A scene from the 1998 film Fear and Loathing in Las Vegas as interpreted by Google's neural network.
Google's artificial neural network, Deep Dream, is capable of producing some amazingly trippy images.
A typical neural network looks for features of an image that match a particular concept, like a banana. But the Google program turns that on its head, by tweaking a random image to produce features that eventually resemble a banana.
Recently, YouTube user Roelof Pieters had the bright idea of running the program on scenes from Fear and Loathing in Las Vegas, the 1998 cult classic which is basically Johnny Depp on one long acid trip.
The results, needless to say, were disturbing. Creepy animal heads with eyes like gaping orbs materialize in a darkened room.
Google's image recognition systems are trained to look for specific objects, such as cats and dogs. But when the company's engineers fed in images of landscapes and static noise into Deep Dream, the AI spat out strangely artistic images, like this one featuring fountains and pagodas:
Or this one, which looks like a "magic eye" pattern:
But applying the AI to a clip of Fear and Loathing takes the trippy to a whole new level.
See for yourself, if you dare: