Pixels vs. Polygons

I play games from a pretty broad time perspective, which means I often get to see technology from one era juxtaposed against technology from another era.  It struck me recently that there are really two distinct styles of game art -- both are still in use today, but one was not possible during the early days.

In the early days of gaming, the pixels ("picture elements") WERE the artwork, and the artist had to design a recognizable image within strict limitations on memory and display capability.  But the artist did have choices -- even on a limited-resolution system like the Atari 2600, there were 256 colors available, and graphics and colors could be changed once per scan line, or once on every other scan line, or even changed at varying rates to provide more detail in some areas of the screen than others.  Of course, the game's programmer had to agree that the proposed design was technically feasible, but as more often than not the programmer was the artist, this provided a degree of flexibility.

As time went on, a broader palette of colors emerged, and individual sprites could be defined using more than one color; the NES managed four colors (usually implemented as three plus a transparent color to let the background show through), and the SNES allowed 16 along with transparency effects.  The Atari Jaguar, 3DO and Sega Saturn upped the ante further, to 256, 32,768 or even 16.7 million colors.

And then, just as pixel art was starting to mature and look really good, the technology changed radically.  Great-looking 2-D games were shunted aside in favor of early, clunky 3-D efforts.  In 3-D, pixel art was used primarily for overlays, and also as texturing to dress up the low-polygon models of the PSX era.

Game designers had challenges to deal with in this new era -- it took a while before Mario 64 and Tomb Raider established that platform games could in fact be played in 3-D environments.  But artists had work to do too; suddenly, the end result of the artist's vision could not be realized with a graphics tablet and a pixel editor.  Graphics production became more like an animated film -- start with a concept painting, sculpt the objects in polygons, texture them with artwork, and finally light them to look like the artist's original intent.

The first wave of 3-D hardware was barely able to get recognizable images onscreen -- lighting was flat and simple, textures were warped and grainy, with visible individual pixels, and there weren't enough polygons to provide much detail.  The N64 smoothed out the textures, but didn't have enough memory for very detailed artwork.  The Dreamcast and Playstation 2 technology made significant improvements -- resolution was higher, colors more varied, lighting more sophisticated, and enough polygons were available to provide for wide vistas and deep draw distances.

The current generation of hardware has made even more headway -- lighting is now accompanied by (admittedly low-resolution, often fuzzy) shadowing, and bump mapping allows a flat texture to look richer and more detailed when properly lit.

And pixels?  Well, these new machines can certainly handle them with aplomb.  HD graphics, deep color palettes, and more layers of software-based parallax scrolling than any 2-D hardware system ever dreamed of have opened up a broad range of possibilities.  Fortunately, game designers seem to be realizing this -- we do still see decent 2-D games come out on occasion, and most of them look absolutely gorgeous.

It reminds me that it's important to look back once in a while -- just because a graphical technique originated decades ago doesn't mean it's tapped out.