Fast Company

3M's Visual Impact Scanner Knows What Your Eyes Want

After decades' worth of vision research, Minnesota-based conglomerate 3M has tuned up its Visual Attention Service, an algorithm that can scan all types of content to determine exactly where the average human eye is most likely to be drawn.

3M Visual Attention Service

Open a website today, and you'll be struck with a circus of activity: bold headlines, splash images, drop-down displays, flashing ads. What grabs your attention the most?

That's the question that advertisers and designers alike have tried to figure out over the years. But until now, the best answers came from either trial-and-error practices or eye-tracking studies, which use software and cameras to monitor a user's gaze--and are often slow, impractical, and costly. Minnesota-based conglomerate 3M has come up with a different solution, however, and it's based on decades' worth of vision research from PhDs and neuroscientists. Called the Visual Attention Service (VAS), 3M has created an algorithm that can scan all types of content (graphical, text) to determine exactly where the average human eye is most likely to be drawn.

"It predicts visual impact," says William Smyth, a business manager at 3M's graphics division, who has roughly 30 years experience in image hardware and software. "Whatever creative content you're working on, we can help make sure it's going to get people's attention."

VAS is a web-based scanning tool. Simply open the service on 3M's website, upload an image or enter a URL, and VAS will scan the image--in about 20 seconds--and build heat maps and spit our probability scores for the content that is most eye-catching. "When we view things, our eyes actually produce more data than our brains can process, so our brains make decisions about what we should pay attention to," Smyth says. VAS's algorithm processes those visual cues--color, edges, faces, shapes, contrast--analyzing and predicting our instinctive responses (that is, our focus for the first three to five seconds) to stimuli.

"That list of elements--color, shape, contrast, and so on--those are things designers play with," says 3M's Kelly Canavan. "It is all meant to help designers balance the trade-offs they have to make when designing, say, an ad."

NYT heat map

It's remarkable how specific 3M's system can be. Analyze The New York Times, for example, and you can see exactly what attracts our attention: The eye is drawn down the center of the page, following the large and colorful images, while the list of text on the left-hand column is completely ignored. But let's dive deeper. Say we want to determine which advertisements on the page are most effective. VAS allows users to draw "areas of interest" boxes around any spots for comparison. After analyzing the selections, VAS will spit back percentage figures which represent the probability that a visitor will notice a particular design element when first opening the page. In this case, Lincoln's SUV ad is far more noticeable than Hublot's logo and silver watch sitting dark in the upper corners.

NYT VAS

On Fast Company, we can see that Google's ad performs almost twice as effectively as iMeet's.

FC VAS

"It's so convenient that you can use it during the design process," says Smyth, "and compared to an eye-tracking study, it's very affordable" (For individual users, a three-month subscription will cost about $250 per month. For large groups and the enterprise, that figure lowers depending on the number of users. Alternatively, you can always check out 3M's free trial first, although the scanned images are watermarked.)

In addition to the web-based scanning tool, 3M offers a VAS plug-in for Photoshop as well as an iPhone app. (Android is on its way.) And the company is gearing up for some big upgrades, which should make VAS even more valuable to advertisers. Soon, users will be able to do side-by-side comparisons of content, and automate an ad's effectiveness across myriad environments--say, if you wanted to test a banner ad on 10 different websites.

And in the next version of VAS? The company is already testing beta versions for video analysis. Imagine walking through Times Square with a video camera and determining which billboards are the most appealing for consumers.

Oh wait. You won't have to: 3M's VAS algorithm will do the work for you.

Add New Comment

0 Comments