AI image generators have been brewing (generating?) up a storm for the past year. If you've been on social media, watched prime time news shows, or read a magazine, AI-generated images have been impossible to miss. They're everywhere, and it's easy to see why: the tools necessary to make them are now good—and available to the public. If you want to join in the fun, or add some AI-powered features to your business workflows, the apps on this list will give you what you're looking for.
I've been writing about AI image generators since Google Deep Dream in 2015. That's about as long as anyone realistically has been thinking about these tools, and it's really exciting for me to see how far they've come. I'm going to try to avoid the thorny discussions around artistic merit and copyright infringement in training data. Instead, I'll focus on the fact that these AI image generators can now produce fascinating results from written prompts. It's worth taking a few hours to play around with one of these text-to-image AI apps—even just so you can appreciate them from a technical perspective.
The best AI image generators
DALL·E 2 for an easy-to-use AI image generator
Midjourney for the best AI image results
DreamStudio (Stable Diffusion) for customization and control of your AI images
How does an AI image generator work?
AI image generators take a text prompt and then turn it—as best they can—into a matching image. This opens up some wild possibilities, since your prompt can be anything from "an impressionist oil painting of a Canadian man riding a moose through a forest of maple trees" to "a painting in the style of Vermeer of a large fluffy Irish wolfhound enjoying a pint of beer in a traditional pub" or "a photograph of a donkey on the moon."
Seriously, the only real limits are your imagination, the AI image generator's ability to comprehend your prompt, and any content filters put in place to stop bad actors flooding the internet with AI-generated violence or other NSFW content.
Most AI image generators work in a pretty similar way. Billions of image-text pairs are used to train a neural network (basically, a very fancy computer algorithm modeled loosely on the human brain) on what things are. By allowing it to process near-countless images, it learns what dogs, the color red, Vermeers, and everything else are. Once this is done, you have an AI that can interpret almost any prompt.
The next step is to actually render the AI-generated image. The latest generation of AI image generators do that using a process called diffusion. In essence, they start with a random field of noise and then edit it in a series of steps to match their interpretation of the prompt. It's kind of like looking up at a cloudy sky, finding a cloud that looks kind of like a dog, and then being able to snap your fingers to keep making it more and more dog-like.
What makes an AI picture generator great?
There's a reason that AI image generators have become incredibly popular over the past 12 months: before that, they were pretty bad. The technology underlying them was incredibly cool and impressive, at least to research scientists, but the images they could output were underwhelming. Even the original DALL·E was more of a fun novelty than a world-shaking revelation when it launched in 2021.
Since they're such a recent development, there isn't a huge amount of great AI image creators to choose from. The three big names—DALL·E 2, Midjourney, and Stable Diffusion—are all on this list for different reasons, but they're also the only apps that really met my criteria for inclusion for two reasons:
I was looking for apps that allowed you to generate AI images from a text prompt. Tools that have you upload a dozen of your photos and then spit out AI-generated portraits are fun (and normally built using Stable Diffusion), but they aren't the kind of general-purpose image generators I was considering.
I was looking at the AI image generators themselves, not tools built on top of them. For example, NightCafe has a great community and app, but it just enables you to use the DALL·E 2 and Stable Diffusion algorithms (as well as some of the older ones). It's worth checking out, but it doesn't meet my criteria for this list.
Aside from all that, I also considered how easy each AI image creator is to use, what kinds of controls and customization options it provides (for things like AI image upscale), what pricing model it has, and most important of all: how good were the results?
I've been using and writing about DALL·E 2, Midjourney, and Stable Diffusion for roughly the past year, so I'm pretty familiar with how they all work—and their various pros, cons, and weird behaviors. But writing this article was actually the first time I've put all three head-to-head with the same prompts. The results were fascinating, and I'm delighted to say all three offer genuine reasons to use them.
Before we dive in, one more thing to note: all these tools are technically in beta, and I suspect they'll remain that way for a while. While they're getting more and more impressive every day, AI image generators have a long way to go before they're able to consistently produce great results and reliably fit into commercial workflows.
The best AI image generator for ease of use
DALL·E 2 is arguably the biggest name in AI image generators—and with good reason. It was the first AI-powered image generator that was good enough to create wildly interesting images and was widely available to enough people to go viral. And while first doesn't necessarily mean best, DALL·E 2 has a lot going for it.
The biggest thing is that DALL·E 2 is ridiculously simple to use. Sign up for a free OpenAI account, type what you want to see, and click Generate. Within a few seconds, you'll have four AI-generated image variations to choose from. OpenAI also built ChatGPT, so if you've used that, you'll be right at home.
When you sign up, you get 50 free credits and another 15 for free every month. Each credit allows you to generate four image variations from a single prompt (whether it's a written prompt, image upload, or a request for more variations). It's a pretty generous allotment if you're just looking to play around with AI, and you can buy packs of 115 more for $15. That works out to around $0.13 per prompt or $0.0325 per image variation.
