DALL-E 3 launch: Did OpenAI just take the AI image throne?
OpenAI’s DALL-E 3 just solved the three biggest problems with AI image generation, and integrated it with ChatGPT Plus — is this the end for everyone else?
A few short weeks ago, I wrote a blog series on Ideogram.ai’s groundbreaking new solution to achieving rendered text in AI generated images:
At the time, I was wondering if that specific feature, coupled with really impressive image quality, would put Ideogram ahead in the race for the AI art throne.
I still think Ideogram is great, but the AI image generation world just got turned upside down by DALL-E 3.
Open AI releases DALL-E 3
The tech world has been abuzz with anticipation for the upcoming release of DALL-E 3 for weeks, and the latest innovation from OpenAI looks very exciting.
Rumor has it, the model has been officially released to beta users, and will be integrated into ChatGPT for Plus users starting in October.
DALL-E 3 known features
It’s important to note that OpenAI has been fairly tightlipped until this week when they released their pre-launch material to the public here:
Building on the success of its predecessors, DALL-E 3 promises to revolutionize the field of image generation through advanced machine learning algorithms. This new version aims to address previous limitations while introducing a slew of cutting-edge features that are set to redefine what’s possible in the realms of art, advertising, and beyond.
Like, for example, solving the three biggest problems that exist in AI image generators:
- Prompt engineering
- Text generation
- Consistent characters
If this is as frictionless of an experience as I believe it will be from OpenAI, this could be a major moat creating value driver.
Let’s get into those features.
ChatGPT Plus integration eliminates prompt engineering
I wrote a blog a few weeks ago about a ChatGPT plugin that acts as your meta prompt engineer to optimize your prompts.
OpenAI just built that in with this new DALL-E 3 x ChatGPT collab.
The beauty of this, as was clearly already identified by the creators of the Prompt Perfect plugin, is that it eliminates a great deal of the need to learn prompt engineering to get usable output.
I spent months with Midjourney to understand how to get exactly what I wanted out of prompts. That’s no longer required.
And, for OpenAI, this means a much larger, immediately addressable market. This will absolutely take dollar for dollar market share from Midjourney and Stable Diffusion, and the single integration will widen their moat.
This is a competitive masterclass from Sam Altman and team.
Text rendering in DALL-E 3
The first thing that caught my eye in their example images was this cartoon avocado with readable text:
The next thing was a movie poster with high quality text rendering:
Let me point out that I don’t see much on the page that talks about text rendering as a specific feature of DALL-E 3. But, knowing how big of an issue text generation has been in the space, it would be a very curious choice to use these example images if it was not a feature.
If the model couldn’t do it well or consistently, that’d be a big disappointment, and OpenAI knows that.
Consistent characters in DALL-E 3
This has been another complaint of AI image generators, specifically with Midjourney. It’s been quite difficult to date to do it well without lots of work.
Check out this video that Sam Altman posted to X this week:
This also feels like a masterclass in Marketing from OpenAI. I can envision the marketing team having a conversation like this:
“Hmmm, easily creating consistent characters is a top 3 feature request. Maybe we just subtly create a video of a sunflower hedgehog named Larry, and then generate Larry in a bunch of different scenes to see if anyone notices.”
Again, I haven’t seen them claim this is a feature, but knowing how big of a need this is for their customers, it would be a poor choice for them to highlight consistent character generation in their launch video if it wasn’t a feature.
DALL-E 3 launch
I’m trying to temper my expectations until this shows up in my ChatGPT Plus, but y’all, I’m so excited. I can see tons of my current workflows that this could immediately improve.
We’ll have to wait another week or two, but here’s to hoping that this launch is everything that we’re hoping for!
I hope you found this blog helpful. If you like this type of content, I post daily news, tips, and tutorials to help you navigate the digital world. Follow for more!
Related Content:
