Home Top

The face that is EVERYWHERE: Creepy image combines thousands of photos of inanimate objects that look like people

Image created by New York-based artist, Robby Kraft, using software
He combined 2,500 images on Instagram tagged under #FacesInThings
Used on Instagram to show photos of ordinary objects that look like faces
Kraft used same software to create average face of selfies and white noise

By Ellie Zolfagharifard For Dailymail.com

Published: 15:58 EST, 16 November 2015 | Updated: 18:09 EST, 16 November 2015

View
comments

From grilled cheese to a piece of toast, many of us think we see faces in random places.

This is down to a psychological phenomenon known pareidolia, which causes our brains to perceive familiar sights in everyday objects.

Now one artist has revealed the creepy face lurking beneath these objects, by using software to average the appearance of Instagram images tagged as #FacesInThings.

Scroll down for video

This is the average face that appeared from all the 2,500 images taken from the Instagram hashtag #FacesInThings. '#FacesInThings makes no distinction between human or animal, yet this image looks human,' said Robby Kraft. 'When I look for a gender, male and female both seem to fade in and out, very androgynous'

The hashtag is used on social media when people see a lot of random objects that are similar to a human face. This can be in anything from a coffee cup to a toothbrush holder.

To create the image, New York-based artist Robby Kraft, used an algorithm designed to average out the faces of real people.

According to the Daily Dot, the idea came to Kraft after he spent a week studying artist Jason Salavon, who is famous for manipulating data and images using software.

Kraft used the algorithm on 2,500 images taken posted with the hashtag #FacesInThings.

While only around 20 were identified as a face, the final pattern is an image that looks strikingly human.

An example of some of the images on Instagram tagged under #FacesInThings. Many of us think we see faces in random places. This is down to a psychological phenomenon known pareidolia, which causes our brains to perceive familiar sights in everyday objects

Robby Kraft used the same algorithm to average out images of human faces using the hashtag #Selfie on Instagram. The algorithm aligned, layered and blended 200 of these images to produce this 'average' face

Kraft attempted the technique with pure noise to see if a face would appear. The images were found by creating random photos of light and dark spots. Pictured is one example

Kraft also used the algorithm to average out images of human faces using the hashtag #Selfie on Instagram.

The algorithm aligned, layered and blended 200 of these images to produce this 'average' face.

'The algorithm was able to detect faces in about one in three,' Kraft writes on his website. 'These were layered and alpha channels were blended evenly.'

When using #selfie images, a recognisably human face would appear after combining about 15 photos.

WHAT IS FACE PAREIDOLIA?

Pareidolia is the psychological response to seeing faces and other significant and everyday items in random stimulus.

It is a form of apophenia, which is when people see patterns or connections in random, unconnected data.

There have been multiple occasions when people have claimed to see religious images and themes in unexpected places, especially the faces of religious figures.

Many involve images of Jesus, the Virgin Mary and the word Allah.

For example, in September 2007 a callus on a tree resembled a monkey, leading believers in Singapore to pay homage to the Monkey god.

Another famous instance was when Mary's face was a grilled cheese sandwich.

Images of Jesus have even been spotted inside the lid of a jar of Marmite and in a potato.

Remarkably, this was the same number needed to get an average human face using the #FacesInThings hashtag.

'#FacesInThings makes no distinction between human or animal, yet this image looks human,' said Kraft.

'Also when I look for a gender, male and female both seem to fade in and out, very androgynous.'

Kraft attempted the same technique with pure noise to see if a face would appear.

The images were found by creating random photos of light and dark spots.

By generating over 7,000 images, faces were detected in around 0.6 per cent of them, or in 47 images.

While it's less defined, the result still bares some feature to a human face, with dark areas where the eyes and mouth would be placed and a lighter vertical strip for a nose.

'This is a visualisation of the face-detection algorithm; or a computer peering out from the void, learning a human face,' he says.

By generating over 7,000 images, faces were detected in around 0.6 per cent of them, or in 47 images. 'This is a visualisation of the face-detection algorithm; or a computer peering out from the void, learning a human face,' he says.