Just over a year ago, we were all obsessed with a visual trick: was the dress white and gold, or black and blue? Why wouldn’t we be? It was fascinating! Why would we see something – the same image – completely differently to our closest friends? Well, just a month after that, an eye test using research from the Massachusetts Institute of Technology (MIT) came out on YouTube. That’s right: YouTube! Take a look!
Now, we at Concept don’t recommend that you take this as a full eye test – it isn’t. The video suggests that if you see Marilyn Monroe as the image gets closer, then you need an eye test. If you think something’s wrong, get a check. You need to have an eye test every two years to stay on top of things too. However, the science behind this is amazing!
Everything that we’re talking about is based on research by Aude Oliva, who works on Computational Perception and Cognition at MIT. She originally made the “Marilyn Einstien” image for a 2007 New Scientist article. What’s happening in the image isn’t as complex as it might seem. To make the image, Oliva added two filters – one to a photo of Marilyn Monroe and one to a photo of Albert Einstein. The filter over Monroe is a low-frequency filter, which removed all the sharp edges from her image. The opposite was applied to Einstein – a high-frequency filter, which leaves only the sharp edges of his face. When the image is overlaid, we see Monroe at a distance and Einstein up close.
The question is why? In some ways, that’s an open question too. Hybrid images are still being used and researched to better understand how our eyes process images. However, a paper on hybrid images produced by Oliva and MIT in 2006 can shed some light on the question. To make a successful hybrid, there are quite a few factors to take into account, most of them to do with how the images are blended or combined.
Low spatial frequencies – or blobs – like Monroe lack precise definition, which is normally given by sharp edges – boundaries and shapes. Our brains tend to make groups of the blobs in order to understand them. Almost the opposite is true of high spatial frequencies: we understand the definition and fill in the rest of the image. When there appears to be a high spatial frequency, that isn’t contradicted by the rest of the image, we go from there. That’s why we see Einstein up close, when we can make out the details and Monroe when we can’t.
The secret to a good hybrid is that we shouldn't be able to consciously change which image we see, as the alternative image should only appear as noise or not at all. This means that the images need to be blended well. For example, in the picture below, the elephant had his back blended with the horizon. The blobs he's made up of make sense in both perceptions.
As we said, weird. Whilst the video above is certainly no substitute for an eye test, it’s certainly one of the most interesting things we’ve seen in a while!