Artificial intelligence has been taking a victory lap these past few months. You’ve probably heard of that little AI conversational chatbot, ChatGPT. In the wake of this incredible technology being made available to the public, we've been treated to some pretty awesome tools that push the boundaries of possibility. One such example of this is Midjourney.
If you're a casual observer who's thought about bringing your wildest ideas to life through digital art but lacks artistic skills, you're in for a treat.
Or, you might want to create slick-looking graphics for your business's website, social media content or landing page, this AI tool will blow your mind.
Or, if you have no idea what all of this is and have never heard of Midjourney, don’t worry.
We’ll be breaking down what it is, what it can do and how you can use it, in this blog.
Midjourney is an independent research lab dedicated to "exploring new mediums of thought and expanding the imaginative powers of the human species".
Yeah, it is very sci-fi indeed.
What this small team of mad scientists have produced is also called Midjourney, a text-to-image generative AI program.
This basically means it can create captivating images from simple text inputs or "prompts". Similar AI image generators exist, such as DALL-E 2 and Stable Diffusion.
In order to use Midjourney, you’ll need to create or log in to a Discord account.
You can find the option for this under "Join the Beta" in the bottom right of the homepage.
Once you're in, join a newcomer group where you'll be able to start prompting. It's worth noting that currently, Midjourney is only offering paid plans for conjuring digital art.
Plans range from $10 to $60 per month, with discounts available on annual subscriptions.
If you've had a chance to use ChatGPT, you're halfway there. With ChatGPT, you submit a prompt to an AI chatbot and you get a response. That response is essentially an amalgamation of billions of data points sourced from across the internet transformed into a coherent reply.
Midjourney uses the same natural language processing and machine learning technologies to formulate a response, but instead of seeing a text reply, you're presented with an image.
The inner workings of these are mind-bogglingly complex, but using Midjourney is an astonishingly simple feat. It all happens in seconds.
Just type a prompt and wait for Midjourney to send back an image.
In practice, it works like this. 👇
Let’s say you type blacklight volcano, this is what you'll get:
Pretty neat, right?
Now, if your prompt game is a little stronger, you can tap into more complex commands like this: diagrammatic drawing of the structure of Tokyo Tower.
Let’s see what we get:
Notice how a more complex prompt results in a more complex image.
With Midjourney, you’re free to create pretty much anything your mind can imagine.
From photorealistic images to futuristic cityscapes and bustling city streets to detailed illustrations, cartoon characters and wild animals.
As you learn how to structure advanced image prompts, you’ll get to play around with different lighting, such as dramatic, cinematic or natural lighting.
You can also adjust standard settings, for example, you can tinker with the aspect ratio and depth of field to enhance your images further.
The only thing you need to remember is this: You must type /imagine before any image prompt.
Then, let ideas rip.
A “prompt" is a short phrase that Midjourney interprets to produce an image.
When you type a prompt, the AI bot dissects the words and phrases into smaller pieces called tokens.
These are cross-referenced against Midjourney’s training data and an image is produced that best represents your request.
Prompts can be strikingly simple (which works well) but Midjourney has the ability to interpret granular detail too.
We'll explore the many types of prompts you can try over the next few sections to give you a guide on what to do, but the real beauty of AI tools like Midjourney is in the trial and error and discovery of creating stunning visuals.
Let's go! 🚀
Basic prompts are nice and straightforward, even for the artistic illiterates. They can be a single word, phrase or even an emoji.
Note: the /imagine text that precedes the prompt (ALWAYS type this first), then add whatever words you want Midjourney to generate after.
Advanced prompts give you wiggle room to unleash your creative desires fully.
Midjourey is a very clever AI tool and as such, it can break down prompts that contain multiple image URLs, text phrases and up to several parameters (more on those in a minute).
This means you can really test Midjouney to generate a high-quality piece of art that is almost indistinguishable from a professional artist or photographer.
Now, we start getting into the more advanced aspects of Midjourney's generative AI model.
