AI in graphic design: can robots be artists too?

As one of spark*’s motion designers, I was tasked with creating an animation to celebrate this past National Coffee Day – if you know the spark* team, you know we drink a lot of coffee, and this day couldn’t pass by unrecognized. I was inspired to try something new: developing an animation through the power of AI. You may remember this video – these unique visuals for social media were designed using open-source artificial intelligence, a cutting-edge technology that is quickly taking over the world of experimental design.

What is open-source AI, and how does it work?

Understanding AI may seem overwhelming, but it doesn’t have to be. Here’s a simple breakdown: there are open-source datasets available to the public that have been trained with all the images on the internet. A dataset is a collection of images that have been labelled and categorised according to the content of the image. For example, an image of a cat will be labelled as “cat”. An AI image generator (like Dall-E) takes word prompts from the user and creates an image using these datasets.

How did we create our Coffee Day animation?

To gather the inputs for our animation, I collected video prompts from spark* team members about what coffee meant to them. I then added them to Dall-E, and the results were amazing.

While I was creating this, another AI program, Stable Diffusion, had just been released to the public. This program was faster, had less restrictions, and unlike Dall-E, allowed me to make animations. In earlier attempts, I gave the AI all the prompts together and hoped it would make a seamless animation. As shown below, it is easy to generate images with this AI, and it is equally probable that those images will turn into beautiful noise.

In a way, I had to guide the AI to navigate through the animation. To avoid uncertainty, I generated the animation for each prompt individually.

Other uses of image generation

Artificial intelligence has many uses when it comes to generating images. “Inpainting” is a common one. When given one half of an image, the AI can guess and create the other half.

“Stylization” can make an image look like it was created in a specific style, such as art deco or impressionism. Images can even be stylized as if they were created by other artists, like Van Gogh. For motion designers, image generative technology is exciting because of its ability to help stylize simple animations.

The future of image generators

AI image generators have an opportunity to be the future of graphic design. This relatively new technology can do what no others have: give anyone the opportunity to visualize their imagination. No coding, design, or artistic experience required.

Dall-E has only been available since 2020 and has already made a splash in the world of AI and design. You can browse the hashtags #AIart or #StableDiffusion on Twitter to see examples of AI generated art in marketing, entertainment, and more. This thread shows examples of how AI can learn and reproduce a product on demand for advertising uses, essentially enabling unlimited product photography for marketers.

New image generators are quickly being released to the public. When I first began this project, Dall-E was the best open-source AI image generator available. Then came Stable Diffusion, Midjourney, NightCafe, and many more, each with their own unique capabilities.

Parting thoughts

Only time will tell what the future of AI has in store for marketing. Although open-source image generative programs are impressive and allow anyone to create stunning visuals, regardless of design experience, many believe it will never replace the organic talents of artists. Instead, this tool will likely be used to complement the skills of designers, beginners and experienced alike.