Two computer science researchers have come up with an algorithm that can take a low-resolution piece of pixel art and upscale it accurately to vector graphics.

Microsoft’s Johannes Kopf and Dani Lischinski from the Hebrew University of Jerusalem built the algorithm by blending a number of approaches, including pixel analysis and spline curves. These are already used in the vectorization of bitmaps, but as the new algorithm focuses solely on 8-bit pixel art, it can take the art form’s particular quirks into account and produce results with far fewer graphical artifacts than more generalized approaches.

In the research paper (.pdf) — offline at the time of writing but handily mirrored at Imgur — the pair say:

Because of the hardware constraints at the time, artists were forced to work with only a small indexed palette of colors, and meticulously arrange every pixel by hand, rather than mechanically downscaling higher-resolution artwork. For this reason, classical pixel art is usually marked by an economy of means, minimalism and inherent modesty, which some say is lost in modern computer graphics.

As a result, says the paper, every pixel can be a feature on its own, or carry important meaning. Other vectorization algorithms tend to lose detail when they’re given pixel art as an input. The researchers claim their approach is “well-suited to pixel art graphics with features at the scale of a single pixel.”

You can see some of the results in Wired UK’s gallery. The algorithm doesn’t always work perfectly, admit Kopf and Lischinski, especially when it comes to the anti-aliased Doom face, and there’s also the question of whether certain aspects of pixel art — like Space Invaders — should have nice rounded edges.

It remains a research project for now, but how awesome would it be to see SNES games upscaled to HD resolution on the forthcoming Wii 2? We hope someone at Nintendo is paying attention.