The pros, cons, and techniques of procedural asset creation (renamed)

Laa-Yosh

I can has custom title?
Legend
Supporter
More like, they can't afford to build content that won't be seen. Just think about the math.

Free roaming means the player may complete the game with only 50% of the levels seen. In this case half your art budget was wasted because the customer hasn't seen it - so if you knew what 50% that is, you could've spent two times as much work on that 50% and make it two times better looking (and throw out the rest).

So, make only 50% of the levels, but make sure everyone sees every part of it. Hence the corridors.
 
Free roaming means the player may complete the game with only 50% of the levels seen. In this case half your art budget was wasted because the customer hasn't seen it - so if you knew what 50% that is, you could've spent two times as much work on that 50% and make it two times better looking (and throw out the rest).
Or you can use good enough (guided) procedural generators for the free roaming stuff and leave the heavy editing for the main storyline and accept the compromise of uneven quality for offering a grander scope.
 
Or you can use good enough (guided) procedural generators for the free roaming stuff and leave the heavy editing for the main storyline and accept the compromise of uneven quality for offering a grander scope.

Procedural stuff will never, ever look as good as what artists like the Naughty Dog guys can put together for the Uncharted games.

Most of the larger open world games already use as many shortcuts and tricks as possible but that's still not procedurally generated. Like brush based tools to shape terrain, automatic tiled texture variation and height-based layers for grass/rock/snow and brush based vegetation placement tools and so on. This is all quite streamlined after decades of research and development and I think its limitations are also evident. It's a lot better looking than procedurally generated stuff but it's nowhere as nice as anything in UC1-2-3 or similar games.

Rage and BF3 and Lionhead's Megameshes represent new, different approaches to build outdoor environments but all of them rely on more artist time to get better results.


Indoors is always going to be limited by artist time, it's not possible to procedurally generate and decorate stuff like that.
 
Yes, generally they are but they don't have to be when designed right. GTA and Prototype are pretty much just a bunch of block-buildings with some cars and trees.

Would you want COD's gameplay to take place in GTA4's cities? I don't think the results would be nearly as exciting. They need to manually build COD levels and thus they need to make decisions on where and how they spend their art budget. Also, people apparently like these corridor games...
 
Procedural stuff should stay on the development side IMHO. Give the artists tools to efficiently (procedurally) create forests/trees and such, but the final product should be something that was tweaked and perfected by a professional.
 
Procedural stuff should stay on the development side IMHO. Give the artists tools to efficiently (procedurally) create forests/trees and such, but the final product should be something that was tweaked and perfected by a professional.
So artists aren't professionals?
 
The dilemma here is that they have a limited budget and having to build 10 times the content means 1/10th of the effort and detail, so a much uglier game.
I don't understand why I have to stress this so many times.

I really couldn't care less if a bad game have the best gfx ever....




Anyway if there is anything to blame here is the mindset that if you arent the best seller you dont worth ******.



The industry will need to start doing small budget projects that while dont cater to everyone it does cater to a segment that like the specifcs of that game, if that mean open world FPS (like Far Cry or Crysis) with Crysis (1) gfx there is a market to that, just make sure your budget/price is appropriated to it.



It is like music, does everybody like death metal? No, only a (relative) few, but there is great death metal bands, they just dont have the studio budgets of Lady Gaga and they dont need it to make great music. The same happens to movies or books or anything else in the media and arts segment.


This is not a contest, as long as the game sell enough to pay costs, make a (decent) living and start the next one is it very sucessful .


More and higher budget |= good game (apply that to everything else in life too).
 
Last edited by a moderator:
So artists aren't professionals?

Yeah that's what I meant. A professional artist. Should be the one creating things. If stuff is just procedurally generated at runtime it will get boring fast.
 
Also it has absolutely nothing to do with AI :)

My little theory is that most games are corridors now because of the mass market not wanting a challenge but an interactive movie instead.

It's mainly to do to the fact if you have a wide open environment most players will get lost and won't know what to do. Which is frustrating for your average player.

Playtesting drives a lot of the 'dumbing down' and ubiquitious signposting' and handholding seen in modern games.
 
Indoors is always going to be limited by artist time, it's not possible to procedurally generate and decorate stuff like that.
Lets take Witcher 2 ... do you really think the kind of quality of interior decorations in the majority of the homes is impossible to procedurally generate?
 
