DALL-E 3 vs. Midjourney: A Side by Side Quality Comparison

There’s no use sugar-coating it: algorithms have taken over the world. What started as the subject of popular science fiction films has now become essential to our daily lives. We live in a world where artificial intelligence can do your schoolwork, write poetry, automate business processes, even count how many oranges we have in the refrigerator.

My favorite use of AI? Making art.

A few days ago, after much speculation, OpenAI made an announcement that got the AI community talking: the release of DALL-E 3.

A lot of people on Twitter were thinking Midjourney was dead. I mean, a viable competitor that has API access and doesn't rely on discord? It's over – right?

Obviously both of these tools work better in their own worlds, but I think the gap is narrowing as to which is the best choice to use. DALL-E 3 looks so much more serious than what came before it.

So now, does DALL-E 3 actually have what it takes to beat Midjourney as the most popular AI image generator? Let's compare them both.

What are DALL-E 3 and Midjourney?

DALL-E 3 is OpenAI’s new AI image generation model that was released earlier in the week. Like its predecessors, DALL-E 3 is able to generate more realistic and detailed images from prompts.

With this new model, OpenAI guarantees a more "nuanced and contextual image generation" that doesn’t ignore a single line of prompt.

From their website examples, it looks like DALL-E 3 can generate images with much better creative expression, text generation, and sophistication than DALL-E 2.

DALL-E 3 is set to be available to the public through ChatGPT Plus in October 2023, but you can try it now using Bing's image creator.

Midjourney has been king for more than a year now. On a surface level, what makes Midjourney different is that it doesn’t exist as a separate platform but rather a bot integrated in Discord.

Even though it's kind of annoying, it maintains a significant and loyal following because of its creativity and attention to detail. I think up until this week it was undoubtably the best AI image generator for the average artist. That line just got blurry.

I’ve been using Midjourney for many months now and I’m way more than satisfied with their capabilities than DALL-E 2. After seeing the new examples, DALL-E 3 is extremely promising.

How DALL-E 3 Compares to Midjourney

As of right now, nobody has DALL-E 3 access until October, so we can only compare based on the samples they put on their website.

I logged into Discord, used the same prompts with Midjourney to compare, and put them side-by-side for comparison. Here's what I found out

Note that all DALL-E 3 artwork will either be on the left or top of the image comparisons, while Midjourney's will either be on the right or the bottom.

Realistic Generations

PROMPT: A chic chair with a design reminiscent of a pumpkin's form, with deep orange cushioning, in a stylish loft setting.

Both DALL-E 3 and Midjourney did an awesome job at capturing the essence of this prompt. The former produced a bright, photo-realistic loft during the afternoon with the sofa firmly in the middle of the art, while the latter’s output is a lot more moody and has an aura of sophistication.

For this prompt, I’d say DALL-E 3’s image is a lot more detailed but I still prefer Midjourney because of the contrast and the incredible job it did at simulating light and shadows.

Stylized Generations

PROMPT: A silhouette of a grand piano overlooking a dusky cityscape viewed from a top-floor penthouse, rendered in the bold and vivid style of a vintage poster.

I must say, I’m surprised at how similar these two images look. Apart from the missing music rack in the Midjourney one, I’d say these pianos are identical. The prompt didn’t even specify New York City, but they both have the Chrysler building’s silhouette in the background.

Similar to the last one, I’d describe DALL-E 3’s output as being more detailed, especially in the background. It has bolder colors that evoke feelings similar to jazz music. On the other hand, Midjourney’s output has a softer texture and palette. But, if I were to pick one, I’d give this to DALL-E 3 — it feels more coherent as an art piece and I particularly love the way they stylized the clouds.

Digital Illustration

PROMPT: Digital illustration of a beach scene crafted with yarn. The sandy beach is depicted with beige yarn, waves are made of blue and white yarn crashing into the shore. A yarn sun sets on the horizon, casting a warm glow. Yarn palm trees sway gently, and little yarn seashells dot the shoreline.

This is where we start seeing a difference between outputs of DALL-E 3 and Midjourney. On first impression, I vastly preferred Midjourney’s output: it feels more like a digital illustration and the art style they went for is more easy on the eyes. It’s beautiful, though I’d say some of the yarns looked like pasta.

