troll.

[Reg]

Geeze Schild it's not like I've based my whole Internet persona on hating anything popular and loving obscure stuff that nobody has ever heard of.

[Morat20]

One reason I'm following Spore is that Wright has identified the critical problem in modern game design -- that graphics and art are eating up an ever increasing percentage of the game budget and time -- and has tried to make a game that gets around that problem, to an extent.

Yeah, that really worked out well for him, seeing as development started seven years ago and the projected release isn't until early -> mid 2008. :P

Hence the "to an extent". :) He's running into the same problem from a different direction.

[Murgos]

Yea, instead of bogging down my computer with shitty textures (par for the course with previous Will Wright titles), we're gonna bog down our computers doing math. Meh. Critical problem my ass.

Bog our computers down by doing math?  Oy Vey.   

[Morfiend]

Yea, instead of bogging down my computer with shitty textures (par for the course with previous Will Wright titles), we're gonna bog down our computers doing math. Meh. Critical problem my ass.

Bog our computers down by doing math?  Oy Vey.   

WINNAH

[schild]

Guys, procedural shit is pretty intensive. Remember how long it took that 300kbps shooter to load the first time? Just saying. I'm not saying it'll be intolerable, I'm just saying, they didn't realize any sort of critical problem and set out to "fix it."

[Morat20]

Guys, procedural shit is pretty intensive. Remember how long it took that 300kbps shooter to load the first time? Just saying. I'm not saying it'll be intolerable, I'm just saying, they didn't realize any sort of critical problem and set out to "fix it."

I didn't say -- or at least didn't mean to say -- that he fixed it. I said he recognized it as a problem, and Spore was one attempt to find a way around it. He's been pretty open in interviews about how insanely huge a part of his games budget art assets have become, and how each new game seems to find art assets becoming larger and more expensive a part of the game.

He coupled that with the fact that the fans of his games seemed to like nothing better than to make their own crap for the Sims and share it around, and came up with a notion wherein he could bypass the bulk of the "Art generation" and make "creating your own shit to share with people" an integral part of the game.

Doing so was a huge technical challenge (I make no claims as to whether he can actually make it work) since no one has really tried this on this great a scale.

His 'solution' such as it is doesn't really scale to anyone else, or probably to any other game (save sequals to Spore, assuming there are any) -- other than it's generally worth it to include tools to build art and game assets because some people like to mod, and a lot more like to download them and play them.

I only mentioned him as an example of someone noting that games these days spend a hell of a lot more time on "teh pretty" then they used to, and that the more time and money you devote to art assets, the less you have to gameplay. Will Wright's trying to turn "creating art assets" into the gameplay -- but that's a pretty limited solution.

[Falconeer]

Yea, instead of bogging down my computer with shitty textures (par for the course with previous Will Wright titles), we're gonna bog down our computers doing math. Meh. Critical problem my ass.

Bog our computers down by doing math?  Oy Vey.   

You know.... 1954 was a great year for sci-fi.
(Yes, Murgos for winnah).

[Murgos]

Guys, procedural shit is pretty intensive. Remember how long it took that 300kbps shooter to load the first time? Just saying. I'm not saying it'll be intolerable, I'm just saying, they didn't realize any sort of critical problem and set out to "fix it."

Yes, it's easier to just look up what color and shape to make a set of pixels from a set list (i.e. a 'texture map' or model) than to compute what color it should be based on dynamic inputs (procedurally).  I was just goofing on your 'math is hard for computers' line.

Seriously though, loading textures into and out of memory is what takes a great deal of the work of a computer in a modern game and does a lot to restrict how detailed a scene can be.  A single 512x512 resolution texture is 1,048,576 bytes of memory in 32bit color.  Procedurally generated graphics can be [but not necessarily will be] much more efficient.  I'll even go so far as to say that most video lag you see today is due to textures being swapped in and out of memory.

[schild]

http://www.theprodukkt.com/kkrieger

Now granted, my computer is dogshit. But loading that took more time than loading up Doom 3 because it was procedurally texturing everything. Which is to say, I was merely saying that Will Wright's "Fix" isn't much of a Fix at all, it's just something to make his shit smell like roses.

