As the controversial role of artificial intelligence grows in society, so too is its application in the arts and mainstream media. In fact, the best AI generators of 2023 can create surprisingly detailed and accurate pieces that you can pass off as real (well, except the hands).
These artworks can often be created with surprising ease from the user’s side – many of the best AI art generators only require a simple ChatGPT-style worded prompt to spawn a painting, sketch, or apparently life-like photograph.
The technology has come a long way from the viral Dall-E mini meme machine of 2022.
Big players in the world of tech have now plunged their flag into the new realm, with Adobe Firefly recently announcing Google Bard integration to help create AI-generated art.
So, what’s the best AI art generator? Below, we’ve rounded up the best available in 2023, considering cost, results, usability and adaptability. And, to show what each is capable of, we had each AI generator create an image of a bear fighting a dragon.
Don’t say we do nothing for you.
Best AI art generators in 2023
Specs | Developers: Midjourney (Discord) | Price: $10 (£8) a month+ | Platform: Discord | midjourney.com
- High-quality, photo-realistic images
- The Discord community can help you create better prompts, for better images
- Tricky interface compared to other AI art generators
- Cost of subscription
This Discord-run AI art generator is capable of producing stunning photorealistic results. Its photorealistic accuracy is extremely impressive – but comes at a price. The basic plan for Midjourney (as a year-long subscription) comes in at $10 (£8) a month, with the most expensive pro plan setting you back $48 (£38) a month.
As its use is based entirely within Discord, it can be a little tricky to learn how to use its interface. The software specialises in gothic-like images suitable for bringing fantasy or science-fiction literature to life.
Specs | Developer: OpenAI | Price: Free for those who created an account before 6 April 2023, $15 (£12) minimum cost for anyone else for 115 credits | Platform: browser-based or API | openai.com/dall-e-2
- 15 free credits every month
- Simple to use
- Can generate good detail in a variety of different styles
- Cost of extra credits
- Lack of negative prompts (meaning you can’t list the elements you don’t want to appear)
- Limited resolution options
Perhaps the most well-known AI art generator of the lot, DALL-E 2 comes from OpenAI, the same people who brought you ChatGPT. This image generator is simple to use and is great for those looking to expand their AI art skills after tinkering with more basic platforms, such as Craiyon.
Its outpainting tool is pretty nifty, too, which expands the backgrounds of pre-existing art and photos.
You’ll get 15 credits to spend in DALL-E 2 each month, allowing you to enter 15 different prompts. As the service generates four photos from each prompt, that’s 60 free images a month.
- Great amount of editing tools with high-quality results
- Able to generate different styles of text and colour vectors
- Impressive generative fill tool
- Included as part of Adobe Creative Cloud
- Cost of Adobe Creative Cloud subscription
- AI is trained on a limited pool of images
Adobe has been around for years with its Photoshop software synonymous with graphic design and image editing. Now, the company has entered into the AI image generator race with Firefly.
Firefly is included as part of the full Adobe Creative Cloud All Apps package but you’ll need to request access to use its beta service.
If you’re unsure of the ethics of AI art (more on that below), Firefly only uses Adobe stock and public domain content, with Adobe promising compensation to the original artists. Its generative fill tool is impressive stuff, working like the outpainting of DALL-E 2 above.
Specs | Developer: Stability AI | Price: Free trial on browser (pay for credits to carry on using), open-source API | Platform: Browser-based (DreamStudio) or API | DreamStudio.ai
- Creates high-quality AI-generated images quickly in a variety of styles
- Open source code is available to download and run if your PC is powerful enough
- Free trial on DreamStudio website if you want to give it a go
- Difficult open-source installation
- Not a free service
Stable Diffusion is a tricky AI image tool to understand. You can head over to DreamStudio to give it a go, with a free trial allowing you to generate a fixed number of images before you need to purchase credits to carry on using the thing ($10/£8 buys you 1,000 credits, which is around 5,000 images).
You can also use the generator for free thanks to it being open-source. Going this route will require plenty of technical know-how, however. You’ll need to have Python up and running and download a host of different software.
Dream by wombo
Specs | Developer: wombo | Price: Free limited access | Platform: App and browser-based | Dream.ai
- Good choice of style templates
- Quick image processing
- The free version gives better results than other platforms (in vertical aspect ratio)
- High cost of subscription for full access
- Just one image per prompt on paid version (four on premium plan)
You can pay to use Dream by wombo, with subscriptions of $10/£8 a month, $90/£70 a year, and $170/£133 for full access forever. Paying for Dream subscriptions allows you to generate four images and create them in a wider variety of styles over its free version.
More like this
Free use of Dream is unlimited in terms of how many prompts you can input but your options are more limited in terms of image style and the service will only generate one image instead of four. Generating images on Dream is a quick process and you’ll get some nifty phone backgrounds out of it on the free version of the site.
Specs | Developer: Craiyon LLC | Price: Free | Platform: Browser-based | Craiyon.com
- Free – can be used as often as you like
- Easy to use
- Waiting time for generated images can be long
- Image distortion is commonplace
Formerly known as DALL-E mini, Craiyon is the best AI art generator that’s completely free. You don’t even need to create an account to use it. You can prompt it as often as you like, too, but it does get busy and its creation wait time can be long.
As a free service, its generated images usually come with some level of distortion. This can make for some horrifying creations, but those unlimited free prompts allow you to fine-tune your AI-generated images as much as you can.
How do the best AI art generators work?
AI art generators work by employing learned data. Developers of the models feed countless images into them, with which the artificial intelligence identifies trends and characteristics. Essentially, they scour sites similar to Google Images for images to base your prompts on, using what it’s learnt to give you what you’re asking for.
This does mean there’s a question mark as to how unique generated images are. The generated art will take aspects of paintings, drawings, and photos that have been created by real people and apply them to its generated image. Adobe Firefly’s use of stock images and public domain image learning, however, should allay some of those fears about creative use and ownership.