OpenAI's Dall-E 3 has been on the scene for about a month, and creative enthusiasts everywhere are diving into various use cases. The potential seems limitless, from creating AI images to producing short films.
10 minute experiment to see how fast i could generate a film with AI.
— Oliver McCann (@OliverZMcCann) October 14, 2023
I asked #ChatGPT for a story and then to generate images with #DALLE3
Ran the images through #runwayml and added some music.
This was the story GPT4 came up with:
Echoes of the Past:
Intro: In a jazz bar,… pic.twitter.com/6hlPknBYY7
Now you might be asking questions: Is Dall-E 3 really worth the hype? Is it better than Midjourney?
If you've been using Midjourney for your AI image needs, you might wonder if a switch is in order.
In this blog post, we’ll dive into an in-depth comparison, where we put Dall-E 3 against Midjourney using 16 distinct prompts to understand the strengths and shortcomings of each platform.
What are DALL-E 3 and Midjourney?
DALL-E 3 is OpenAI's newest AI art generator.
It's built into ChatGPT, making it user-friendly, and is available through ChatGPT Plus for $20 a month. While still in beta, it makes waves in various fields for precise images.
Check out the detailed guide on How to use Dall-E 3.
On the flip side, we have Midjourney, a bot inside Discord.
It's known for its rich styles and emotions in images. For $10 a month, you can start with their basic plan, but be ready to tweak your prompts sometimes.
Here is a detailed guide on how to use Midjourney.
So, DALL-E 3 offers detailed art through a dedicated platform, while Midjourney, within Discord, leans into creativity and emotion. Both have their own advantages. It all comes down to what you are looking for.
Dall-E 3 vs Midjourney: A comparison matrix
| Dall-E 3 | Midjourney | |
|---|---|---|
| Ease of use | Very easy | Medium |
| Cost | $20 per month | Starts at $10 per month |
| Image quality | More nuance and detail | Good |
| Image style | Supports all art styles | Supports all art styles |
| Image size | Square, tall, and wide | Supports custom sizes |
| Creativity | Understands user intent | Adjust creativity levels |
| Image generation speed | A bit slower | A few seconds |
| AI images copyright | Users own the images they created | Users own the images they created |
| Realism | Less life-like but more detail | More realistic |
| Customization | Limited customization options | More customization options |
Dall-E 3 vs Midjourney: The Ultimate Showdown
Looking at a comparison table can give you a brief idea, but you will only understand the strengths and weaknesses of each AI art generator by doing a side-by-side comparison.
In this section, we handpicked some of the best images and art types. We’ll use the same prompt in Dall-E 3 and Midjourney for each type to compare the results.
Note: All the images to the left are created in DALL-E 3, and to the right are created by Midjourney.
Landscapes
Prompt: Golden wheat fields under a stormy sky, with a lone scarecrow wearing a bright red scarf
The Dall-E 3 image has a detailed, illustrative style with a warm, golden hue, showcasing a scarecrow-like figure. In contrast, the Midjourney’s image has a more photographic feel, focusing on a cloaked figure in a looming storm, painted in sepia tones. It completely missed the scarecrow.
Abstract concepts
Prompt: Visual representation of the sound of laughter using vibrant bursts of color and swirling patterns
The Dall-E 3 picture has many mixed colors, looking like they're spinning, with lots of blues, making it feel dreamy. The Midjourney picture has a lady laughing with colorful patterns around her, making the laughter feel alive and real. Both are cool in showing the joy of laughter.
While Midjourney did a great job, the image does not look like abstract art. Dall-E 3 understood the intent of the prompt and generated an abstract visual.
Historical settings
Prompt: A gladiator preparing for battle in a Roman Colosseum, adjusting his helmet and gripping his shield
On the left, the Dall-E 3 shows a gladiator with a detailed and ornate helmet standing before the Colosseum. The ambiance is more serene, and the sunlight illuminates his gear.
On the right, the Midjourney image presents a more rugged gladiator in an intimate moment. This warrior seems lost in thought, perhaps reflecting on the battle ahead. His armor is more battle-worn, and the scene feels darker and more intense. He tightly grips his ornate shield, showcasing his determination.
Both images look real. The Dall-E 3 one has included almost everything we asked in the prompt, but Midjourney missed the helmet and colosseum. Dall-E 3 also missed the ‘adjusting the helmet’ part.
