• Practical AI
  • Posts
  • Is Midjourney Doomed? Crazy New Update from OpenAI!

Is Midjourney Doomed? Crazy New Update from OpenAI!

The new Dall-E version 3 will live inside ChatGPT!

Hey, how’s it going?

Have you noticed how the AI hype has died down a bit - at least in the media?

Don’t be fooled - AI is not going away!

In fact, it is rapidly evolving.

While I have stopped trying to keep up with the latest news, AI awareness is important. That’s why I try to focus on deeper dives, and leave the news to the other guys.

Today, I had plans to cover Google Bard’s insane upgrade, but then OpenAI announced some HUGE news on Wednesday that I simply cannot ignore…


In mid-October, I will launch a brand new newsletter about AI in healthcare. It will focus on cancer and the advances AI brings to help cure this bastard! Do you know someone who suffered or died from cancer, or maybe you work in the healthcare industry? You don’t want to miss this weekly newsletter! Sign up here

OpenAI launches Dall-E 3 with ChatGPT 😮 

Kind of, but both are HUGE announcements on their own.

I decided to focus on Dall-E 3 and ChatGPT today and discuss Bard’s big thing next week.

So what’s the big deal, and what the heck is Dall-E anyway?

There are three major players in the generative AI image space: Midjourney, Stable Diffusion, and Dall-E.

Dall-E has been lagging behind, while Midjourney has been quantum steps ahead of the other two.

On Wednesday this all changed…

OpenAI created a cute little video showcasing what Dall-E 3 can do, and how it integrates with ChatGPT (which is INSANE).

I’ll get back to why this is amazing shortly.

But first, have a look at this video:

Why Dall-E 3 is a big deal

Until now, ChatGPT has been the “darling” of generative AI for text, while Midjourney was the superstar of image generation.

I wonder if Midjourney is about to get crushed to pieces by OpenAI…


Let me first state that Midjourney is AWESOME!

However, in order to use it, you need to first use Discord.

Discord is, for the “average Joe,” complicated and not very user-friendly. In other words, the barrier to entry is high - making it a product (at this point in time) for the more tech-savvy and biggest enthusiasts.

Not a chance in h*ll my wife would even try to figure out Discord in order to play around with Midjourney…

On top of that, if you do make it into Discord and Midjourney, you will need a higher-tier subscription to keep all your artwork and ideas and chat with Midjourney private.

I will not go into details about how it works, but in one sentence, you generate your pictures in a “public” prompt screen that you share with “everyone else.”

Want your stuff to be private? More $$!

Prompting Midjourney, and any generative image AI for that matter, requires a much higher level of prompting skills than ChatGPT or Bard.

You need to know a lot about art, images, and photography terminology to get the maximum benefit from it (such as type of lens, debt of field, and a whole host of other terminologies). On top of that, you can prompt the AI for various styles, types of art, types of images, and much more.

And until recently, it was extremely difficult to rework a specific image you generated from Midjourney.

Not to mention the infamous toes and fingers or trying to get text into an image.

Midjourney 5.2, the latest model, gave its users many new features. But the interface is still via Discord, eliminating most people from using its powers.

It is one super duper powerful tool if you learn how to use Midjourney and all its “hidden” features.

But I think it is faster, easier, and more intuitive for people to use ChatGPT, Bard, or Claude 2 over Midjourney, Stable Diffusion, and even Dall-E 2.

But yesterday, OpenAI disrupted the disruptors…

Enough already, give me the scoop!

One thing all the text-to-image systems (AIs) have in common is that they have a tendency to ignore words or descriptors in the prompt, forcing users to learn prompt engineering to tweak the AI to do what you want it to do.

With Dall-E 3 tightly integrated into ChatGPT, we not only eliminate the need for Discord but also the need to understand and learn complicated prompt engineering.

Need to iterate on the output?

Tell ChatGPT what you want to change.



Dall-E 3 also has full control over adding text to images, which so far has been a big issue in AI (Oh, the Memes I will make, the meeeeemes!!!!)

And lastly, it is effortlessly able to reiterate an image without changing the subject in the image, which is far from easy in Midjourney

In other words, the barrier for people to start using an AI for image generating was just removed… Or at least lowered immensely.

But what about the price?

Check this out…

It will come included with ChatGPT Plus ($20/mo)!

And since it all happens in your private app/chat window, nobody will see your work, prompting, and experimentations.

This means that I can stop using Midjourney and Discord (I dislike it so much), and it saves me $60/mo on a pro plan with stealth mode (private mode).

In short, I think this is a quantum leap forward for ChatGPT, OpenAI and AI in general.

Dall-E 3 will make it so much easier for people to get started with AI image generation AND get meaningful outputs - which you can even use as you please (no copyright issues).

The images we have seen from Dall-E 3 so far are insanely good.

I found this interesting Tweet on the app formerly known as Twitter, where this guy Matt compares outputs from Dall-E 3 to Midjourney.

Here is his prompt:

"A vast landscape made entirely of various meats spreads out before the viewer. tender, succulent hills of roast beef, chicken drumstick trees, bacon rivers, and ham boulders create a surreal, yet appetizing scene. the sky is adorned with pepperoni sun and salami clouds."

Here are the outputs:

They are both great, but personally, I think the Dall-E one is way better! Look closely at the images and the prompt. Do you agree?

You should check out the full thread on Twitter/X and see the original images and many many more by Matt Garcia (it's amazing):

A few words on safety and restrictions

Safety is a big deal for these AI companies and when an AI starts going bananas, the media is having a frenzy.

OpenAI has taken great strides towards limiting Dall-E 3’s ability to generate violent, adult, or hateful content.

It will also not let you generate art in the style of a living artist. I find that wording interesting. So I assume you can generate art in the style of Picasso and Dali still…

And lastly, Dall-E 3 has mitigations to decline requests that ask for a public figure by name.

Fun fact

Dall-E is a portmanteau of the Pixar character Wall-E and the artist Salvador Dali.

That’s all I had for you today.

What do you think?

Is this a big deal?

Do think artists, photographers, designers and stock image companies are doomed?

Will you be using AI images in your business, presentations, webpage, catalogs, etc? Or you will continue to buy and pay for them as before?

Let me know what you think by replying to this email or tweeting me @thomassorheim 

AI’ll be back!

- Thomas

What do you call an AI that makes images of food?



An arti-fish-al intelligence!

(Today’s dad joke courtesy of Claude 2)

This is not the end. It is where the fun begins!

Join the conversation

or to participate.