Get More Variety in AI Image Styles

You get better results with AI when you can clearly describe what you want. Just like working with LLMs like ChatGPT, image generators do best with clear directions. One of the aspects of AI image generation that I appreciate is how I can generate different styles of illustrations. Once I get something I like, I can generate a series of related images in a similar style to have a cohesive look and feel. But first, you have to know how to describe different image styles. Changing just a few words in your prompt can generate significant variety in your AI image styles.

Example AI image styles

I generated a series of images in Midjourney with the same prompt, just changing the style at the beginning of the prompt.

[Image Style], a woman with shoulder-length auburn hair working at a computer, dark blue shirt, modern office

Styles used:

Digital illustration
Flat vector
Graphic novel
Cartoon
Colored pencil
Watercolor and ink
Vector illustration
Anime

8 AI-generated images of a woman with auburn hair and a dark blue shirt working at a computer.

Even though the prompt is almost the same, you can see how different the results are. Some of these styles will be more or less consistent depending on additional parts of your prompt. “Vector illustration,” for example, can have simple shapes or more complex details, realistic body proportions or exaggerated features.

Comparing AI image tools

You’ll get different results depending on which AI image generation tool you use. This comparison shows results of the “vector illustration” version of the prompt above from four different tools: Gemini (Nano Banana), Ideogram, ChatGPT, and Seedream.

4 versions from the same prompt in different AI image tools.

Browse image styles

Some AI image generation tools have built-in styles to pick from. This makes it easier because you don’t have to know what words to use to describe the style; you can look through the examples and select one. For example, Recraft has a wide range of styles for both vector and raster art.

Brushless has a smaller set of built-in styles, but you can add your own styles based on sample images. Brushless also has fairly good consistency within each style. The “Flat Icons” style is useful for training purposes. The “Structured” and “Humanist” styles could easily work for training too.

Screenshot of image style options from Brushless.

Other ways to find image styles

One helpful strategy for figuring out what words to use to describe an image style is to find an image you like and ask ChatGPT or another multimodal tool to describe it.

Another option is to look at other examples. Many AI tools have public feeds for comparison where you can see how other people have prompted their images and learn from them.

You can also check out collections of AI art styles.

Consistent styles

If you want to create a series of images with a unified look and feel, check out my previous post on generating AI Images with Consistent Style in Midjourney.