Midjourney vs DALL-E vs Firefly vs Meta Prompt Comparison

I remember the awe I felt after using Midjourney and DALL-E for the first time. It felt like dark magic the way it weaved images using my own ideas. But, after a while, that magic wore off. There's nothing really new, and it's plagued by the same problems it had for years like nuance and text generation.

The space is in dire need of new competition. And, just last week, Meta pulled through with its own AI image generator.

But how good is it when compared directly against established faces like Midjourney, DALL-E, and Firefly? And, more importantly, what does it bring to the table that the others don't? More on that in this article.

Midjourney vs. DALL-E vs. Firefly vs. Meta: Image Comparison

Realistic Portraits

Prompt: a student, close up shot, urban setting, blue hour

As usual, Midjourney comes in with a stylized image. It’s not bad — but it’s not the right art style for a portrait image. The hair is too smooth, and the face is too flawless. It’s too unrealistic to pass as an actual photograph.

DALL-E is worse. Everything is a little too perfect…“uncanny valley” levels of perfect. You could’ve told me that this was a render of Chad Squidward and I would’ve believed you.

To its credit, Firefly does try. If we’re talking about realism only, it’s way ahead of the previous two. However, between the photo composition and the style of the shot, it feels too much like a stock image — which is great if you’re using Firefly for stock photos along with the rest of the Adobe suite.

Without a doubt, our new contender wins this category. Meta produced an image that could pass as human. It’s well-shot while retaining that candid feel, has amazing lighting, and the shadows are perfectly rendered. It even has that cinematic feel to it. This isn't really surprising, considering that the model is trained on over a billion Facebook and Instagram images (Wait, did I give permission for that?!)

Landscape

Prompt: landscape photo, alaska, night, aurora borealis

All of these are pretty good. I like Firefly's image the least because the colors are too punchy for my taste. There's also something wrong with the proportion of the trees, at least in my opinion.

DALL-E would've been my favorite but it did overcompensate a little when it comes to the mountains, which fell victim to AI image generators' tendency to repeat elements of an image. Other than that, the color grading is good, details are amazing, and I particularly like the decision to set it at sunrise — giving it a different color to play with other than dark tones.

Midjourney is incredibly detailed, down to the reflections. The colors are vivid, but not obnoxiously so. Same with DALL-E, it had some repetitions, particularly in the pine trees, but it's less noticeable.

Somehow, Meta notches another win. Some may argue that the absence of color takes away from the beauty of the Northern Lights, but I disagree. There's something about how muted the surroundings are that contributes to the grandeur of the Aurora Borealis.

Digital Art

Prompt: mountain range, digital art, low poly, geometric, colorful gradients

To be honest, I can't separate DALL-E, Midjourney, and Meta in this category. They used the same color scheme for the gradient, which looks good but is also so common nowadays. Their biggest difference is with their geometric composition. Meta is sharper, DALL-E is softer, and Midjourney is flatter.

If there's a common thread between these Firefly images, it's the vividness of their colors. This time, though, it works. Unlike the other three, Firefly had more than three colors in its gradient, and it's still a coherent artwork. I also love the detailed and sharp lines separating each polygon, which gives the overall look some depth.

Pixel Art

Prompt: pixel art scene, a quiet library, atmospheric, 16-bit

Meta had a good idea for what it wanted to do. I even like the addition of plants, giving it a sort-of well-kept, dystopian vibe to it. However, if you zoom in, it's not really pixel art. It's most noticeable in the chair and the table.

DALL-E has the most pixel art vibe to it. It genuinely looks like a still from an 8-bit game mystery game, where you finally encounter the killer. However, one thing I don't like is how the second floor morphed into the book case.

For pixel art, I've yet to see anything beat Midjourney. It's simple, cozy, cute. The windows and clouds adds a dreamy vibe to the entire artwork. What sets is apart though is the lighting, which is just superb.

As for Firefly, yeah, I don't know what happened there either.

Surrealism

Prompt: the entire universe held inside a sunflower, surrealism, expressive oil painting

Meta's output is simple but effective. It satisfied the prompt well. The image of the universe inside of the sunflower reminds me a lot of Van Gogh.

I like Midjourney's artwork a lot, but I don't think that it prioritized the second part of the prompt well enough. There are some stars, and the darkness suggests that the sunflower is subjected in space, but overall, it's pretty vague.

DALL-E did amazing. The way it's constructed is Lovecraftian. I like the enormity of the entire artwork, and the addition of eyeballs in the planets adds a lot to the creepiness and whimsy of the piece.

Once again, Firefly's image is a disappointment. It didn't even include the universe, which is a critical part of the prompt.

Text Generation

Prompt: a speakeasy's neon sign saying "Open"

Unfortunately, text generation still hasn't progressed well enough in AI image generators. This is most apparent in the images from Meta and Midjourney.

Firefly did manage to create readable text, but the uneven lines and overlapping letters suggest that it got there by mimicking the shapes and not by understanding what the word means.

DALL-E 3 remains to be the industry standard when it comes to text. Sure, it's not perfect, but it's readable. It even used cursive and put some flair into the neon sign. This is as good as it gets (for now).

High Context

Prompt: an epic battle scene between a medieval army and a horde of dragons with knights charging on horseback, a brave dark knight standing in the middle of the battlefield, archers firing arrows, dragons breathing fire from the sky, a sense of chaos and heroism, dark fantasy illustration, atmospheric, gothic

Midjourney had, by far, the best artwork here but, in terms of context understanding, it failed the test. All elements are there but because of the sheer volume of the prompt, some of the subjects were not properly rendered. For example, look closely at the dragons. Notice how some of them are just wings?

Meta understood the subject of the prompt properly, but wasn't able to follow the styles that I mentioned in the latter parts.

DALL-E 3 is still the most nuanced AI image generator out there. Between its level of understanding and the intricacy of the artwork, this is an easy win for DALL-E.

And Firefly, you're doing great, honey. Keep it up or, you know, it might be time for a new update. Sooner than later, I hope.

What Are They Good At?

In Summary

In this four-way competition, the winner is a tie between DALL-E 3 and our newest entry, Meta. However, always remember that this is just our own testing.

For instance, I would still use Midjourney over these two because of my line of work because I don't really deal with high context prompts, text generation, or realistic images that often. And even though I had some harsh words for Firefly earlier, I do recognize the value of its accessibility when paired with other Adobe software.

What about you? Do you have some opinions about these AI image generators?