DALL-E 2 and more fun apps

Unless you’ve been hiding under a rock for the last year, the chances are you’ve heard about what’s happening with AI. Artificial Intelligence chatbots like ChatGPT and Google Bard are wowing people with the possible potential of what AI can do for us.

Alongside these text-based systems though are several AI image-generation tools that have appeared and caused quite a stir because of what they can do. The technology has come on in leaps and bounds in a very short space of time and suddenly there are image generators that can create images that look like a human made them.

What tools are available though? We’re rounding up the best we’ve seen and the ones worth experimenting with.

Stable Diffusion

Stable Diffusion is a fantastic AI image generation tool that’s free to use. It’s being used to create some really impressive images.

We’ve previously shown off the sorts of things being created by it and those include everything from cartoon artwork to impressive life-like photos of fake animals.

This is a text-based tool which is able to create photo-realistic images based on your own text prompts. As with other AI tools, the better the prompts, the better the images. Fortunately, Stable Diffusion gives you some example prompts to get you started. There’s also a subreddit dedicated to Stable Diffusion where you can get some ideas of just what’s possible.

Chad Nelson/DALL·E

DALL-E 2

OpenAI’s ChatGPT is perhaps one of the most well-known AI tools to appear on the scene. That same organisation also has an AI imager generator and that’s known as DALL-E 2.

Like other AI image generators, DALL-E 2 is creating some seriously impressive images with just text input.

This AI image generator was created by training a neural network with images that already had image descriptions. Over time this training only gets better as the system sees and learns about more images. Deep learning is key to the success of these tools and the images they create.

DALL-E 2 does have a cost to it – you have to buy credits – but it is one of the better tools available.

Microsoft/OpenAI

Bing Image Creator

Did you know that Bing AI and Microsoft Copilot are based on OpenAI technology?

Armed with that knowledge you’ll be unsurprised to learn that Bing Image Creator is another interesting image generator tool. It’s powered by DALL-E and is easy to access. In fact, you can even access Bing’s image generators through Bing Chat. So as long as you have a Microsoft account you can quickly create images with text prompts. It’s in preview mode at the moment but is open to everyone to use for free and it’s pretty fast as well.

Midjourney 

Midjourney

Midjourney is one of the most impressive AI image-generation tools we’ve seen. It’s able to create some absolutely stunning images that could easily be mistaken for being crafted by real people.It’s unusual though because you need to join the beta Discord server to access it properly, and you also need to pay to make the most of it, but if you want the best then it’s worth investing in.

Nvidia Canvas

Most of the AI image tools featured in this article use a mix of text prompts to generate the images. If you want a more hands-on experience though Nvidia Canvas is slightly different. This is a tool that takes your digital brushstrokes and uses those along with AI to create life-like images. Just like the other tools it has been trained with generative adversarial networks and Canvas has seen well over 5 million images.

Unlike the other tools, it’s limited more to landscape-style images, but you can use tools to paint all sorts of things including sky, clouds, grass, hills, stone, sea, sand, rivers, fog and more. We’ve tried it out and shown what’s possible. It’s impressive and well worth dabbling with.

Craiyon 

Craiyon

• Try the AI image generator: Craiyon

Craiyon is not quite as fancy as the other AI generators we’ve listed, but it is open source and free to use at a base level with pro plans available. It’s also fast and gives you a lot of different variations with each prompt that you use. You might have seen some of the images Craiyon has created appearing online in meme form as they’re often pretty hilarious, but it also is highly capable and can create some impressive works of art.

Google Magic Editor

With the popularity of these AI image generators, it’s only a matter of time before we see more and more apps appearing too.

At Google I/O the company showed off Magic Editor which is set to be part of Google Photos and will use generative AI technology to make complex edits to your photos. This includes all sorts of things like changing colours, repositioning subjects, deleting unwanted objects and more. Think of this as a more powerful version of Magic Eraser, but with much more in the way of AI capabilities.

How to use AI image generators

There are clearly a few different image generators to choose from. Some of them are free and others require a payment for credits. Most of the AI image tools work in the same way though. They require text prompts in order to generate images. These prompts are essentially logical descriptions of the image you want to create. The more detailed you are the better, but you can also use both positive and negative prompts. Use terms to include or exclude elements you do or don’t want in the images and use descriptive text to influence the design.

For example, you can use terms like “high-resolution photo” or “black and white image” or “pop style artwork” to change the look of the images.

If you find that you’re not getting the results you want then it’s worth researching how prompts work and how you can change the wording of your prompts to improve the wording to get better results.

Ethical considerations of AI image generators

There are many concerns about how AI image generators work and the images they create. Especially the impact it will have on creators and artists. Also, the copyright implications of these tools are constantly being discussed.

If you’re using them to create images and artworks, who owns the copyright? You or the company behind the AI? And what if you use other people’s work to influence the AI? It is already trained on image models in the public domain. So there are a lot of considerations.

Beyond these discussions, there are often rules in place with the various AI tools. You often aren’t able to use the AI tools to create images of celebrities, for example. This is to prevent people using AI tools to create Deep Fake content.