However, if we’re being strict, I’d have to give this round to DALL-E 3 for one reason: it followed the prompt. It may be little, but there was a line there that said the sun had to be made of yarn too. And, as someone who sometimes becomes too frustrated with Midjourney’s stubbornness, attention to detail is a quality I greatly appreciate.

Pixel Art

PROMPT: Pixel art scene of Coit Tower standing tall on Telegraph Hill, with a panoramic view of the city below and birds flying around.

For me, there’s a clear winner in this round. While Midjourney’s artwork is a valiant and beautiful attempt, DALL-E 3 is the one who produced true pixel art. If you zoom into Midjourney’s piece, you’d see that it’s a lot softer and is similarly animated to some Disney movies while DALL-E 3 has such rich details while staying true to its prompt — you could’ve said it was a still from an 8-bit game and I wouldn’t bat an eye.

Surrealist Art

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.

This is another instance where these two tools went in a completely different direction. There is a lot to love with these generated artworks, but DALL-E 3 once again triumphs over Midjourney in this prompt. 

I couldn’t describe it with any words other than it’s so bizarre. DALL-E 3 managed to capture that dreamlike quality that enthusiasts seek in surrealist art. It’s absurd, subversive, and a little psychedelic, which is precisely what the prompt asked for.

Meanwhile, Midjourney has a more grounded output. It manages to keep that sense of whimsy that’s present in surrealism but it’s subdued to the point of mainstream. Nevertheless, it managed to follow the prompt reasonably well. I only have one question for Midjourney though: where is the pepperoni sun?!

Flat Design

PROMPT: Flat design illustration of a diverse family of monsters. The group includes a furry brown monster, a sleek black monster with antennae, a spotted green monster, and a tiny polka-dotted monster, all interacting in a playful manner.

As far as creativity goes, Midjourney and DALL-E both gave a satisfactory output for the prompt. However, this round belongs to DALL-E solely because it managed to produce all of the characters listed in the prompt whereas Midjourney only gave us the fuzzy brown and the green spotted monster.

Sketches

PROMPT: An ink sketch style illustration of a small hedgehog holding a piece of watermelon with its tiny paws, taking little bites with its eyes closed in delight.

Hedgehog? Check. Watermelon? Check. Cute tiny paws? Check.

For simple prompts, it seems like DALL-E 3 and Midjourney produce similar artwork. That said, I’m going to have to award DALL-E 3 another point because it understood the prompt to a tee, including the closed eyes and taking bites of the watermelon.

Botanical Illustration

PROMPT: An antique botanical illustration drawn with fine lines and a touch of watercolor whimsy, depicting a strange lily crossed with a Venus flytrap, its petals poised as if ready to snap shut on any unsuspecting insects.

As a fan of this art style, these are phenomenal. They both evoke a slight feeling of wonder that’s present in similar artworks. As far as context goes, these look exactly like a weird mash-up of Venus flytraps and lilies, in a good way. DALL-E 3’s drawing has a lot more contrast while Midjourney’s has a lot more subtle details that blend together.

I have no choice — I must give this round a tie.

Oil Painting

PROMPT: A detailed oil painting of an old sea captain, steering his ship through a storm. Saltwater is splashing against his weathered face, determination in his eyes. Twirling malevolent clouds are seen above and stern waves threaten to submerge the ship while seagulls dive and twirl through the chaotic landscape. Thunder and lights embark in the distance, illuminating the scene with an eerie green glow.

If I were to describe these in one word, I’d use “breathtaking.” Generative AI art has really stepped up its game the last few years — and these are proof of that.

Let’s start with Midjourney’s piece: it’s derivative of artwork from the Baroque period. It does well to capture the weariness of an old sea captain. You can almost hear the crashing waves in this painting. The details like the water crashing on the ship cabin and the sun peeking on the horizon are also top-notch. I vastly prefer this one to DALL-E’s.

That said, DALL-E’s generated output manages to capture every detail in the prompt, from the seagulls to the green glow. However, I’m not a big fan of the personification of the clouds to make it appear malevolent; sometimes, simplicity triumphs style.

