Visual magnetism – what draws you into an image?
In an ever increasingly complex and crowded visual environment, what do we actually pay attention to? As all should know by now, visual attention comes in at least two specific forms: bottom-up and top-down.
While top-down attention is typically viewed as volitional, effortful and motivated, bottom-up attention is the term for conditions in which sensory stimuli automatically attract attention. One way to conceptualise this, especially in terms of vision, is visual magnetism. A magnetic item is a piece of information that has specific properties that inherently attract attention more or less automatically. This could be due to intensity, specific colouring, positioning or overt changes in e.g. an image.
Eye-tracking is known to measure visual attention, but part of the limitation of this method is that it has a hard time in disambiguating between top-down and bottom-up attention.
But there is is (obviously) good news: I’ve devised a method that can determine visual magnetism in an image. By relying on decades of research on the visual system – and a decent amount of time gone in testing, adjusting and validating models, I have made an analytic tool that models visual attention to a degree that it is almost spooky.
Yes, you read right: you can now know where people will look, even without testing a single subject.
The method is (somewhat humouristically) labelled eye2D2, and you can go to this page to see more documentation emerging these days.
The model has been tested against eye-tracking studies of many different kinds of images, including ads, outdoor banners, in-store layouts, newspaper layouts, and much much more.
The method has an accuracy that is 80% compared to traditional eye-tracking. It can do the analysis at a fraction of the time (hours) and cost (5 images = $10.000). This is a huge improvement compared to traditional eye-tracking which uses 2-4 weeks (or more) and at a much higher price ($20.000 per image). Even better, since eye2D2 does not remember images, one can run multiple versions/tweaks of the same image
Below you can see some direct comparisons between traditional eye-tracking (left) and the new automatic measure of visual magnetism (right).