[Morat20]

http://www.theprodukkt.com/kkrieger

Now granted, my computer is dogshit. But loading that took more time than loading up Doom 3 because it was procedurally texturing everything. Which is to say, I was merely saying that Will Wright's "Fix" isn't much of a Fix at all, it's just something to make his shit smell like roses.

You might -- and this is just a total stab in the dark -- wait until you can actually play it, which admittedly might be 2024. :)

More seriously -- as I noted, Wright went with procedural generation for two reasons. First, art assets were taking up a huge chunk of his budget. Second, people seemed to like making shit FOR use in his games almost (or sometimes more) than actually playing his games. Procedural generation killed two birds with one stone, allowing him to to make a game that was, pretty much, people making up the art assets for the game.

On a tangent, you should only have to crunch the numbers on a creature once and cache it after that. Once the software has worked out what it looks like, how it moves, and how to properly draw and animate it, you don't have to crunch all that shit again -- just display as normal.

[Murgos]

http://www.theprodukkt.com/kkrieger

Now granted, my computer is dogshit. But loading that took more time than loading up Doom 3 because it was procedurally texturing everything. Which is to say, I was merely saying that Will Wright's "Fix" isn't much of a Fix at all, it's just something to make his shit smell like roses.

You do realize you aren't actually saying anything, right?  Pointing at one dudes tech demo of procedural graphics and going "La, it's all shite." is a little sweeping, eh?

Procedural graphics are hard and to do them right is difficult.  Usually, they are only used when EXTREME optimizations are needed.  That particular game takes so long to load because he creates all the graphics from formula to keep the size of it down to 300kb.  You do realize that the whole point of that tech demo is that it is a 'look what I can do if I push this concept to it's utter limit.' sort of thing, yes?

[Yegolev]

Will Wright should cancel Spore and concentrate on making a game with lots of effeminate big eyed characters in it.

That would be The Sims 2.

Will Wright's trying to turn "creating art assets" into the gameplay -- but that's a pretty limited solution.

Also from The Sims 2.  I think it could work but I am still in a serious wait-and-see mode with Spore.

In the absence of pewpewnicorns, I would like to point out that I have easy-clicky access to Space Quest and X-COM via GameTap, but I still don't play them.  Well, I started to play each of them but have not made much progress.  Space Quest, man, I forgot how angrifying it is to type in wrong word after wrong word.

[Trippy]

Guys, procedural shit is pretty intensive. Remember how long it took that 300kbps shooter to load the first time? Just saying. I'm not saying it'll be intolerable, I'm just saying, they didn't realize any sort of critical problem and set out to "fix it."

You are missing the point.

First, the textures and other procedurally-generated stuff can be cached after they've been "rendered".

Second the point of generating textures, surfaces, and other objects procedurally in a large game is that you don't have to have armies of people creating those art assets for you by hand. When people complain about game development getting more and more expensive it's the cost of the art asset production work that they are mostly complaining about, and also why many game companies are starting to ship that work overseas to China.

[Morat20]

Second the point of generating textures, surfaces, and other objects procedurally in a large game is that you don't have to have armies of people creating those art assets for you by hand. When people complain about game development getting more and more expensive it's the cost of the art asset production work that they are mostly complaining about, and also why many game companies are starting to ship that work overseas to China.

Yep. You can buy and modify a database, an engine, various middleware, tons of solutions in a can that have the bulk of the work done for you. You cannot really buy "Art assets". That shit has to be new. The best you can buy is tools that fit whatever engine you've bought, and even then the tools have to be modified to work with whatever engine changes you've made. And then all that art has to be done from scratch each and every time.

Will Wright's lucky -- the people that play his games view 'creating the art' as a huge portion of the game, so he makes a game where creating the art assets IS the game. It's not something most games can do. According to a few interviews, he got the idea for Spore -- and using procedural programming -- because his art costs ballooned in The Sims 2 and even then, users quickly exhausted it and moved to user-created content.

[tmp]

According to a few interviews, he got the idea for Spore -- and using procedural programming -- because his art costs ballooned in The Sims 2 and even then, users quickly exhausted it and moved to user-created content.