While very simple to use, DALL·E 2 has some powerful features. The image editor (which is also in beta) enables you to add additional generated frames, so you can expand an image, whether you generated it using DALL·E 2 or uploaded it. This technique, called out-painting, means you can create larger works of AI art. There's also an eraser, so you can remove bits of an image and replace them with AI-generated elements (this technique is called in-painting).
In addition to the DALL·E 2 web app, OpenAI offers an API, which allows developers to build apps that integrate with DALL·E 2. Because of that, you can connect DALL·E 2 to Zapier to do things like automatically create images from Google Forms or HubSpot responses—or any other apps you use.
Create images with DALL.E based on Slack messages and send as a channel message
Generate OpenAI (DALL.E) images from new Airtable records and send as an email
Generate images based on Google Form responses and save them in Google Sheets
Create NFTs with images in OpenAI plus metadata with One Call in idexo from new Google Forms responses
Create NFTs with images in OpenAI plus metadata with One Call in idexo from new HubSpot form submissions
Zapier is a no-code automation tool that lets you connect your apps into automated workflows, so that every person and every business can move forward at growth speed. Learn more about how it works.
DALL·E 2 pricing: 50 free credits on sign up, plus 15 free credits each month. 115 additional credits can be purchased for $15. API pricing is more complex, but starts from $0.016/image.
The AI image generator with the best results
Midjourney consistently produces my favorite results of all of the image generators on this list. The images it creates seem more coherent, with better textures and colors—and overall, the results are just more interesting and visually appealing. In particular, people and real-world objects look more lifelike and natural than they do with other AI image generators, at least without lots of prompting. It's telling that it was the first AI image generator to win an art competition.
Unfortunately, Midjourney is let down by one major quirk: the beta is only accessible through Discord. Once you join Midjourney's Discord server or invite the Midjourney bot to one you control, you can enter a prompt by typing
/imagine [whatever you want to see]. The bot will then generate four variations of your prompt, which you can then download, upscale, re-edit, and more.
By default, every image you generate is posted publicly in Midjourney's Discord. It gives everything a cool community aspect, but it means that anyone who cares to look can see what you're creating. While not necessarily a problem for artists, this might be a dealbreaker if you're looking to use Midjourney for business purposes.
If all this sounds confusing, don't worry. Midjourney's help docs are really good and walk you through getting started as well as all its advanced features, like the different model versions, upscaling your images, blending multiple images, and using different parameters to control things. Once you understand the different options, the results you can get are genuinely amazing.
Midjourney's free trial limits you to 0.4 hours of GPU time in total, or around 25 images; your images are also limited to a CC BY-NC 4.0 Creative Commons non-commercial license. After that, the Basic Plan starts at $10/month and comes with 3.3 hours of GPU time per month, or around 200 images. You also get the option to buy additional GPU time, and you can use your images commercially.
Midjourney pricing: Free for ~25 images; from $10/month for the Basic Plan that allows you to generate ~200 images/month and provides commercial usage rights.
Best AI image generator for customization and control
DreamStudio (Stable Diffusion)
Unlike DALL·E 2 and Midjourney, Stable Diffusion is open source. This means anyone with the requisite technical skills can download it and run it locally on their own computer. It also means that you can train and fine-tune the model for specific purposes. Almost all the services that use AI to generate artistic portraits, historical portraits, architectural renders, and everything else use Stable Diffusion this way. If you've got the chops and want to build something awesome with AI, Stable Diffusion is the best way to do it right now. There's even an API.
But Stable Diffusion is also available in a powerful public app called DreamStudio, built by its developers, Stability AI. (Like everything on this list, it's in beta.)
DreamStudio gives you a huge amount of control over the various aspects of generating an image with AI. When you type in your prompt, there are sliders that allow you to determine how large the final image is, how closely it matches the prompt you give it, how many steps the diffusion model takes, and how many images are generated. You can also select what version of the algorithm it uses, and even enter a specific seed and sampling method so that you get repeatable results (otherwise, they're randomly generated). DreamStudio also has in-painting and out-painting, though you need to use Chrome to access them.
DreamStudio works on a credit system. When you sign up, you get 100 free credits, which are good for 500 images with the default settings. Using a more powerful model, generating larger images, or iterating them through more steps will all use up your credits faster. Once you're done, you'll need to buy more, starting at $10 for 1,000 credits.
All in all, DreamStudio and Stable Diffusion give you the most customization and control over the whole AI image generation process. They enable you to go as deep into AI as you want to go—and even build your own AI services.
DreamStudio pricing: Free for 100 credits. From $10 for 1,000 credits.
Expect things to change fast when it comes to AI-generated images
AI image generating and art is a rapidly evolving space. It's incredible to see how far the different engines have come over the space of a year. With hundreds of thousands of people now using them, the developers are getting huge amounts of data to train and refine their models more, so we can expect things to continue to improve.
It's also likely that we'll soon see some new image generators get released. Google hasn't yet made Imagen publicly available, and Meta hasn't released anything based on its Make-A-Scene algorithms to the public—exciting times are ahead.