Wednesday, April 29, 2009

Game Graphics: Diminishing Returns?

From a historical perspective, we've seen huge leaps in the graphical sophistication of videogames -- compare Pong to Bioshock and they're just worlds apart, even though under the hood they're doing exactly the same sorts of things. The current generation of consoles just does it all much, much faster, with a bigger palette to draw from (and on) and hardcore support for 3-D math. And that has produced some very impressive results.

I am inclined to think that the law of diminishing returns is starting to kick in, however. It was a huge leap from black-and-white, color-stripped Space Invaders to Galaxian in full color; the NES era had a limited color palette but delivered sufficient resolution to allow some differentiation in art style among publishers and developers. The SNES generation improved on that with better color (and sound); then the Playstation brought consistent 3-D to the table. And then... what? Well, the Dreamcast upped the resolution to 480 lines; the PS2 included hardware support for 1080i, though it went largely unused; and the modern consoles support full HD, except the Wii which tops out at 480p. There are other measures, of course -- draw distance essentially became a minor concern with the PS2 generation, and bump mapping has only come into its own recently. There are lots of little things that can be done to improve the realism of computer graphics.

But where do we really expect to go from here? It's hard to imagine that the next generation of consoles will do very much that looks different from this generation. There's always the "real time ray-tracing" target, of course, but ray-tracing tends to look geometric and computerish, not organic, and I'm not sure it's really the holy grail of graphics technology it's made out to be. Resolution (for televisions, anyway) and color depth are maxing out. I think we will continue to see better physics, more consistent frame rates, more heavily-textured and layered polygons, and more moving objects. But all of these advances will be incremental improvements, not revolutionary steps forward. Eventually we will have filled the screen with pixels doing interesting things, and find that there's not really anywhere else to go on that front. There's true 3-D, of course, but that's more of a display technology (possibly requiring a 120 Hz refresh rate, I should note) than an image generation advance.

So it will be interesting to see what happens in the "next generation". My expectation is that Wii 2.0 will look more like the PS3/360 generation, because it could definitely stand to catch up a bit. But what about the PS4/720/what-have-you? Seriously, from here on out I expect the "wow" factor of graphical advances to start wearing off. Even handhelds will eventually catch up with their console brethren on the cosmetic front.

And I hope that means we'll see game design return to the forefront. Innovation for the past few decades has gone largely into improving the audiovisuals, and I appreciate the benefits of those advances. But there's something about technological constraints that seemed to unleash developers' creativity back in the day. The current retro/downloadable trend has produced some fine contemporary examples like World of Goo and Braid -- it's easier to take risks with lower budgets, and technological targets that don't require doubling your graphics staff each generation. And as the investment curve in new graphics possibilities starts to level off, I hope we will see developers freed to take bigger, more interesting risks.


  1. One of the things game designers seem to ignore is how beneficial MORE, rather than BETTER, can be; World of Warcraft exploits this by having its sub-par graphics as a trade-off for no slow-down for countless hordes of characters on the screen at one time. I find it hard to believe that in, say, Half-Life 3, there will ever be a time when Gordon Freeman faces down an onrushing army of 500 enemies. Or that there will be a game anytime soon that can realistically display an entire city without taking shortcuts like popping-up buildings or loading sections at a time or something. Immersion in the future should be more and more possible if the focus shifts like this, but unfortunately it will instead probably remain dedicated to "more realistic eyelashes!!!"

  2. Thanks for sharing your thoughts. More memory and faster storage media have made "seamless" environments more feasible -- popup and texture catch-up are still challenges, especially in open world "sandbox" games where the designer can't take advantage of linear progression to hide the missing faces and such, but there's a visible difference between GTA IV and Driver. A game that calls for hordes of enemies might be hard to pull off in 3-D, but doing it impressively in 2-D is feasible with current technology. Sometimes it's a matter of fitting the display approach to the design goals. I'm just hoping that the current generation gets a chance to breathe and let game design mature -- that does seem to be happening, actually, as the XBox 360 is about five years old and no successor has been announced.

    Any art is a matter of compromise and choices -- a thoughtful designer will know that, say, Lara Croft's eyelashes only count in the cutscenes, and those polygons and lighting shaders can be better spent on the environments. The better the technology, the fewer the contraints on the technical side, but the push for more and more horsepower on the purely visual side is waning. Physics is likely the next frontier. AI will remain elusive for quite some time to come, I think. But I'll be interested in how it all plays out!