Yeah that's what I meant. A professional artist. Should be the one creating things. If stuff is just procedurally generated at runtime it will get boring fast.

Depends on the game. If Rise of Nations, Civilizations, and Diablos weren't procedurally generated at runtime they'd get boring real fast.
 
Yes, decoration is by the definition of the word an artistic process.
A computer program might throw in some random stuff, even at predefined locations but it's not going to be the same as an artist manually placing items using rules of composition and balance and such, putting a story behind it, etc etc.

Just as you can random generate dungeons in Diablo - but it won't be like a hand crafted labyrinth...
 
Yes, decoration is by the definition of the word an artistic process.
A computer program might throw in some random stuff, even at predefined locations but it's not going to be the same as an artist manually placing items using rules of composition and balance and such, putting a story behind it, etc etc...
Most of that is irrelevant to the background art that makes up the base of any game. Composition is meaningless where the artist doesn't control the camera, as you don't know the viewpoint so can't compose. Story is mostly irrelevant in scenery save a few key points you'd want to draw attention to. It's more important to me for levels not to appear repetitious than for the repetitions to be great works of art. Take something like Uncharted or Resistance, running through houses. We see the same content over and over and it destroys the illusion of these being real people's homes. Dedicate a bit of run-time processing to tweaking colours and furniture models, arranging rooms via a formula for realistic object placement, generating some varied photos and decorations based on some formulas (like photofit for varied people photographs), and these worlds would be a lot more inviting. This would also free the artists to work on the stuff that needs artistic attention like key scenes describing part of the story of with essential gameplay elements (cover etc.) - better games overall without the insane cost of having artists hand-craft every component and every scene.
 
I don't have time right now but your preconceptions about how good these changes would look are quote wrong in practice. We've tried a lot of these approaches and they usually fail to work.
 
Depends on the game. If Rise of Nations, Civilizations, and Diablos weren't procedurally generated at runtime they'd get boring real fast.
Disagree, its fun at the beginning but in the end its annoying. Titan Quest has hand made areas and it isnt more boring than Diablo.
First two examples are correct though.
 
I don't have time right now but your preconceptions about how good these changes would look are quote wrong in practice. We've tried a lot of these approaches and they usually fail to work.
I look forward to that post! ;)

As more meat to my position, the scenery around me that I encounter day to day, home's and streets etc., aren't created and positioned via artists but via various human behaviours or natural forces. A bed is always against a wall, as is the wardrobe, but there's no particular formula for where to put them other than space considerations (unless it's a home owned by a Feng Shui enthusiast) and being able to open the door. The TV is placed either well or badly in the sitting room and then the 2/3 piece suite is positioned around it. Or around the fire place if you're going medieval. Pictures are hung at a reasonable height and spacing. There are often a set of matching mugs and a load of varied mugs. You may have a single light fixture in the roof, or a number of uplights or spots if they are a bit more classy. A castle will have torches. I reckon a pretty convincing typical house or office or castle can be created very well via a mathematical model. Now I know in the world of cinematography everything is crafted including lights and set dressing and the like, but every scene in a film or cutscene is managed. Lots of gameplay operates...tangentially to the art? Certainly a game like Infamous doesn't have every street and every building crafted with complete artistic care. The accumulation of rubbish could be completely automated (according to rules) and I wouldn't see the difference. What was very apparent in that game is the limited number of assets and general same-y-ness, punctuated with key buildings. I can well see a next-gen Infamous using procedural content creation to make a far more natural city without requiring artists to sculpt every piece of it.
 
Disagree, its fun at the beginning but in the end its annoying. Titan Quest has hand made areas and it isnt more boring than Diablo.
First two examples are correct though.
I played CON time and again from the beginning because the random dungeons made for some varied encouters. I found games with fixed dungeons like BGDA2 get very boring and the same content becomes tiresome. Good random level creation is always a plus IMO, but as ever it all comes down to implementation. Random fields of nothing of interest aren't going to have any value.
 
I would like procedural generation on the development side, if that makes sense. Like, the artist can use the algorithm to get the basic layout and then tweak it from there. If it all happens in runtime it will get repetitive.
 
I would like procedural generation on the development side, if that makes sense. Like, the artist can use the algorithm to get the basic layout and then tweak it from there. If it all happens in runtime it will get repetitive.


Dont know if it is the case but often repair and tweak takes more time than doing from scratch in so many things....
 
Back
Top