# AI Deep Dream Art Hi guys, These are images an AI generated. It transferred it's style onto a new images. The images are created using a Neural Network, specifically a Convolutional Neural Network. ## The Art From this image: ![](https://www.pictshare.net/0q6i412jkp.jpg) the AI generated: ![](https://www.pictshare.net/o5rkxgutuz.jpg) ![](https://www.pictshare.net/ud0nl45cmy.jpg) ![](https://www.pictshare.net/y4519ztoub.jpg) ![](https://www.pictshare.net/1izdh14ta7.jpg) ![](https://www.pictshare.net/hwuga63m91.jpg) ![](https://www.pictshare.net/8blugsmyco.jpg) ![](https://www.pictshare.net/e0qro5p5sx.jpg) ![](https://www.pictshare.net/713sgpt5w5.jpg) ![](https://www.pictshare.net/isq0e0pack.jpg) ![](https://www.pictshare.net/wez912xz7v.jpg) ![](https://www.pictshare.net/ej4jahihqt.jpg) ![](https://www.pictshare.net/mzv9ssq2n9.jpg) ![](https://www.pictshare.net/8e3wf57z1k.jpg) ![](https://www.pictshare.net/dau13cupt1.jpg) ![](https://www.pictshare.net/b7i6mfx6xf.jpg) ## Explaining the AI This Neural Network falls into the category of a generative model and demonstrates how AI can be creative. The steps the Artificial Intelligence takes to produce these images put simply are: 1. The Neural Network which is best at handling images, the Convolutional Neural Network, is trained on a dataset with millions of images. After training the Neural Network is at a point where it can detect objects, animals, people ect. 2. The Neural Network is then fed forward a new image. 3. Since the Neural Network was pretrained it is able to pecieve where in the image an object is most likely to appear. This perception the Neural Network makes is then merged with the original image on a pixel level. 4. The merged image is then fed to the Neural Network. 5. Steps 2-4 are then repeated. For more information you can check Google's research [blog post](https://research.googleblog.com/2015/06/inceptionism-going-deeper-into-neural.html). Happy AI blogging!! Make sure to up vote and follow if you enjoyed. From neurallearner :)