Here, you have the option to flesh out ideas by instructing the bot to produce specific artistic styles using a variety of keyword terms.
Let’s look at some of these now.
Let’s say you want your image to convey a certain look, this would be your "style”. You can prompt Midjourney to incorporate different styles into its visuals by adding keywords based on design, an art form, artist, genre or more.
You can change between low and high stylise options by entering - - s <some number>. So if you wanted low, you would type - - s 500 or for high, you would type - - s 2000.
Similar to the above, type a value from 0-100 to increase or decrease the level of abstraction in your image. The prompt goes like this "/imagine <text prompt> - - chaos 100"
You can change the output resolution of your images by specifying the quality with keywords. Use prompts that include 4K, 8K, HD render, ultra-detailed, ultra-photoreal, intricate details, photorealistic, ultra-sharp and more. Look how great this child’s portrait is, the detail is stunning!
There are standard settings for simpler, more predictable outcomes by denoting "HD" or "quality”.
For example /imagine tree - - hd. Or /imagine tree - - quality 8.
Unless you want a standard square every time (which you might), you need to tell Midjourney the aspect ratio parameters it should adhere to.
So if you want a larger image for wallpapers or a cinematic view or anything in between, you have to type /imagine <text prompt> - - ar <aspect ratio>. You might choose standard ratios like 4:3, 16:9, 5:1, or others, but you must specify the size for the correct image output.
If you want a custom size, then simply enter the desired height and width like this /imagine <text prompt> - - w 100 - - h 400, for example. The image below has an aspect ratio of 2:3.
Let’s say you want to create a beautiful work of art that is similar to an image you’ve come across online. We’ll call that "inspiration ".
To give consistent results across multiple images, you can use the image URL in your prompt.
You can add multiple image URLs and multiple text prompts for Midjourney to interpret. For example, /imagine <paste URL link> fresh fruit.
If you want your image to look closer to the URL image provided, you can dictate the weight of the image. Like this /imagine <paste URL link> fresh fruit - - iw:2.
You can ask Midjourey to disregard any element of an image that you don't want it to include. For example, if you have an image of beautiful scenery with a few trees that are ruining the shot, you can remove these by typing "- - no trees" in your text prompt.
If you want to create realistic images in Midjourney, then this section is for you. From landscapes to city skylines to wildflowers, bustling cityscapes to coral reefs.
In your prompts, include details about the type of lens, maybe a specific camera like Nikon or Sony, and consider the type of lighting (dramatic lighting, cinematic lighting, volumetric lighting, etc).
Adjust the aperture settings to create a blurry background and use resolution keywords like 4K, 8K, ultra-realistic, etc, to add extreme detail.
Check out these Midjourney prompts for realism, all of which are taken from the official Midjourney Discord channel.
Prompt: A girl takes an English test, sits at a desk with a sheet of paper and writes text to a software company for an HR job, the girl comes for an interview, some cyberbanking, warm tones, a yellow accent, cyberpunk in the background, dark tones in the background, accent on people, realism, photo, v 5.1 --q 5 --chaos 30 --ar 2:1
Prompt: a woman taking a bite of pizza, and her face is amazed, she is chewing, there is food in her mouth, her eyes are big, she looks excited and happy, quiet, photorealism, real life, real people, real woman.
Prompt: medieval farmer’s yellow saxophone incrusted with steel and yellow diamonds; 3d, 4k, realism, photo
Prompt: realistic cityscape from the middle of the street at ground level
Prompt: extremely closeup shot of basketball Ball moving trail of fire, A hyper-realistic image, Octane render, 8K --quality 5 --s 1000 --ar 16:9 --v 5.1 --style raw.
Portraits are notoriously difficult to create. Thankfully, Midjourney makes this a whole lot easier! There’s a lot to consider in your portrait prompts.
Think about the use of lighting to give depth and enhance the image.
Also, pay attention to small details like hair, clothing and textures. And experiment with different colours and artistic styles that draw the eye.
