Image generators work like text generators do — only instead of text, they spit out an image/several images in response to your prompt. Some generators will also gladly accept reference images from you to help create what you need — most notably, Adobe’s Firefly.
AI image generators are trained on a large dataset. They sponge up every bit of info fed to them — via the internet or a photo bank — to understand what’s what: what a dog looks like, who Mickey Mouse is, and how to create photography in Johannes Vermeer’s style.
Some image generators work better on reference pictures, some can add characters like Goofy, and others draw from a licensed picture bank, so you don’t have to worry about copyright infringement.
And, of course, the quality may vary greatly. Paid subscriptions with big companies backing the AI image tools will generally offer better results than free tools created from an open-source code.
There are plenty of ways you can use AI in email marketing, so you have to understand what goal you are trying to accomplish when choosing an AI image generator. Keep in mind:
There is another important difference worth mentioning. Image generation can be tricky from a legal perspective. Some AI image generators may refuse certain tasks due to copyright restrictions. Others might not warn you about such limitations, so proceed with care and check the “Terms of Use” page. When working with AI, this should always be your first step.
“Best” is a relative notion, of course. We’ve looked at the most popular — in terms of monthly site visits (all of them clock over 1M) on Similarweb and some options we thought worth mentioning because of the specific benefits they offer.
Finally, those tools we had the chance to test personally, we asked to do the exact same thing: produce an image of Winnie the Pooh and Piglet, in pencil, based on Alan Milne’s book. The results were hit-and-miss, but the prompt was the same — which serves to highlight the difference between various AI image generators. Here’s how Winnie and Piglet looked in Milne’s book.
With all that settled, let’s look at the standout generators on the market right now.
DALL-E 3 is an updated version of DALL-E 2, released because competing AI image-generating software has started to catch up. DALL-E belongs to OpenAI — the same company that developed ChatGPT — and this simplifies things for users.
You don’t need a separate account to use DALL-E, just a ChatGPT one. You can generate images and tweak them just like you would generate and tweak text: simply input your prompt into ChatGPT and watch it produce results.
And the results are stunning. Just look at this ink drawing of a hedgehog eating a watermelon.
Being part of ChatGPT, DALL-E is easy to integrate into your workflow, meaning you can rely on it daily. However, there’s a downside: you can’t use DALL-E as a standalone service from OpenAI. You’ll have to buy a ChatGPT Plus or better subscription (from $20 per month and more). Unless you need the rest of ChatGPT with DALL-E, you might want to look at other options.
But wait, is that it? No legal way to try DALL-E without paying? Why, yes, we are glad you’ve asked! You can try it as a standalone service with Microsoft Bing or Microsoft Designer since it is built on DALL-E 3 (more on that below).
Leonardo AI’s main competitive edge over other tools is that you can create characters, including those that resemble real-life people. Many other image generators implemented guardrails, and for a good reason: so that people wouldn’t create fake images that might catch on. But there are much more harmless cases — for example, where generating a Disney character for a birthday card would be an asset to you.
We went ahead and input a prompt for Piglet and Winnie the Pooh — in pencil, based on Milne’s book. Here are some of the better-looking results. The characters, though, look more like Disney versions.
While there are no restrictions on creating characters, there are guardrails on violence: Leonardo refused to create an image of Piglet firing a gun, depicting the scene from the cartoon where he shoots the balloon.
Another interesting feature is the recently-added Flow State: a tool that creates a series of images after a simple prompt and lets you choose which ones you like best so that it can further iterate. In an attempt to generate more interest, Leonardo made Flow State iteration dirt-cheap: just one token, whereas creating a new image from scratch costs 40 tokens.
Speaking of, the token is Leonardo’s currency — the amount you need depends on task complexity. For example, the free plan gives you 150 tokens — you can create three images with that limit. The first paid plan — the Apprentice — is $10 per month, and it hands you 5,500 tokens, which equals to a little fewer than 140 images.
But still, Leonardo looks like a thoroughly professional tool. You can input a short and easy prompt, but then AI works on that — the resulting prompt is 3-4 sentences long and highly detailed. Leonardo creates 4 images per prompt, which you can then use for the Flow State — or edit to your heart’s content. And there are many, many ways you can edit these images — you’ll likely need a professional designer to make sense of all the functionality on offer.
One of the most popular AI image-creating software, Midjourney, is so good that it won an art competition back in 2022. Or rather, the artist behind the prompt did. Yes, as of fall 2024, he has been suing for copyright protection of his generated art.
While the quality of output is great, there are two potential problems you might face:
Until recently, there was a third problem: Midjourney was only available through Discord. The company has launched a web app, but some of the functionality didn’t make the jump from Discord to web app — that’s why it’s important to mention it.
The fourth problem with Midjourney is the lack of a free plan. There are four paid options, though, and a paid subscription gets you commercial rights to the images you produce. All plans revolve around Graphics Processing Units (GPU), which is the time AI uses to generate an image based on your prompt. Fast GPU times are limited (from 3 hours to 60 hours a month), while relaxed GPU time is unlimited.
The good thing about Microsoft Designer is that it lets you access DALL-E 3 without paying the subscription fee for ChatGPT. The tool itself is also in the same place as Copilot — or Microsoft Designer web app — which means you can ask Copilot to draw up an image for you. All you need is a Microsoft account.
