We’re getting to a point where it’s hard to tell what’s a real photograph and what’s been AI-generated. It’s scary on one level because now it’s far easier for the CIA to create more fake mom-and-pop Facebook profiles. On the other hand, the art that is being created from text prompts is wildly impressive. This technology has come a long way from when we first saw these images posted back in 2021.
Of course, everyone has their favourite generators (mine is Midjourney) but whatever startup you back at this point, only a few are going to emerge victorious. And those will probably be the ones that can generate high-quality images in various styles (including hyper-realistic) and those that can be edited and moulded to further suit the user. So let’s take a look at the most promising AI generators of 2023.
1. Adobe Firefly
I chose to start with Firefly not because I think it’s the ‘best’ image generator but because it has a bunch of useful features to enhance image creation that are completely new and unique. And when partnered with other Adobe products, Firefly is going to increase productivity tenfold for designers.
This thread does a great job of summarising the new Firefly features.
One of the biggest criticisms of AI image generators has been the fact that they are taking inspiration from already existing images across the web. So basically using other artists, photographers, and designers' work as sources to create similar images. That’s an ethical grey area yet to be resolved. What’s cool about Firefly is the image generator has been trained on Adobe stock images. So images have been brought and paid for and Adobe has plans to create a compensation model for creators once Firefly exits beta.
While most AI image generators have text-to-image, Adobe Firefly takes it one step further with text-to-vector and text-to-pattern. With access to things like editable vectors, we can work around some of the limitations of image generators.
Firefly’s main features:
- Text to vector
- 3D to images
- Sketch to image
- Text to template
- Text effects
Firefly pricing: Abode has yet to define a pricing model for Firefly.
It makes sense to follow Firefly with Canva’s new generative AI. To be clear, while Canva has created an image generator, it’s not in-house technology. What they’ve done is use ChatGPT and Stable Diffusion as the foundation for their new text-to-image tool. So technically this is Stable Diffusion technology with a twist. Similar to what Adobe’s done, Canva has also created a host of photo editing features that go beyond simply prompting an image generator.
Like it or not, this is going to be far more practical for the everyday office worker compared to straight-up text-to-image generators like Stable Diffusion or Midjourney. Why do I say that? If editing features are built into the product, users are more likely to get the results they want (even if it requires a bit of tweaking).
Along with all the other editing features Canva already offered, you can now use something called Magic Edit which allows you to replace something in the image. For example, you might have generated an image of a person holding a rose, you can use Magic Edit to replace that rose with something else like a pencil with a simple text prompt.
Canva’s main features:
- Magic design
- Draw your ideas
- Magic edit
Canva pricing: Image generation and most Magic design features are FREE. The Pro version is A$165 (per year), and this unlocks some additional editing features.
Midjourney is currently blowing everyone’s minds with the latest version, V5. Maybe you’ve seen Drippy pope in NYC? Yeah, that’s Midjourney.
It’s been stated multiple times that DALL-E has the more realistic images while Midjourney is far more stylistic. I personally think, with V5, Midjourney takes the lead. It has now solved the finger issues (not entirely but close enough) and is pumping out hyper-realistic images.
Midjourney is a little bit different from some of these other image generators in the way that it’s hosted entirely in Discord. Everything is public, from your prompts to your images to your edits, it’s all done in Discord rooms alongside other users. It takes some getting used to, but I think the community aspect is a huge bonus because you get a broader sense of what’s possible to create plus how to create it. We all know it’s about the prompt — so seeing users work in public will help you with your prompting.
Midjourney’s main features:
- Style transfer with text prompts
- Creating variations and remixes
- Mashups (2-3 images combined)
- Upsizing options
- Rendering options
Midjourney pricing: You can generate around 25 images for free. The Basic plan starts at $10/month where you can generate around 200 images with commercial usage rights.
4. DALL-E 2
DALL-E was created by OpenAI who also created ChatGPT (don’t know if you’ve heard of it!). Until recently, DALL-E was one of the main contenders but with these new updates, I think they’ve dropped behind. We haven’t seen an update since 2022, so at this point, it’s a waiting game.
The interface for DALL-E is actually really clean. Type in your text prompt and generate. It will show you four variations and you can do simple edits or upload your own images to play around with. All your generations are saved, and you can access them at any time.
DALL-E main features:
- Editing and retouching
- Multiple iterations of an image
DALL-E 2 pricing: Free 15 credits (or 15 images) per month. It’s an additional $15 USD for 115 credits.
Other AI image generators...
Similar to what we’ve seen with AI chatbots, all the big tech companies are determined to create their own text-to-image AI tools. Google and Meta are two of those companies. The Google Research team has been working on Imagen, which is described to have “an unprecedented degree of photorealism.” There are some images to see on their website but we’ve yet to see an actual product.
It’s the same case with Meta with their text-to-image generator called Make-A-Scene. This was announced back in July of 2022, and we’re still waiting for an update. One interesting feature explained in Meta’s blog post was the sketch-to-image, where create a text prompt and roughly sketch how you want the image to be arranged and the AI will base its creation on that.
Well, that’s all from me! I’ll be sure to update this article as new and better versions of AI generators are released. Thanks for reading ✌️
What’s a Rich Text element?
The rich text element allows you to create and format headings, paragraphs, blockquotes, images, and video all in one place instead of having to add and format them individually. Just double-click and easily create content.
Static and dynamic content editing
A rich text element can be used with static or dynamic content. For static content, just drop it into any page and begin editing. For dynamic content, add a rich text field to any collection and then connect a rich text element to that field in the settings panel. Voila!
How to customize formatting for each rich text
Headings, paragraphs, blockquotes, figures, images, and figure captions can all be styled after a class is added to the rich text element using the "When inside of" nested selector system.