Prompts: a bunch of young Greek guys posing, in the style of arte povera, tabloid photography, precise nautical detail, realist portraits, associated press photo, domestic intimacy, les nabis, colourized, herbert list --ar 4:5 --seed 2023 --v 5.1
Prompt: A realistic portrait of an elf from Dungeons and Dragons appears to be in her late 30s. She resembles Indira Varma. She wears a traditional sari gown. long ears, elf ears, pointy ears. her long hair is braided. the background is a Hindu temple in a jungle
Prompt: A realistic family portrait of the Simpsons if they were human. Hyperrealistic. Shot on film
Prompt: create a realistic portrait of geishas in the Brazilian hinterland, shot in an analogic camera with some grain texture
Prompt: hyper-realistic 3D portrait of a 100% perfect Melanin goddess, queen, flamboyant sass, flawless.
Prompt: a realistic portrait of a young woman with great hair, smiling up at the camera, stock photo, emotionally complex, fast wave, precise detailing, clear edge definition, photorealistic, textural sensations cinematic, in the style of Zeiss batis 18mm f/2.8, white background, soft tonal transitions, pot core, digital gradient blends, grey and amber, accurate and detailed, 8k resolution --uplight --ar 1:1 --v 5.1
Prompt: Done in an abstract realistic portrait with a small white dove next to the man, As I look through the window, I see the old man on the other side a bit away, sitting in the rain, he sits on a bench in the park all alone, without an umbrella for his own, sometimes I wonder if he sits in the rain, to hide his pain, to blend the tears from his eyes with the showers from the skies, was it there in that park where he met his true love? the only thing he really needed in this life, for it was here, where he met his wife, his true love, who, not long ago, flew like a dove to the heavens above.
Ever wanted to expedite a prototype design for products? Totally get it.
With Midjourney, you can generate mind-blowing product photography images faster than ever.
Here’s what to consider in your prompts:
Avoid using “realistic" prompts as they can make your products look digitised, instead, just stick to photography terms to maintain a consistent image quality.
Prompt: branding and packaging for The Flamingo Cocktail, trendy, elegant, South Beach Miami aesthetic, product photography --s 750 --s 250
Prompt: <https://s.mj.run/DrTsMr2V9Ww> Product photography, luxury fragrance, advertising editorial, photography by Zachary Goulko, 3:4 size image
Prompt: The eight side seal coffee bean packaging bag on a light blue background in the style of a perfume advertisement, product photography, luxury bags, advertising, photography, smoky, elegant, men, 4k, realistic, professional --s 750 --v 5
Prompt: product photography, a 3D jigsaw puzzle of Turing era Bletchley Park, in the style of jewellery by painters and sculptors, unique and one-of-a-kind pieces, hand-painted details, grocery art, primitivist, white and amber, meticulously detailed --ar 4:3 --s 750 --ar 5:4 --s 250
Prompt: A black woman, 25% Dutch, 75% African, holding a small lipgloss in her right hand, wearing a white t-shirt, wearing jeans, is captured in an 8k picture with a vibrant, pink background that womans energy is contagious, as she smiles. The lighting is a mix of natural and studio lighting, creating a bright and fresh atmosphere. Photographed by Mario Testino, using a Hasselblad H6D-400c with a 100mm lens, the photo style is a mix of lifestyle and product photography, capturing the womans enthusiasm for the product --v 5
Prompt: 1950s toy for shark teeth, blue shark box, 1950s magazine advertisement, mid-century toys, vintage product photography --ar 4:5 --c 10 --s 1000 --upbeta --v 5.1 --v 5.1
All good brands need a great logo. It's like cheese and toast, they just go together.
Midjourney is an excellent tool that can help generate professional-looking logos for any business with minimal effort.