Microsoft Designer gives you 15 boosts to use daily. One prompt equals one boost, so you can generate 15 images per day. It’s actually 60 — because much like Leonardo, Microsoft Designer gives you 4 images per prompt.
However, 15 daily boosts is a lot — you will probably never need a paid subscription with such an amount. If you do, you can sign up for Microsoft Copilot Pro for $20 per user/month and receive 100 boosts instead of 15.
We asked a Microsoft Designer to do what Leonardo did: a pencil drawing of Piglet and Winnie the Pooh. It came up with incredibly heartwarming imagery but didn’t quite nail the book version, recreating the Disney cartoon instead.
We have to mention multiple issues some users reported, though. One of them is accessing the service between 1 P.M. and 9 P.M. EST, when the servers seem overloaded. Others report that sometimes it takes a minute to generate an image — which is still not bad, all things considered. There is also a problem with certain features, like generating GIFs, for example. If GIFs in emails are a must for you, you might want to look for another service.
Adobe Firefly is part of the Adobe package. You can access it through Adobe Express or the web app, but Firefly shines brightest when you use it via Adobe Photoshop. This is not surprising: Photoshop has been an industry standard for years.
There are several features that make Firefly an interesting option:
Adobe Firefly also works off of Adobe’s stock photos — which means images you create can be used commercially without breaking copyright laws. However, there’s a flip side to this coin: you can’t create characters using it. It didn’t produce a coherent result when we asked it to create a pencil drawing of Winnie the Pooh and Piglet. But the images it can create look great — albeit we found there’s often a futuristic spin on those.
The pricing plan revolves around “generative credits” — again, think of it as a currency. You can’t buy pictures with a certain amount, though, since that depends on task complexity or AI’s amount of effort. While it’s unclear how many images a certain number of credits buys you, there’s a free plan with “limited generative credits”, and the two standard ones cost $10 and $30 per month, respectively, giving you 2,000 and 7000 credits to toy with.
The best thing about Generative AI is that it’s trained on the vast library of Getty Stock Images. It makes the images safe to use commercially, plus the artists of the original photos you use to create new ones are compensated.
Generative AI gets you four images per prompt. As well as inputting text prompts, you can upload a reference image for the AI to work with — a recent addition from Getty. This AI generator can also incorporate the color palette and mood you want and offers you a choice of lens types and depth settings. Here are some of the images users created with it, as advertised on Getty’s website.
While there is no free plan, you can buy a certain amount of generations for one-time payments without struggling to calculate some in-house currency or time limits. There are only two options, though: either 25 generations for $50 or 100 generations for $150. Both come with automatic legal protection of up to $50,000 per image. And that’s what is fascinating about Generative AI by Getty: few bonus features listed first for the custom plan are not about the unlimited pictures you’ll get but all the legal safeguards.
We couldn’t not mention it. Google is the company behind Imagen 3’s AI image generator, which is considered one of the best around. Google was far from the first on the scene in this regard, but the end result is well worth it.
Imagen 3 is completely free (you only need a Google account), generates one quality image per prompt (only in 1:1 ratio, sadly), and you can easily access it through Gemini, Google’s AI assistant, as well as ImageFX.
We asked Imagen for the same thing as Leonardo and Microsoft Designer: a pencil drawing of Piglet and Winnie the Pooh. The end result was not nearly as kind to characters as the one from Microsoft Designer — AI also added text for some reason, though we didn’t ask for it, and struggled with it. And it looks once again like an image from a cartoon, not the book.
Thus, Imagen might let you down in some cases and tweak your prompt. Other than that, the picture is nice.
A lot of AI image creators still struggle with adding text to images. That’s definitely one flaw to address. Another is drawing complex objects — like hands and eyes. We tested out Imagen 3 for a pic of Jerry the Mouse behind bars: the result was so creepy we decided not to even add it. Suffice to say, Jerry had two sets of pupils and four hands, in addition to two legs. Imagine how it would look with generated portraits!
Then, of course, there’s the balancing act between being able to use characters — or people — in generated images without users creating such pics to spread misinformation or draw up adult imagery/videos.
From the functional perspective, it would be interesting to see whether AI models will be able to create 3D images. Stability AI — the creators of Stable Diffusion and FLUX.1 — are already working on generating 3D videos from one-dimensional images. OpenAI, Google, Microsoft, and Adobe can’t be too far behind.
There are a bunch of AI image generators on offer, both free and paid, that can get the job done for you. They offer different functionality, guardrails, image license, and output of different quality.
It all depends on your needs and intended use. Adobe Firefly and Generative AI are best if you don’t want to worry about breaching copyright (or use Photoshop a lot, in the former case). Microsoft Designer offers free access to DALL-E 3 and 15 free daily prompts.
As for DALL-E 3 by OpenAI and Midjourney, well, you’ll have to pay for a monthly/yearly subscription — and get the ChatGPT in the former case, whether you need it or not. However, these two, along with Leonardo AI, are the best options on the market in terms of quality.
Easter Emails 2025: Egg-cellent Ideas for Email Marketing Campaigns and Subject Lines
Read and Enjoy: 50+ Best Newsletters To Subscribe To
AI-Based Email Marketing: What You Need to Know
Underconsumption and Marketing: How Brands Can Adapt and Still Sell
Top 10 Email Marketing Software for Small Businesses That Will Save You Time and Costs
100+ Free Helpful Email Marketing Resources Every Marketer Will Find Useful: From Tools to Communities
Check your email — the guide is on it’s way to your inbox.
