The larger goal, says Bala, is to create artificial worlds that are as convincing and natural to inhabit as the real one. Bala is forging toward this goal in at least two areas. Using a computational technique she and her colleagues call “lightcuts”, she is finding economical ways to create graphics that—like many natural scenes—feature multiple light sources that differ in quality and intensity. With feature-based textures, she is addressing a problem common to graphics in computer games.

“When you get close to an object in a game, you are signaling a desire to see more,” she says. “But the image tends to look worse. If you use feature-based textures, you can get arbitrarily close, and it still looks good.” To illustrate this point, she shows me an example of a character from the best-selling game Warcraft III: where an extreme close-up of the wizard’s tunic looks blurry in the original game, but you can follow every crease and fold in Bala’s version.

Of course, solving such problems conceptually isn’t enough to make the solutions practical. The second, perhaps more challenging problem is applying them in the face of “real world” limitations such as processing speed or memory. “We can have this insight (about the visual system), but then there’s the cleverness of making [our solution] work so that the program is still fast; it’s still lightweight,” she says.

Although computer game makers are big consumers of her work, the Bombay-born Bala sees a world of other applications for her NSF-supported work. It applies, in principle, to anything that requires creating plausible scenes, such as making art, virtual or “augmented” reality, lighting design, architectural rendering and, inevitably, to applications that haven’t been thought of yet.

“When you interact with this”—and she picks up a computer mouse—“it’s not a very natural metaphor for dealing with the world.” In the future, though, we will live naturally—and productively—in the sophisticated yet intuitive artificial worlds Bala is helping to create.