AI image generators are brewing (generating?) Since almost two years, AI image generators have been brewing (generating?) AI-generated pictures are everywhere. Whether you were on social media or watching primetime news, or reading a magazine, they're hard to miss. It's no wonder they're everywhere: the tools to create them are good and available to everyone. These apps will help you join the fun or add AI-powered features into your business workflows.

Since Google Deep Dream, in 2015, I have been writing about AI-based image generators. This is the longest anyone outside a computer lab has been thinking about AI image generators, and I am really excited about how far they have come.

At least in the places I can, I will try to avoid the difficult discussions about artistic merit, copyright violations and whether these tools replace or augment artists. Instead, I will focus on the fact these AI image creators can produce fascinating results based on a variety of text prompts.

Even if you just want to see how they work from a technical standpoint, it's worth spending a few minutes playing around with these AI text-to image apps. You're going to see a lot of their work in the months and years ahead, whether you like it.

The best AI image creators

DALL*E 3, an AI image generator that is easy to use

The best AI images results are midway through the journey

Stable diffusion for customizing and controlling your AI images

Adobe Firefly is a tool for integrating AI generated images into photographs

Getty offers Generative AI for commercially-safe images.

How does AI image generation work?

These AI image generators all take a text prompt, and turn it into a matching picture as best they can. The possibilities are endless, as your prompt could be anything: "an impressionist painting of a Canadian riding a moose in a forest of Maple trees", "a Vermeer-style painting of an Irish wolfhound drinking beer in a traditional Irish pub" or even "a photo of a donkey in space."

The only limits to your creativity are your imagination and the AI image creator's ability understand your prompt. Also, any content filters that have been put in place by the content provider to prevent plagiarism, copyright infringement and bad actors from flooding the web with AI-generated violence and other NSFW material. The Vermeer prompt was reliable, but now some image generators block it as it is a named artist.

The majority of AI image generators operate in a similar manner. The neural network, which is a fancy computer algorithm loosely based on the brain, is trained by millions or billions image-text pairs. It learns to recognize dogs, Vermeers and the color red by processing countless images. After this, your AI can understand almost any prompt. However, it takes a certain amount of skill to set things up correctly.

Next, the AI generated image must be rendered. This is done by the latest generation of AI-generated images using a diffusion process. They start out with a random noise field and then edit it to match the prompt. This is like looking at the sky and finding a cloud that resembles a dog. Then you can snap your fingers and make it look more dog-like.

Before we get started, I do not want to oversell anything. These text-to image generators are impressive but won't save you from having to take product photos again. They can be very helpful if you need a few weird or unusual images. If you need something very specific, it's better to hire a photographer or license the image. It can also take much longer to create a header for a blog than to find a header on a stock image site. It won't look as customized, but it is more likely that the model will have ten finger.

What is the best AI image creator?

Before that, AI image generators were not very good. Although the technology behind them was impressive and cool, their output images were not. When it was launched in 2021, the original DALL*E lacked any real impact. It was a novelty that people enjoyed.

There aren't many AI image creators available, as they're so new and require such high computing power. There are three major names: DALL*E 3, Midjourney and Stable Diffusion, as well as models by AI powerhouses such Google, Meta Adobe and NVIDIA. This is partly due to the strict criteria that I set.

I was searching for apps that would allow you to create AI images using a text prompt. The tools that let you upload photos, then generate AI-generated images are cool (and are usually built using Stable Diffusion), however they weren't what I was looking for.

I looked at AI image generators, not the tools that were built on top. NightCafe, for example, is an AI image generator with a wonderful community and app. However, it only allows you to use Stable Diffusion or fine-tuned versions based on it, DALL*E 3's API, and a few older generative models. Although it's worth checking, this app doesn't fit my criteria.

In addition to all of that, I also looked at how easy it was to use each AI image creator, what kind of controls and customization it offered (for things such as AI image upscale), the pricing model, and, most importantly, how good the results were.

Since the launch of DALL*E, I have been writing and using text-to-image generators. I also write about photography and arts for over a decade. I am familiar with all these tools, their pros and cons, and strange behaviors. This was the first time that I had compared so many AI image creators with the same prompts. The results were amazing, and all of the apps in the list have real reasons to be used.

Be aware that all of these tools are still in beta and will likely remain in this state for some time. The AI image generators are improving every day but they still have a way to go to be able to produce consistently great results that fit reliably into commercial workflows.