Futuristic scenes
Prompt: Cybernetic street musicians playing luminous instruments in a neon-lit alley of a metropolis
The left image by Dall-E 3 shows a calm, long alley with alien-like musicians and bright neon signs. It made sure to have perfect details of the background, too. The right image by Midjourney feels busier, with a mix of humans and robots and a wider, vibrant alley filled with reflections from neon lights. While both pictures show futuristic musicians in neon-lit alleys, Dall-E's feels more like on another planet, and Midjourney's has a mix of today and future vibes.
Portraits
Prompt: An elderly woman with silver hair tied in a bun, wearing vintage glasses and embroidering a colorful pattern
These two images beautifully capture an elderly woman working on her embroidery. The Dall-E 3 image on the left shows a woman with striking vintage glasses and silver hair tied in a bun. She is working on a vibrant pattern. The ambiance is refined, with soft lighting highlighting her features. The right image by Midjourney seems more candid, where the lady wears more casual, black-rimmed glasses and is dressed in a colorful blouse.
Both images emphasize the art of embroidery, but the Dall-E 3 leans towards elegance while the Midjourney one feels cozy and authentic.
Pixel art
Prompt: A mage casting a spell, with magic particles and a floating spellbook, against a pixelated enchanted forest background
On the left, Dall-E 3 offers a pixelated image of a forest background with the mage cloaked in deep blue with a tall hat, replicating an old-school video game vibe. You can see the magic particles swirling around him and the floating spellbook, which is wide open, showcasing its glowing pages.
Now, on the right, Midjourney paints a more realistic picture. The mage is portrayed as a young, intense-looking man, deeply engrossed in the act of spell-casting. The magic particles are vividly visible, surrounding the glowing orb-like spellbook he holds. While the forest background is evident, it isn't pixelated as the prompt had asked.
While both images brilliantly depict a mage casting a spell, only Dall-E 3 nailed the 'pixelated' detail.
Surrealist art
Prompt: An oversized butterfly reading a book to a circle of attentive, tiny elephants on a floating island
Both images are created using the same prompt but paint very different scenes. Dall-E 3's image is vibrant and fun, showcasing a butterfly with an elephant's head reading a book to tiny elephants on a floating land.
On the other hand, Midjourney's image has an enchanted jungle feel with a giant elephant island and many small elephants doing different activities. But, Midjourney's version misses the central element of the "oversized butterfly."
Flat design
Prompt: A minimalist postcard showcasing Tokyo's essence through iconic silhouettes like Tokyo Tower, a sushi roll, and a cherry blossom branch
Both images capture Tokyo's essence using Tokyo Tower, sushi, and cherry blossoms. Dall-E 3's version is vibrant, showing a detailed cityscape and sushi roll against a bright backdrop, and the cherry blossoms are lush.
In contrast, Midjourney has a calm and minimalist approach with a pastel palette, simplified structures, and fewer cherry blossoms.
While both creations encompass the requested elements, Dall-E 3 adds extra features like a river and bridge. Quality-wise, Dall-E's image is richer in detail, while Midjourney's prioritizes simplicity and open space.
3D renders
Prompt: A detailed 3D rendered jade dragon pendant with ruby eyes, suspended on a delicate silver chain against a velvet backdrop
Dall-E's pendant (on the left) closely matches the ‘jade’ look with its green color and has ruby-red eyes, but the silver chain seems thicker than expected. The backdrop looks like velvet.
Midjourney's pendant (on the right) doesn't look as much like jade and has a more metallic feel, but its ruby eyes are prominent. The chain here is more detailed, and the background is plain dark. Compared with the prompt, Dall-E's image aligns better with the ‘jade’ and ‘velvet backdrop’ details, while Midjourney nails the ‘silver chain’ aspect.
Digital illustration
Prompt: A digital illustration of a mischievous cat trying to sneak a fish out of a bowl while a parrot nearby shouts a warning
Both pictures show a cat trying to get a fish from a bowl with a parrot nearby. Dall-E 3's image on the left has a gray-striped cat calmly touching the water, and the parrot is just watching.
In the Midjourney picture on the right, the cat looks surprised, and there's no parrot. Dall-E's picture has more detail and texture, making it look more polished. Midjourney's image feels rushed and has a darker setting with missing elements.
Oil painting
Prompt: A solemn sailor lost in thought, holding an old compass, with the tumultuous sea and storm clouds in the backdrop
The left image, made by Dall-E 3, has an older sailor looking thoughtful with a stormy sea behind him. The right one, by Midjourney, features a younger sailor looking out to a calmer sea. Both pictures match the prompt, but Dall-E's seems closer because of the stormier backdrop. The image quality is good in both, but they give different feelings: one feels like looking back on past adventures, and the other feels like getting ready for a new one.
