Procedurally Generated Art – Good or Bad (or Neither)?
A friend who’s really new to the game industry recently let me know that something I’d said struck him as not supportive of video game artists. Having many artist friends, I was more than a little curious to find out what it was I might have said. I certainly didn’t mean to offend anyone.
“What did I say?” I asked.
“You supported procedurally generated art,” he replied.
I had to think about the ins and outs of it for a while before I decided to write about it here, and I am curious about your view on it. I think mine comes with the perspective of having worked on a number of projects that used procedurally generated art and content.
In my view, procedurally generated art doesn’t ask artists to leave the party. Rather, it invites programmers to it. In a nutshell, I support procedurally generated anything since it allows us to make bigger and better games if the code is doing the work for us. In my experience on teams, the procedurally generated content that we worked with never once displaced a team member, be it artist or designer (I’ve worked with procedurally generated content, too). Rather, it freed these individuals to do other critical things on the game and allowed the game to be bigger than it otherwise would have been.
But if there had been no procedurally generated art, wouldn’t there have been more artists working on the team? Well, no. There is a finite amount of artists that any game team can support and still be profitable. For instance, consider Spore. It is well known for its goal of producing procedural and user-generated content. Does this goal displace artists? No. If someone had suggested creating a game with the amount of paid, human-created content that Spore hopes to have, the elevator pitch would have ended right there. The game would not have been sustainable or profitable, even with Will Wright’s name on it.
Beyond what the paid artists can create, everything else is gravy. Procedurally generated gravy. There would be no games with 1000+ levels, each created by an individual artist that never sleeps. Instead, there would be a smaller game that fit within the possibilities allowed by the team. The much greater danger to the practicing video game artist is outsourcing, unless that artist is located in Malaysia, Singapore, India or any of the other outsourcing hot spots.
Procedurally generated levels did not kill off level designers or the need for them. Motion capture didn’t starve animators. Zbrush didn’t put hundreds of artists out of work because now fewer artists can do it faster. Procedurally generated content didn’t kill off game designers. Rather, it allowed the guys who would have created content to focus on game balancing way earlier than they otherwise would have.
This is an industry dictated by dollars. Technology – procedurally generated art included – allows us to do more within our constraints.
Or so I think, anyway. Am I missing the boat?
Applied Game Design 
No.
You’re standing on the boat, waving at a handful of luddites standing on the shore who are squinting at the brightness of the sun towards which you are sailing.
That sounds a bit harsh, but I am the guy who wrote Death of the Level Designer: Procedural Content in Games, so I guess I’m not used to mincing words in this area…
We’re also looking at procedurally generated textures, now, which supposedly could mean the “death” of texture artists, as well.
http://www.psu.com/4D-Graphics–A-reality-only-for-PlayStation-3-Feature–a1063-p0.php
Known as “4D”, procedurally generated textures allow for the use of low-resolution images that are rendered using algorithms. So the coders become the artists, much like you said, creating algorithms that are more and more accurate to the material they are trying to portray.
Unfortunately, this is also rather advanced technology, and really can only run on high-end PCs and Playstation 3s. But hey, any step forward isn’t a step back.
Not everything has to be procedurally generated, either, so texture artists won’t be “not needed”, rather, they’ll be able to work on more simplistic scales for material-based textures and then spend time detailing hi-res textures that really require the personal touch.
When television was introduced, it was going to kill movie theaters. Same thing with home video. CGI was going to kill pen and ink animation. It was also going to replace actors, artists and stuntmen.
Technology provides tools to be*used by* people not *instead of*.
You compare outsourcing and procedurally-generated content as if they’re two different things. Aren’t they essentially the same, except in the case of procedural generation you’re “outsourcing” the task to a computer?
I don’t think procedurally-generated levels kill level design jobs, nor do they even force level designers to do game balance tasks (which actually WOULD be cause for concern — not all level designers are strong at systems design, they’re different skills). But more to the point, the level designers are still on the team… they’ve just essentially been promoted to Leads, with the computer algorithms as their reporting team members!
Back to art… I mean, who makes procedurally-generated art? Is it just the programmers, or is there an artist at some point saying whether the procedural generation looks good, offering ideas for how to make the procedures better in ways that only an artist would know? Are we to believe that the Spore team has absolutely no game designers or artists working for it, that it’s all programmers? (If so, one wonders what Will Wright is doing on the team. Surely he’s not a tech lead?)
As Andrew said, you’re on the boat. And on a nice seat.
It’s like Isaac Asimov wrote on a introduction for a short-stories collection I have, the robots shall be created to perform the non-creative tasks. The artists would still create art, or in this case define the foundation in which the algorithms will build up the assets.
To answer the heading’s question: Procedural art is good.
I’m a veteran game developer (now turning to academia now) and I can assure you Ian, you said nothing offensive or threatening to professional game artists.
No game artist I know is even remotely worried about being displaced by procedural artwork. Why? for the same reason the spirograph-style visualizations in iTunes, while cool, are never going to put music video producers out of work.
Why not? Game players want artwork that resonates with their real-world experience. The abstract beauty of procedural art, however beautiful at first glance, is boring to our audience. They want a world they believe in. That world can only come from a fellow human’s artistic interpretation of the world we share.
If a programmer spends enough time to handcraft a world of “procedural” textures/animations/etc, then i call them they’re an artist that codes. I predict this person will add intolerable risk and cost to any game (except Spore-like games – see below), when , compared to a typical team of texture painting artists.