These tools are not free because they're in beta. Since I wrote this article both DALL*E & Midjourney have discontinued their free trials. Although it's understandable given the high computing load required to create AI images, they are still more difficult to play with.

DALL·E 3

Text-to-image AI allows you to create the best AI images.

AI image generators are brewing (generating?) Since almost two years, AI image generators have been brewing (generating?) AI-generated pictures are everywhere. Whether you were on social media or watching primetime news, or reading a magazine, they're hard to miss. It's no wonder they're everywhere: the tools to create them are good and available to everyone. These apps will help you join the fun or add AI-powered features into your business workflows.

Since Google Deep Dream, in 2015, I have been writing about AI-based image generators. This is the longest anyone outside a computer lab has been thinking about AI image generators, and I am really excited about how far they have come.

At least in the places I can, I will try to avoid the difficult discussions about artistic merit, copyright violations and whether these tools replace or augment artists. Instead, I will focus on the fact these AI image creators can produce fascinating results based on a variety of text prompts.

Even if you just want to see how they work from a technical standpoint, it's worth spending a few minutes playing around with these AI text-to image apps. You're going to see a lot of their work in the months and years ahead, whether you like it.

The best AI image creators

DALL*E 3, an AI image generator that is easy to use

The best AI images results are midway through the journey

Stable diffusion for customizing and controlling your AI images

Adobe Firefly is a tool for integrating AI generated images into photographs

Getty offers Generative AI for commercially-safe images.

How does AI image generation work?

These AI image generators all take a text prompt, and turn it into a matching picture as best they can. The possibilities are endless, as your prompt could be anything: "an impressionist painting of a Canadian riding a moose in a forest of Maple trees", "a Vermeer-style painting of an Irish wolfhound drinking beer in a traditional Irish pub" or even "a photo of a donkey in space."

The only limits to your creativity are your imagination and the AI image creator's ability understand your prompt. Also, any content filters that have been put in place by the content provider to prevent plagiarism, copyright infringement and bad actors from flooding the web with AI-generated violence and other NSFW material. The Vermeer prompt was reliable, but now some image generators block it as it is a named artist.

The majority of AI image generators operate in a similar manner. The neural network, which is a fancy computer algorithm loosely based on the brain, is trained by millions or billions image-text pairs. It learns to recognize dogs, Vermeers and the color red by processing countless images. After this, your AI can understand almost any prompt. However, it takes a certain amount of skill to set things up correctly.

Next, the AI generated image must be rendered. This is done by the latest generation of AI-generated images using a diffusion process. They start out with a random noise field and then edit it to match the prompt. This is like looking at the sky and finding a cloud that resembles a dog. Then you can snap your fingers and make it look more dog-like.

Before we get started, I do not want to oversell anything. These text-to image generators are impressive but won't save you from having to take product photos again. They can be very helpful if you need a few weird or unusual images. If you need something very specific, it's better to hire a photographer or license the image. It can also take much longer to create a header for a blog than to find a header on a stock image site. It won't look as customized, but it is more likely that the model will have ten finger.

What is the best AI image creator?

Before that, AI image generators were not very good. Although the technology behind them was impressive and cool, their output images were not. When it was launched in 2021, the original DALL*E lacked any real impact. It was a novelty that people enjoyed.

There aren't many AI image creators available, as they're so new and require such high computing power. There are three major names: DALL*E 3, Midjourney and Stable Diffusion, as well as models by AI powerhouses such Google, Meta Adobe and NVIDIA. This is partly due to the strict criteria that I set.

I was searching for apps that would allow you to create AI images using a text prompt. The tools that let you upload photos, then generate AI-generated images are cool (and are usually built using Stable Diffusion), however they weren't what I was looking for.

I looked at AI image generators, not the tools that were built on top. NightCafe, for example, is an AI image generator with a wonderful community and app. However, it only allows you to use Stable Diffusion or fine-tuned versions based on it, DALL*E 3's API, and a few older generative models. Although it's worth checking, this app doesn't fit my criteria.

In addition to all of that, I also looked at how easy it was to use each AI image creator, what kind of controls and customization it offered (for things such as AI image upscale), the pricing model, and, most importantly, how good the results were.

Since the launch of DALL*E, I have been writing and using text-to-image generators. I also write about photography and arts for over a decade. I am familiar with all these tools, their pros and cons, and strange behaviors. This was the first time that I had compared so many AI image creators with the same prompts. The results were amazing, and all of the apps in the list have real reasons to be used.