3D Renders

PROMPT: A 3D render of a coffee mug placed on a window sill during a stormy day. The storm outside the window is reflected in the coffee, with miniature lightning bolts and turbulent waves seen inside the mug. The room is dimly lit, adding to the dramatic atmosphere.

I’m simply blown away at how well DALL-E does at this round. Don’t get me wrong — I still like Midjourney’s output, but it’s simply too (for the lack of a better word) pedestrian. It’s similar to other artwork I’ve seen online, and I couldn’t even see the waves inside the mug.

On the other hand, DALL-E 3 was able to provide an exceptional rendering of the storm and crashing waves inside the coffee cup. Apart from that, the lighting from the little light bulbs on the side was a nice touch and well-executed.

Architecture

PROMPT: A modern architectural building with large glass windows, situated on a cliff overlooking a serene ocean at sunset.

I’d love to live in either of these two houses but, if I were to choose which one is better designed, I’m going with DALL-E 3’s. Now, I’m not an architect but surely putting the design on Midjourney’s render isn’t safe. 

For the background, I still prefer DALL-E’s light blue and orange hue to Midjourney’s more subdued palette. I also really like the details on the other cliffs from the former, as well as the reflection of the clouds on the sea.

Diorama

PROMPT: A minimap diorama of a cafe adorned with indoor plants. Wooden beams crisscross above, and a cold brew station stands out with tiny bottles and glasses.

This one’s a no-brainer, in my opinion. DALL-E 3’s coffee shop has a more welcoming ambiance and I particularly love the little “Cold Brew” sign on the wall, which is already a massive improvement over DALL-E 2 considering my experience with trying to generate text with it.

However, the lighting and contrast on Midjourney’s coffee shop is just immaculate. More than that, I love how detailed it is. From the plants to the espresso machine, every little corner has its own personality. For that reason, this one’s clearly a point for Midjourney.

High-Context Prompts

PROMPT: A middle-aged woman of Asian descent, her dark hair streaked with silver, appears fractured and splintered, intricately embedded within a sea of broken porcelain. The porcelain glistens with splatter paint patterns in a harmonious blend of glossy and matte blues, greens, oranges, and reds, capturing her dance in a surreal juxtaposition of movement and stillness. Her skin tone, a light hue like the porcelain, adds an almost mystical quality to her form.

Let’s be real: Most of the earlier prompts were already high-context and Midjourney failed to follow every single one of them. But, let’s give it a chance — maybe, this time, it’ll do a better job of following the prompt.

Unfortunately, that’s not the case. To be fair, both of them failed from the very beginning. The prompt asked for a middle-aged woman but DALL-E 3’s main subject is too old, while Midjourney’s is too young. What saves DALL-E 3, however, is that it’s right more often than it’s wrong.

That’s why DALL-E 3 notches another point.

Low-Context Prompts

PROMPT: Lychee-inspired spherical chair, with a bumpy white exterior and plush interior, set against a tropical wallpaper.

Let’s end this the same way we started: with a chair. It might be my personal preference, but these have to be some of the most uncomfortable chairs I’ve seen in my life. I mean, at my height how would you even fit in those?!

Anyway, I’m getting off-track. Once again, DALL-E 3 wins out of nuance and contextual understanding alone. While I prefer everything in Midjourney’s output, the exterior of the chair isn’t bumpy at all.

The Verdict

With a score of 11.5 out of 14, DALL-E 3 wins this head-to-head battle with Midjourney if we're going to be comparing results based on the literal words in each prompt.

While I personally preferred Midjourney’s art style overall, what made DALL-E 3 stand out is its better understanding of prompts. It handled hyper-specific requests really well and rarely made a mistake in its generations.

It’s also worth mentioning that this is the third iteration of DALL-E 3, while Midjourney is technically still an open beta (even though it's on version 5). With rumors of Midjourney v6 being around the corner, I’m excited to see how well it shapes up to DALL-E.

For now, I have to crown DALL-E 3 the winner. And, if this is a preview of what’s to come, I can’t wait till we actually have access to customize these comparisons and personally figure out which one works best for what you're looking to create.