The need for humanity in our game art is present even when the worlds being portrayed are highly unrealistic. What made Star Wars so great compared to previous scifi? The sterile perfection of 2001 was trumped by a bustling bar full of flesh-and-blood aliens in a barely disguised honky-tonk western bar. Everything compelling in the movie revolved around the same dirty humanity that all good stories are based on. That humanness can never be encapsulated in an algorithm that’s more efficient than hand-building art.
Why is humanity so crucial to success? Because there’s no pattern to hand-built art that’s any less complex than Culture. By Culture I mean the whole mind-boggling mountain of raw data of minute variation that comprises the aesthetic world around us. For example, consider the art on t-shirts. Would hiring a programmer to attempt to build a procedural generator of unique yet convincing art for t-shirts be more successful and cost effective than hiring artists to make t-shirt textures?
On the other hand, procedural art is an important tool that can help artists (including artists who code) create more art faster, as Ian originally said. Every artist I know wants better tools, even though it theoretically means less billable hours, because they know it means they create more content at the same price – makes them look good.
There are a few exceptions. For example, we humans have traditionally valued the complex results from simple algorithms that is so often found in Nature. Woodgrain, ferns, fractal mountains, SpeedTree – we have a few examples where we’ve cost-effectively algorithmically created content that artists used to hand-build. However, they are a tiny slice of the art needed for a typical commercial game, and their application is constrained by their lack of style (meaning it’s hard for the art director to tell SpeedTree to make the trees ‘more Tim Burton’.
Spore is a great example of how even an incredibly well-funded effort by geniuses failed to reduce their need for lots of hand-built art to ship a fun game. The Spore team has lots of human game artists who create important parts of all the art in the game. Spore’s walking animations are not magically generated. They are based on a library of hand-built, funky, entertaining walk cycles, then heavily modified to the shape provided. Yes, they allow limitless character designs…but from a kit of parts that are based on hand built art. That kind of procedural art will never threaten game artists’ jobs. For more info, see Chris Hecker’s 2007 GDC talk detailed Spore’s successes and failures in using a variety of coding tricks to leverage hand-created artwork into a limitless variety of creatures. (http://www.chrishecker.com/How_To_Animate_a_Character_You%27ve_Never_Seen_Before)
Wow, that turned into quite a rant! I don’t actually feel as strongly about this topic as my post implies, but apparently I certainly enjoy writing about it. 🙂
cheers
-Josh Whitkin
Computers should be doing the boring work. If procedural art means that no junior artist is stuck texturing rocks and sand all week, then that is a good thing. If it means there are fewer junior artist positions… that’s fine too. I’d rather have 50 artists employed doing awesome stuff than 500 artists employed, and 450 of them doing soul-crushingly boring work.
@Ian – “You compare outsourcing and procedurally-generated content as if they’re two different things.” I think they are. Human beings are inherently better at art than computers, if only because we can make illogical decisions that are intentionally beautiful.
Procedurally generated anything doesn’t threaten to replace artists, designers, etc. because not every project will have a need for them.
Procedural generation is a powerful tool that only the video game medium can employ to its full potential, one which can allow some truly amazing things to happen that will push us well beyond the capabilities of what anyone else can do.
But that doesn’t mean every game will want to use that tool.
A hammer is a great tool, but not every job requires a hammer.
Spore is amazing, but I don’t think neither players nor developers want nothing but Spore-like products from here on out. We’ll still want hand-crafted things, and to craft things by hand. All procedural generation does is give us the potential for new flavors we’ve never tasted before…. but that doesn’t mean they’re necessarily better flavors, or one we’d want to taste all the time.
It’s just new. That newness is truly remarkable, and something I look forward to. Procedural content is a new frontier. It is something to be excited about.
And just like everything new, some people are going to view it as something vaguely threatening.
Like, y’know… video games themselves.
That doesn’t make it a serious threat.
It’s like the pen tool and the fill tool in visual editing programs. The fill tool allows for more time and effort to be put into the details made with the pen tool, while still reducing the total time and effort put into the piece. I’d say that definitely makes procedurally generated art good, especially for artists.
Hey Brenda, I agree with you. I feel this is one of those issues (like ‘are games art’ or ‘are games bad for society’) that need to be patiently explained over the course of a few decades until everyone gets it.
It seems to have worked for ‘ZOMG branching storylines are teh awesome’. Or so I hope: sometimes I feel that particular meme is like a super-resistant bug. (Please don’t come to me with the ‘but in situation X branching storylines are fine’ counter-meme – I KNOW.)
Someone with more knowledge of 20th century art than me (ie most people) and some time on their hands could probably write a really persuasive article showing how proceduralism, or rather indirect control, can be used to create art.
Oh, and there seems to be an obvious analogy between procedural content and video games. The ‘loss of control’ of an artist vis-a-vis procedural content is pretty much the same as the ‘loss of control’ of a writer or director vis-a-vis a video game. NOOO YOU BARBARIANS GO LEFT THAT IS WHERE THE STORY IS.
I couldn’t agree more with you. Procedures are tools. And since art gets more and more complex, we need more complex tools to create it.
On the other hand, I think your statement is not supportive for pixel artists 🙂
Check out SYNTH(tm) the video game for a 100% mathematically gen’d game,. very crazzy looking
You are dead on, art in all arenas is evolving, by the time you catch up, it will be evolving ahead of you.