Be aware that all of these tools are still in beta and will likely remain in this state for some time. The AI image generators are improving every day but they still have a way to go to be able to produce consistently great results that fit reliably into commercial workflows.

 

These tools are not free because they're in beta. Since I wrote this article both DALL*E & Midjourney have discontinued their free trials. Although it's understandable given the high computing load required to create AI images, they are still more difficult to play with.

The AI image generator with the best results

Midjourney

Midjourney pros

AI images that are consistently the best-looking

Community is a wonderful way to find inspiration

Midjourney cons:

It is strange that Discord can only be used to communicate.

Images generated by you are made public by default

The free trials have been suspended.

 

Midjourney consistently produces the best results out of all the image generators in this list. It produces images that are more appealing and interesting. They have better textures, colors, and overall, they seem to be more coherent. People and real-world items look more natural and lifelike than with other AI image creators. The latest versions are even able to get the hands right. It is telling that this was the first AI-generated image to win an arts competition.

Midjourney's beta can only be accessed through Discord. You can create a prompt using /imagine (whatever you wish to see) once you have joined Midjourney's Discord or invited the Midjourney bot onto a server you control. The bot will generate four variations of the prompt you entered. You can download, upscale or re-edit them as desired.

This is about to change. Midjourney has begun alpha testing a web application. The company has said that it will continue to support Discord in the long-term. There are also bound to be some other quirks, but it is great that this best AI image creator will soon be available more widely.

By default, all images you create are posted in Midjourney Discord. They can also be seen on your profile page on the website. This gives your work a community feel, and anyone can view it. This is not a problem for the artists but could be a deal breaker for those who want to use Midjourney as a business tool.

Don't worry if this all sounds confusing. Midjourney has excellent help documents that will guide you through the process of getting started, as well as its advanced features. These include upscaling images, blending images, and controlling things with different parameters. The results are truly amazing once you learn how to use the options.

Midjourney has suspended its free trials due to the high number of users trying to access it. However, they are occasionally reinstated for several days. If you missed the free trial, then you can get the Basic Plan for $10/month, which includes 3.3 hours per month of GPU time, or approximately 200 images. You can also buy more GPU time and use your images for commercial purposes.

Pricing for Midjourney: Starting at $10/month, the Basic Plan allows you to create 200 images/month with commercial usage rights.

Best AI image generator for customization and control

Stable Diffusion

Stable Diffusion pros:

DreamStudio, the web-based AI image generator from Stable Diffusion, is currently the only major AI picture maker that offers free credits.

This app is extremely powerful and customizable. It also has great results.

 

Stable Diffusion Cons:

Learning curve is steeper than with some other AI art creators

DreamStudio's editing tools aren't intuitive.

 

Stable Diffusion, unlike DALL*E or Midjourney is open-source. Anyone with the required technical skills can download and run it on their computer. You can also train and fine tune the model to suit specific purposes. This is how Stable Diffusion is used by almost all services that generate AI-generated artistic portraits and historical portraits. Stable Diffusion offers the best solution for anyone who wants to create something amazing with AI. Clipdrop is an API that allows you to break it down into different tools.

Stability AI, the developers of Stable Diffusion, have also built a powerful app for public use called DreamStudio. It's still in beta, like everything else on this list.

DreamStudio allows you to have a great deal of control over how an AI-generated image is generated. You can use sliders to control the size of the final image, the degree it matches your prompt, the number of steps that the diffusion model requires, and the number images generated. You can choose the version of algorithm (the most recent is SDXL1.0) and enter a seed to get repeatable results. DreamStudio has also inpainting or outpainting. I haven't had any success with these features.

DreamStudio uses a credit-based system. Sign up for DreamStudio and you'll get 25 credits free. These are enough to create around 30 prompts, or 120 images, using the default settings. Credits will be used up faster if you use a powerful model, generate larger images or go through more steps. You'll have to purchase more credits once you're finished. Prices start at $10 for 1000 credits.

Stable Diffusion and DreamStudio give you complete control and customization over the AI image creation process. You can go as far into AI as you like, and even create your own AI services.

If you run out of time with your free trial, you can try the Stable Diffusion model for free via Clipdrop. However, they will be watermarked and you'll have less control.

 

DreamStudio pricing is free for the first 25 credits. Starting at $10 for 1,000 credits.