A large problem with procedurally generated content (aside from intitial research/implementation time) is imo, it can greatly hinder any sort of effective QA. When you have an artist create walk cycle for 5-legged dog, then it's not too difficult to watch it, decide it looks like shit and send it back for rework. With procedural content you need the QA guy to come up with idea of 5-legged dog to test in the first place, then after it's found to look like shit the guy that brewed spaghetti code responsible for this disaster not only has to figure out which part of his mess is making that dog walk like shit, but also to come up with fix to underlaying theory that _may_ fix the issue ... and then finally actually fix the code so it conforms with this upgraded theory without introducing series of bugs in 3-winged fish and undersized gorillas with two heads. And all this vs tweaking few animation keyframes.

[Typhon]

But that would meant that you'd have to hire QA people that were creative... and you'd see your QA costs balloon... so mabye a game developer would think, "hmm, if I use evolutionary coding styles, I could replace my QA staff with code and servers... servers are much cheaper then people!... and I can spin this with marketing!  GENIUS /HUG ME!"

Nah, no ones is that crazy.

[Murgos]

A large problem with procedurally generated content (aside from intitial research/implementation time) is imo, it can greatly hinder any sort of effective QA.

Um, no?  The power of procedural animation is that it is easy to tweak it.  You change the constraints on the range of motion allowable on the joint, compile, run, change range of motion, compile, run, change length of leg, compile, run, etc... etc... etc...

The same cycle involving artists, rendering and all that other crap has to be much more time consuming and intensive.  The initial investment is building a mathematical model of a 5 legged dog.  Once you have it tweaking it is the easy part.

[Sky]

You say it like there's a problem with a 5-legged dog walking like shit. Maybe that's a clue that 5-legged dogs walk like shit. It is an evolution game, after all. Not every being that evolves is a winnar, the vast majority are dead-ends.

This is one reason I am actually wary of Spore. He's got such a great design concept, I hope he doesn't ruin it by trying to put too much game into it for the plebes. I hope he doesn't water down all the scientific background that inspired it, but it already looks like it's heading that way. And unfortunately it still won't sway the people who need directed gameplay.

If it's a science-heavy massive sandbox...I'll be happy.

[schild]

If it's a science-heavy massive sandbox...I'll be happy.

Unfortunately, his publishers won't. The Sims is the best selling PC franchise of all time. Though WoW will eventually overtake it...

With the money and years that have gone into Spore, I don't think anyone at the developer or publisher wants a niche title.

[Sky]

Thus my concerns.

[tmp]

Um, no?  The power of procedural animation is that it is easy to tweak it.  You change the constraints on the range of motion allowable on the joint, compile, run, change range of motion, compile, run, change length of leg, compile, run, etc... etc... etc...

Precisely: change, compile, run, change, compile, run, change, compile, run, *rinse, repeat* until it starts to resemble something reasonable. It's easy to *tweak the numbers* but tweaking numbers is very small part of getting good looking end results. What you describe is actually typical process your "old way" 3d animation guy will go through when they create IK rig to _help_ them animate that walk cycle by hand. Note: help, because all the tweaking and often complicated setups at the end of day still require manual keyframing, cleaning up and corrections... simply because the end result of what essentially is 'procedural animation' generated by automated solution just doesn't quite cut it. IK can and does go weird, especially if you are trying to use some kind of universal rig to handle range variety of motions. The problem is while with traditional approach you have someone who will see it and manually enforce changes for that particular instance so it "looks right", with 100% automated solution you have to get the procedural handler that's so universal and robust it'll get things look right entirely on its own. And that's something that's much harder... to the point where there's only one company that I know of (NaturalMotion) that advertises this kind of middleware/solutions. Even though these automated IK helpers have been around for years, and they could save movie industry tons of time and money long before games started to get anywhere near such budgets... so it's not like inventive to create these is anything new.

The same cycle involving artists, rendering and all that other crap has to be much more time consuming and intensive.  The initial investment is building a mathematical model of a 5 legged dog.  Once you have it tweaking it is the easy part.

If it was indeed the easy part, the need for hand-created animation would cease some 5-6 years ago. Because mathematical models of human joint range, scale etc were created aplenty and are readily available. In other worlds, if this is all that's needed then the necessary tools have been out there for long time. And we should've been playing games with 100% generated procedural characters for years now.

