DALL-E is a neural network that turns images into paintings. It was created by OpenAI, a synthetic intelligence research lab. The name DALL-E comes from the Disney film Wall-E and the name of the famous artist Salvador Dali. DALL-E is trained on a large dataset of images from the web. The goal of DALL-E is to generate images from textual descriptions, corresponding to “an orange with a blue stripe.” To do that, DALL-E uses a Recurrent Neural Network (RNN). RNNs are a style of neural network that may process sequences of knowledge, corresponding to text. DALL-E will not be the primary neural network to generate images from text descriptions, nevertheless it is the primary one to achieve this in such a high resolution. For example, the image to the fitting was generated by DALL-E from the outline “computer science, digital art.” As you possibly can see, DALL-E was capable of create a wonderful high-resolution image. This shows that DALL-E has a deep understanding of the content of images.

We used the outpainting prompt “fill with flowers“ to realize this incredible result.

So how does “outpainting” work?

The “outpainting” feature of DALL-E allows users to expand a picture in any direction. For example, if you could have a picture that is simply too small to fill the specified frame, you should use outpainting to expand it toward the left and right sides of the image. This is an ideal solution to add more detail to a picture without having to begin from scratch. To use outpainting, simply select the image editing tool from the menu after which click and drag a box within the direction you ought to expand the image. The further you drag, the more the image will probably be expanded. Outpainting is an ideal solution to add more interest to an otherwise plain image.

And technically speaking (but still for AI noobs)?

Outpainting analyses the unique image and labels it (e.g. chair, sunset, computer, digital art, purple floor, wall on the left of the window, and really many more). Then latest images are generated, let’s say in a random fashion for now – this varies by model and is slightly exhaustive as a subject, and these generated images are labeled too and matched to the tens of millions of original labels to see how well they match. The user is presented with the very best variants (and may opt to re-generate more suggestions) and once confirmed the image is prolonged and the method can start over. Outpainting is definitely quite fun to play with using DALL-E, which is now available at a really low price with a couple of free samples to start.

Interest in outpainting over time

As you possibly can see outpainting is a brand latest topic and interest is rapidly rising, especially in China.

This article was originally published at www.artificial-intelligence.blog