How to create images with AI

By the end of this, you will have an AI-generated image you are happy with, made from a description you wrote and refined until it matched what you had in mind.

What you will need

• An image tool: a text-to-image feature inside a chat tool, or a dedicated image generator. There are many, and they change often, so pick one that is available to you rather than chasing the newest name.
• A clear idea of what you want, even if it is only a rough picture in your head.
• About 20 to 30 minutes.

The kind of AI that makes pictures from words is called generative AI, and the feature that turns a written description into an image is often called text-to-image. You do not need to know how it works to use it well. You describe what you want, and it makes a picture.

Step 1: Describe what you want in plain words

Start with a clear, simple description. AI image tools respond to detail, so think about the subject, the setting, the mood, and the style. You do not need special wording, everyday language works fine.

A cosy reading nook by a rain-streaked window, a mug of tea on the sill, warm
lamplight, a stack of books, soft and calm mood, painted illustration style.

What to expect: an image that captures most of your description, though rarely all of it on the first try. Look at what it got right before you change anything.

If it does not work: if the result feels off, do not throw the whole thing away. Note the one or two things you would change and move to the next step.

Step 2: Refine one thing at a time

The quickest way to get lost is to change everything at once. Adjust one element, see what happens, then adjust the next. Most tools let you keep going from your last description.

Keep the same reading nook and warm lamplight, but make it evening rather than
daytime, add a sleeping cat on the chair, and use softer, cooler colours.

What to expect: a new version that keeps what worked and changes what you asked for. Repeat this a few times, nudging it closer with each pass.

If it does not work: if a change is ignored, describe it more plainly and put the most important part first, since the opening of your description tends to carry the most weight.

Step 3: Get help writing the description

If you are not sure how to describe what you want, ask a chat tool to help you put it into words before you generate anything. This is a good way to discover details you had not thought of.

I want to create an image of [describe your rough idea]. Help me write a clear
description for an AI image tool. Ask me about the style, mood, colours, and
setting if that would make the description better.

What to expect: a fuller, more vivid description you can paste into your image tool, often with choices you had not considered. Adjust it to match your taste before you use it.

Step 4: Set the shape and use

Think about where the image will go. A banner, a square social post, and a phone wallpaper all need different shapes, and most tools let you choose.

Create this as a wide landscape image suitable for a website banner, with space
on the left side where text could sit.

What to expect: an image in the shape you need, with room for whatever you plan to add. If the shape is wrong, say the exact format you want, such as square or portrait.

Wrapping up

You now have an AI-generated image made from your own description and shaped for its purpose. Getting a good result is normal to take a few tries, so treat each attempt as a step closer rather than a pass or fail. If you plan to use an image publicly, check the tool’s rules on how its images may be used, as these differ from one tool to the next.

A tip to improve faster: when an image turns out well, save the description that made it. Over time you will build a personal collection of wordings that work, and describing your next idea will come far more easily.