Google's Deep Dream image recognition tool is now available for anyone to use, with results ranging from eerily beautiful photographs to absolutely horrifying snaps.
The software is currently available to download from code-sharing platform Github, and allows users to create dreamscape versions of their own images using the image recognising capabilities of Deep Dream.
Small details of the pictures are picked out by the software and exaggerated in strange and mesmerising ways.
Google says the tool was primarily created to research artificial neural networks, providing new insight into image classification and speech recognition.
The artistic images appear to have been an accidental by-product, but led the company to make the software open source.
"We have seen a lot of interest and received some great questions, from programmers and artists alike, about the details of how these visualisations are made," Google's software team wrote in a blog post.
"We have decided to open source the code we used to generate these images in an IPython notebook, so now you can made neural network inspired images yourself!"
Deep Dream works by over-interpreting images, focusing on benign aspects of a picture and grossly exaggerating them into recognisable shapes - often animals.
To help with the research, Google has asked users to share their results with the hashtag #deepdream, and some interesting images have already emerged (via @hutstaender and @cdotwright).
