Buried in the sea of other exciting AI news like Sora and GPT-4o is the release of Adobe Firefly 3 in late April. This new and improved model from Adobe Firefly 3 adds more features to its already impressive repertoire along with promises of better nuance and creativity.
Here’s the thing about me: I have this compulsive desire to put AI models to the test by comparing them to the most popular models in its segment. So, for Adobe Firefly 3, that would be Midjourney.
In this article, I’ll be going through Adobe Firefly 3 and Midjourney including their features, output quality, and answer why Adobe won’t overtake Midjourney anytime soon.
What are Adobe Firefly 3 and Midjourney?
Adobe Firefly 3 and Midjourney are both AI image generation models, but they exist for completely different reasons. Starting with the former, Adobe Firefly is a model created to coexist alongside other apps in the Adobe suite such as Photoshop, InDesign, and Illustrator.
The third version of this model was released in late April 2024. I’d classify its outputs as “stock photo”-like because it’s made as an aiding tool for designers.
On the other hand, Midjourney sits atop the AI image generation space due to its creativity and popularity. While Firefly exists to help designers, Midjourney is made to test the limits of what AI can create. This platform has more than 15 million concurrent users worldwide and is currently on its sixth iteration. You can catch our full review of Midjourney’s latest model here.
Adobe Firefly 3 vs. Midjourney: Features
As far as AI image generators go, these two have the most features. Let’s list them one-by-one, starting with the features that they share:
- Output Control. Allows you to switch between generating artwork or images.
- Aspect Ratio. Controls the aspect ratio of the output.
- Style Strength. Impacts how the supporting information of your prompt affects the overall look.
- Style Reference. Lets you input an image as a style guide.
- Generative Expand. Makes the canvas of the original image bigger by generating extra details to fill in the blanks.
Two of the biggest features that Adobe Firefly 3 has that Midjourney doesn’t are integration to other Adobe products and structural reference, which allows you to upload an image that can serve as a structural blueprint for the generated photo.
That said, Midjourney has tons more to offer than Adobe Firefly 3. The most significant ones include a character reference feature that allows you to use images so the succeeding output can copy its subject, blending of two images, image-to-prompt conversion, and image variation strength.
There’s also a difference in how these two models implement their features. While Adobe Firefly 3 has accessible buttons for customization, Midjourney relies more on how you structure your prompt for better control. This means the former is a lot more user-friendly, while the latter is more flexible.
Adobe Firefly 3 vs. Midjourney: Output Quality
Realistic Photograph
Prompt: candid photography, film photography, medium shot of a small girl playing with bubbles
Product Photography
Prompt: commercial photography, a glass of whiskey, industrial setting, sunset, high contrast, cinematic lighting, minimalist
Digital Art
Prompt: whimsical and fluid illustrations, dreamlike threads, dreamlike realism, abstraction, honeycore, gradients, inspired by graphic design, minimalist, pastel aesthetic
Character Design
Prompt: a knight in a dark green armor entering a dark forest, 32-bit game, character design, dark fantasy, gloomy atmosphere
Architecture
Prompt: architectural interpretation of open spaces interior, a verdant backyard, minimalist interior design, light brown and light amber, film photography
With Text Generation
Prompt: a teacher writing the word “PYTHAGOREAN THEOREM” on a chalkboard
Thoughts?
I actually thought Adobe Firefly 3 held its ground in this comparison, but it’s still no match for Midjourney. The first set of images were the closest they ever were. And yet I’m still learning towards Midjourney’s output because of the simplicity and the addition of B&W effects to follow the “film photography” aspect of the prompt.
As for the others, it’s clear that Adobe Firefly 3 is a step towards the right direction, but it still lacks the creativity and nuance of Midjourney. Its output for digital art is a little run-of-the-mill and I’m not a big fan of the character design artwork.
One thing that I’ve noticed with Adobe Firefly 3 is the lack of its structural coherence. For example, in their product photography image, the glass has a small bulge in its left which isn’t present on the other side. There are also elements in their architecture output that’s out of the norm, such as the footstool being on top of another chair and the sofa on the right being impossible in real life.
All in all, a good effort for Adobe Firefly 3 — but it still has a long way to go before it approaches Midjourney’s consistency and creativity.
The Bottom Line
Let’s return to the question posed at the very top of this article: does Adobe have what it takes to edge out Midjourney?
My answer is simple: it doesn’t, and I’m not sure it ever will.
The fact of the matter is that these are two completely different platforms. Of course, Adobe Firefly 3 will always have images that look stock-like because that’s what it’s for. On the other hand, Midjourney will always have more creative images because it’s meant to fully explore where AI art could go.
At the end of the day, these AI models can coexist because they have completely different demographics. The only thing that matters is how they both perform in their respective lanes.
Want to learn more about AI art? You can read our biggest comparison to date here. Have fun!
