Artist and programmer Clement Valla has been tracking anomalies like the ones pictured here, for years.

They show bridges and buildings that appear to be melting, twisting, or upside-down.

Valla thinks there about a competition between two ways of building the Earth.

One is a three dimensional model, like the bumps on a topographical globe that show mountains and valleys.

The second are the photographs from satellites, planes and its own Street View cars.

Google Earth is at least in part formed by spreading the latter over the former

That produces strange effects primarily when these two types of information compete.

This is a problem that Google has been working on. Google software engineer Joshua Schpok actually used one of Valla’s images to illustrate a proposed solution to the melting bridges in a research paper.

2013-05-02

Co.Exist

The Strange Art Of The Melting Bridges Of Google Earth

Once an artifact of the race to map the world, these odd anomalies found in satellite photos illustrate the difficulties in truly depicting physical space inside our computers.

Google’s Earth is slowly losing some of its strangest and most distinctive features, according to artist and programmer Clement Valla. He has been tracking these anomalies for years: bridges and buildings that appear to be melting, twisting, or upside-down.

“I discovered them by accident,” he wrote in an article for Rhizome. “These particularly strange snapshots, where the illusion of a seamless and accurate representation of the Earth’s surface seems to break down.”

The distorted images aren’t like the infamous errors in Apple Maps. Those breaks with realism were primarily about navigational functionality: They put things in the wrong place, and, as a result, if you followed the map, you would get lost. The Google Earth anomalies Valla found won’t get you lost, because they aren’t about where things are; they’re about what things are.

Specifically, Valla thinks there about a competition between two ways of building the Earth. One is a three dimensional model, like the bumps on a topographical globe that show mountains and valleys. The second are the photographs from satellites, planes and its own Street View cars. Google Earth is at least in part formed by spreading the latter over the former. “A photograph is a three dimensional surface inscribed on a two dimensional surface,” says Valla. “Google is distorting that back onto the 3D surface.”

That produces strange effects primarily when these two types of information compete. “Two things have to happen for these to really stand out,” says Valla. The first is for photographs to be taken at an angle, which results in “depth cues” like seeing the top and side of a building, or a shadow under a bridge. The second is a landscape with extreme topography, like the sudden drop of a valley, or the rolling hills of Hong Kong. Once such a photograph is spread over such a landscape, the “depth cues” in the photograph don’t match the depth cues of the underlying surface. And the result is, well, weird.

This is a problem that Google has been working on. Google software engineer Joshua Schpok actually used one of Valla’s images to illustrate a proposed solution to the melting bridges in a research paper. “Though these artifacts may have some artistic merit,” he wrote, footnoting Valla, “they can be visually distracting, difficult to parse visually, and break an immersive experience.” He presented an algorithm that would seek out bridges, in particular, by algorithmically looking at roads, and forcing them to follow a more traditional (not suddenly-drooping) pattern.

But when Valla surveys the changes he has seen in Google Earth, it hasn’t primarily been about algorithmic correcting of underlying topography, but what he believes is a greater reliance on more detailed photographs. “It is getting to be a more surface-oriented place.,” he says “And a more detailed place.” Clouds that obscured buildings have disappeared, and some of the upside-down skyscrapers have turned right-side-up.

Valla is the first to say he doesn’t know for certain how or why these changes are taking place. But given the amount of money Google (and Nokia) are pouring into their fleets of photography cars, the photographic explanation seems plausible. “And why are these companies doing this? Why is it such a battleground?” Valla asks rhetorically. “It does have to do with mobile phone technology.”

For an old-school map, specific storefronts are too temporary to bother including. But those are exactly the details that mobile phone users are increasingly interested in. “For the photographs, there’s no difference between a shrub, a garbage can, a van, and a building. It’s all three-dimensional surface information,” Valla says.

As mobile devices have shifted our geographic questions to an increasingly fine-grained level of detail, the mapped world looks less like a map--and more like a photograph.

Add New Comment

1 Comments

  • Driftus Maximus

    Image Generators used in Simulators have been dealing with similar issues for years. Depending on how they are modeled, rivers, fences, overhead wires etc can also have this kind of problem.