Instead games focus on stitching together results of mocap sessions, and procedural animation tends to be limited to ragdolls (animation of these rare situations where internal joint constraints are mostly removed) ... are all game makers just idiots who don't know how much time and money they could save?

[tmp]

You say it like there's a problem with a 5-legged dog walking like shit. Maybe that's a clue that 5-legged dogs walk like shit. It is an evolution game, after all. Not every being that evolves is a winnar, the vast majority are dead-ends.

I wasn't very clear in the wording there, sorry. When I said 'looks like shit' I meant purely technical aspects -- body parts sticking through one another or mesh warping odly because the shape doesn't happen to cope very well with angles skeleton is put through. Stuff that simply *looks* bad. And you only need to read the numerous "zOMG my sword/cape sticks into my horse when I'm riding around" whines to see this sort of things does matter to number of people.

[Morat20]

Instead games focus on stitching together results of mocap sessions, and procedural animation tends to be limited to ragdolls (animation of these rare situations where internal joint constraints are mostly removed) ... are all game makers just idiots who don't know how much time and money they could save?

I suspect at least part of that is mocap and standard animation techniques are considerably faster, and allow the generation of more art assets more quickly. Frankly, if Spore was just a artist's tool for creating creatures and creating animations for them -- it wouldn't be worth the time or money. They've spent millions of dollars and years of work, and STILL someone has to go in there and put the bloody creature together.

It's faster for a professional to do the animation the standard way, and move onto the next creature. Spore -- as a tool for animation and art asset creation -- isn't nearly as good as most of the standard things artists use.

But Will Wright isn't trying to generate art assets for a game -- if so, he'd have tens of thousands of alien creatures already built and populating some game world for a fraction of the cost -- he's trying to make it the game.

So once more -- procedural generation is a good solution for Will Wright for the problems of bloated art asset costs, because what his players really want is to make the damn things themselves in a sandbox. In fact, for him, I suspect a lot of Spore's design work will be reused for later games to allow more user customization.

It won't do shit for, say, the guys making an MMORPG. It's cheaper and faster for them to create Elven Race #505 the normal way. That still doesn't change the fact that, in the end, the costs of creating Elven Race #505 PLUS Elven Land #505 and Elven High-Level Zone #505 keep spiraling upwards.

[HaemishM]

The JRPG genre means more than heroes on one side, monsters on the other, and levels. So much more. Western RPGs? Not so much.

Yes it has effeminate girly men, needless kekekeke, oversized swords and axes, impossibly large eyes, and upskirt shots.

[schild]

The JRPG genre means more than heroes on one side, monsters on the other, and levels. So much more. Western RPGs? Not so much.

Yes it has effeminate girly men, needless kekekeke, oversized swords and axes, impossibly large eyes, and upskirt shots.

Superb argument, good suh.

[Murgos]

The same cycle involving artists, rendering and all that other crap has to be much more time consuming and intensive.  The initial investment is building a mathematical model of a 5 legged dog.  Once you have it tweaking it is the easy part.

If it was indeed the easy part, the need for hand-created animation would cease some 5-6 years ago. Because mathematical models of human joint range, scale etc were created aplenty and are readily available. In other worlds, if this is all that's needed then the necessary tools have been out there for long time. And we should've been playing games with 100% generated procedural characters for years now.

Nah, it's much faster and cheaper to just stick a couple of guys in a body suit with some ping pong balls on it and make them jump around a bit than it is to accurately describe every realistic maneuver a person can make in a sword fight in 3-D calculus for every moment of time.  It's also much quicker to paint something that looks like a single chain mail shirt and give it a couple of predetermined flex points than it is to describe in math the set of all chain mail shirts and their appropriate behaviors.

I haven't done crap loads of modeling and texture work but I have done a few things here and there for game mods in 3d studio max and photoshop and the level of effort for creating a wire diagram and a texture to wrap around it is pretty minimal, not to mention the training time and level of skill required.  I was able to start making mediocre items within an afternoon with the help of a few tutorials.   About the same amount of time it took me to mock up a sword and stick it into oblivion was necessary just to get a simple colored box spinning around procedurally.

edit:  wiki link:  http://en.wikipedia.org/wiki/Procedural_generation