In the past 2 years generative AI programs have reached a threshold where they can create realistic images from a text prompt. This ability is founded on 2 important advances:

• The program has been training on a huge dataset of images, so it has a complex network of data that means it “understands” what is in an image.

• It can “understand” the text input from the user, creating a coded representation of the image from the words and phrases used. This is called natural language processing.

​

Go to one of the following image generation websites to try it out for yourself. You may need to sign up for an account before being able to use the site. Note that on the free tier of the service it may take a little longer to generate each image, and that any images you generate are usually visible by other users in a user gallery.

 

Bing Image https://www.bing.com/images 

Stable Diffusion https://stablediffusionweb.com/#demo 

Leonardo https://leonardo.ai/ 

 

INTERACTIVE

Work with another person to create the most unique image you can think of - try to make something that no one will ever have seen before. Like with chatbots, you should make your prompt specific, include things you want in the image and things you don’t want, and describe the style of image, eg: photograph, illustration, oil painting.

If you are stuck for ideas, here’s a quick exercise to spark your creativity. With your partner:

1. Think of an animal whose name begins with the first letter of your name.

2. Choose an environment where you would like to go for a holiday.

3. Pick an item of clothing that you are wearing.

4. Pick an art style, such as watercolour painting, illustration, cartoon, pixar, photo.

 

Now put these 4 things together into a prompt. For example:

A watercolour painting of a group of otters in a forest clearing. All the animals are wearing red scarves. One otter is wearing a green hat.