How to Create AI Images: A Complete Guide with Expert Advice
A picture says more than a thousand words. But with generative AI, you have to start with words to get that picture.
Generative artificial intelligence technology has exploded in recent years, leading to many creative services that can generate full-length images and video clips. That’s the “generative” aspect of generation AI: it can produce content that appears based on just a simple text prompt. Of course, the actual process is more complicated.
AI image generators rely on a database of existing content (text, images, videos, etc.) and/or scour the open web for information to create new images to fulfill your request. They allow anyone with an account and a prompt to create virtually any kind of image (usually within reason).
There are many AI imaging services available, but not all of them will deliver quality results or adequately protect your data. They also vary in price and features offered. While AI generators can be exciting to explore, getting started can quickly feel overwhelming.
That’s where this guide comes in handy. I tested and reviewed multiple AI image generators for CNET and discussed how the new technology is impacting digital creators. These services are certainly interesting and serve their purpose, but there is a learning curve. While they’re not good enough to replace human creators, they offer a range of creative possibilities.
Here’s how to create custom AI images.
Step 1: Choose the best AI service for your project
I have had hands-on experience with many AI image generators and creative programs. The best program for you depends on your project, budget and experience level. You can check out our full reviews and rankings, but these are my three program recommendations.
What is the best AI image generator for you?
CNET’s Editor’s Choice is Dall-E 3 from OpenAI, the company behind ChatGPT. Dall-E boasts impressive fast compliance and can handle more complex prompts with ease. The conversation flow (which means you can easily follow up by typing in the chat window) allows for easy editing after generation. Dall-E is also one of the more creative programs we tested. You’ll need a ChatGPT Plus account to access the program – $20 per month – and note that OpenAI may use your content for model training. But that’s possible submit a request so that the training stops on your data.
If you are looking for a free program, I recommend Leonardo AI. Canva recently acquired it, but you can still use it on the website. Leonardo offers a robust free plan, with more editing tools behind the paywall and several artistic styles to choose from. It also offers a unique rapid improvement tool that lets you rewrite a simple idea, filling in details that create beautiful end results.
If you’re really on a budget and have no creative experience, check out Canva’s Magic Media generator. It is extremely intuitive and easy to navigate, which is great for beginners and non-creatives. Canva gives you 50 free image generation credits (these are lifetime credits – you’ll need to upgrade for more credits on the paid Pro plan after you use up your free tier) and automatically opt out of training the AI on your contents .
Once you’ve chosen your program, create an account and go to the prompt box.
Step 2: Write your image prompt
Your best chance for success starts with a well-written prompt. Each prompt should include the basics of the characters, setting, other elements, and image dimensions. But to really refine the image you want, add descriptions of the artistic style you want to emulate, the aesthetic, mood, and specific colors.
Leonardo may not understand a prompt that asks for “cottage core coastal grandma,” but he does understand “rustic atmosphere with blue and warm light.”
AI image generators rightly have limitations on what you can generate. This is to prevent people from abusing these services to create illegal content and potentially dangerous or offensive content. Dall-E 3, for example, will do that refuse to generate images asking public figures by name, such as President Biden or Taylor Swift. Make sure you check your service’s security policies before generating.
Prompt engineering – the process of writing the right prompt to get the results you want – is something you get better at with practice. It also varies between each image generator. For example, Canva does better with simple prompts, while Dall-E encourages longer, more detailed prompts. Playing with your program is the best way to get to know it. You can also check out my complete guide to creating image prompts for more tips and tricks. Once you get your best prompt, go ahead and generate!
Step 3: Edit your images if necessary
Even with the best generator, you’ll likely need to edit your images after they’re created. Depending on the service you use, you may have access to editing tools or request changes. Be wary of over-editing your images. I’ve found that services can only handle so many changes before the image becomes too shaky and ultimately unusable.
Take this image I created in Midjourney as a warning about what happens when you edit too much. Sometimes it’s best to start over.
Post-generation editing tools are one area where AI services tend to shine or fall flat. If your service doesn’t offer them (or they don’t work very well), try editing the prompt instead of the image. In my experience, this gives you a better chance of success rather than trying to edit existing images incrementally.
Step 4: Give your images proper credit
It is extremely important to label your images as AI-created content. As AI image services improve, it is essential that all creators clearly identify the origin of an image, especially to distinguish it from human-made art and photography.
In the bottom left corner, Meta AI watermarks all its AI-generated images.
Some services automatically add a watermark or other markup indicating it’s AI-generated, but too many don’t, such as Dall-E, Adobe Firefly, Midjourney, Canva, and Leonardo. If you share your image online, make sure to include the AI service you used in the caption, credit text, and/or alt text. You can also add a disclaimer to your posts when you share them on social media. Meta can flag them as AI content on Instagram and Facebook, but adding a line to your caption will ensure there is no confusion.
Frequently asked questions and other considerations
As of publication, the use of AI image generators is legal. But there are important legal and ethical considerations to keep in mind when creating AI images. Here are a few quick frequently asked questions.
AI image services like Midjourney and Dall-E rely on existing content to help them create images. This content, whether in a database or on the open web, is typically created by humans. That’s why some creators and artists are very concerned about the way these programs are designed and trained. That has led to an uproar in the industry and a number of notable lawsuits, including an ongoing case against Stable Diffusion, an AI model used by popular generators.
If you’re concerned about using a service that could potentially abuse creators, choose a service that clearly shows how it trains its model. Some AI services rely on public databases, such as Adobe Stock and Getty Images, where creators sign agreements with the companies to include their work in those collections for specific uses. For example, Adobe Firefly says it only uses content from Adobe Stock’s public databases and images whose copyrights have expired. It also doesn’t train on your content.
As these AI services evolve, the legalities and ethics surrounding them are about to change as well. If you want to become a regular AI art maker, make sure you stay up to date with the latest requirements, which we cover in our AI Atlas guide.