Diorama
Prompt: A miniature carnival scene, with a working Ferris wheel, tiny visitors enjoying cotton candy, and a clown juggling glowing orbs in diorama style
Both images show miniature carnival scenes with Ferris wheels. The left image by Dall-E 3 has visitors with cotton candy and a clown juggling glowing orbs, fitting the prompt well. The right image by Midjourney has a night-time feel and more complex designs but doesn't show visitors with cotton candy or the juggling clown. While both images have good quality, Dall-E's image aligns closer to the prompt's specifics, whereas Midjourney's offers a unique take, but the tiny visitors are not so clear.
Architecture
Prompt: A whimsical treehouse library with spiral staircases, hanging lanterns, and balconies filled with books
The left image by Dall-E 3 is more fantasy-like, with many details, lanterns, and a bigger treehouse. The right image by Midjourney feels closer to real life, with fewer rooms and lanterns. Both pictures capture the idea of a ‘treehouse library’ with spiral stairs and book balconies. They both follow the prompt well.
However, Dall-E's picture has a more dreamy feel with its greenish glow, while Midjourney's seems set in the evening and feels cozier.
Both images are high-quality, but the choice between them is whether you like a more magical or realistic look.
Interior design
Prompt: A bohemian bedroom with a hammock bed, tapestries on the walls, a mosaic mirror, and plants hanging from the ceiling
Both images capture a bohemian bedroom feel. Dall-E's image (on the left) is colorful with patterns and has a hammock-like seat, clear tapestries, and many hanging plants, but it lacks a mosaic mirror.
Midjourney's image (on the right) is lighter and more spacious, with plants and a lace tapestry, but its bed isn't hammock-styled, and there's no visible mosaic mirror.
While both images have boho elements and hanging plants, neither fully matches the prompt, especially regarding the mosaic mirror and the exact hammock bed description.
High context prompts
Prompt: A blacksmith's workshop during the Renaissance, with detailed tools, glowing forge, intricate armor pieces, and a craftsman at work
The left one by Dall-E has one blacksmith, neatly organized tools, and highlighted armor. The right one by Midjourney has multiple people, scattered tools, and a lively atmosphere. While both depict the workshop, the Dall-E image focuses on a single craftsman and his tools, and the Midjourney one feels more like a busy day with multiple workers.
Low context prompts
Prompt: A moonlit dance
Both images showcase a "moonlit dance." The left image by Dall-E has a vibrant blue tone with silhouetted dancers against a big moon, while the one by Midjourney, on the right, offers a closer, more detailed look at the dancers with a subtler moon glow. Dall-E focuses on the environment and contrasts, and Midjourney highlights the dancers' emotions. Both capture the moonlit dance theme but in different styles.
The showstopper: Midjourney vs Dall-E 3
After evaluating 16 AI-generated images from Dall-E 3 and Midjourney, it's evident that Dall-E 3 excels in capturing intricate details. This platform also surpasses Midjourney in interpreting the intent of prompts to generate relevant images. On the other hand, Midjourney has an edge in crafting visuals that look real. While Dall-E 3 aims for perfection, it can sometimes produce less natural images.
For businesses looking for detail in their AI visuals, Dall-E 3 might be the more suitable choice. You can access it via ChatGPT Plus and also in Photosonic, the best AI image generator, very soon. OpenAI plans to release the Dall-E 3 API soon, making it an integrated feature in Photosonic.
FAQs
1. Is Midjourney better than DALL-E 3?
It's not really about one being outright "better" than the other. They have different styles and capabilities. DALL-E 3 is integrated with ChatGPT Plus and is part of the package you get with GPT-4. Midjourney, on the other hand, might offer variations in its renderings. It's more about your personal preference and the style you're looking for.
2. Is DALL-E 3 free?
No, DALL-E 3 isn't free. It's bundled with ChatGPT Plus, which costs $20/month. This subscription also grants you access to GPT-4.
3. Which is cheaper, DALL-E 3 or Midjourney?
Looking strictly at the numbers, Midjourney starts at a cheaper price of $10/month. DALL-E 3 comes with ChatGPT Plus, which is priced at $20/month. So, if budget is a key factor, Midjourney might be your more cost-effective option.