Keep these pointers in mind before you prompt:
Prompt: The logo is minimalist, with a minimum of detail. Geometry, one colour. Logo for a company that rents out villas
Prompt: brand logo featuring Lion with Strength, Courage, Spoke / Modern in Warm colour palette / Digital art, Flat Vector, Saul Bass style, white background --no realistic photo details --v 5.1
Prompt: An interpretation of a sports club logo in the style of Picasso’s later works, with a basketball, a football, and a tennis racket playfully arranged in a dynamic scene. The Munich Frauenkirche is incorporated into the design as a symbol of the club’s origin, its architecture abstracted into simple yet recognizable shapes. The colour palette is rich and diverse, with strong contrasts and a sense of depth achieved through the use of tones and shades.
Prompt: Design a minimal logo featuring a Rocket in Black on a Cream background. Include Geometric Shapes as an additional design element --s 750 --niji 5.
Prompt: a simple, basic, clean, monochrome, line art logo for a company named Procedure using the letter P only
Prompt: Logo for dog shelter named “Hound Haven”, Siberian Husky, Geometric logo Abstract logo Negative space logo Logo symbol Minimalist logo Modern logo Bold logo Flat logo, adobe illustrator, flat design, shades of green --stylize 750
If you want your images to feature the characteristics of a famous photographer, you can.
By adding the name of a photographer to your prompt, Midjourney will generate an AI image inspired by your chosen artist(s).
Let's have a look at a few examples with their prompts.
Prompt: Subject: One Night in New York City | STYLE: voguing photo by Chantal Regnault and Robert Mapplethorpe
Prompt: the muppets britain’s biggest love saga and the biggest film, froggy and mrs piggy, in the style of light pink and dark emerald, cindy sherman, associated press photo, close-up, award-winning, creative commons attribution, Penelope rosemont --ar 64:47 --v 5.1 --q 2 --style raw --s 250
Prompt: fashion photograph by man ray, 1930s surreal flamingo couture designed by schiaparelli --ar 5:7
Prompt: a chance reflection in a 🪞 by Lee Friedlander --ar 3:2 --v 5
Prompt: Night Photography, Eye-Level, Digital Photography, Canon EOS 5D Mark IV, Brassai, Nightlife Scene in Bandar Abbas, Young Adults in Trendy Clothing, High Contrast, Superior Low Light Performance. --v 5.1 --s 50
Prompt: a photo in the style of Thomas Struth. A large interior space is open to the sky. Water has filled the lower part of the structure. There are two large rockets with what looks like Russian markings. They are similar in design, but one has been partially dismantled. We see everything reflected in the water on the floor. Lonesome. Dystopian. use high-definition textures --are 16:9
As with famous photographers, you can add the flair of your favourite artists to images, too.
Type one or several artist names into your text prompt and wait for the magic to happen.
Let's have a look at a few examples with their prompts.
Prompt: endless universe, buddha meditating, by Agnes Lawrence Pelton, by Akihito Yoshida, by Albert Bierstadt, by Andy Warhol, by Annibale Carracci, by Anthony Van Dyck, Artstation, by Asaf Hanuka, in the style of Assassin's Creed, by Aubrey Beardsley, by Ben Enwonwu, by Bob Byerley, by Camille Corot, by Caravaggio Michelangelo Merisi, by Caspar David Friedrich, by Charlie Bowater, by Claude Monet, by David Mann, by Diego Rivera, by Dorothea Tanning, by Edward Munch, by Fernando Botero, by Frank Frazetta, by Frida Kahlo, by Hassan Mossoudy, by Hayao Miyazaki, by Hazem Taha Hussein, by Hieronymus Bosch, by H.R. Giger, by Ismail Inceoglu, by Ivan Shishkin, by Jeff Koons, by John Constable, by John Kenn Mortensen, by Jose Tapiro Y Baro, by Karol Bak, by Katsushika Hokusai, by Kiki Smith, by Kim Tschang Yeul, by Ko Young Hoon, by Laurie Lipton, by Lee Man Fong, by Lemma Guya, by Lisa Frank, by Mahmoud Sai, by Malika Agueznay, by Maria Sibylla Merian, by Mark Brooks, by Michael William Kaluta, by Nam June Paik, Octane render, by Paolo Uccello, by Pauline Haynes, by Pebble Tay, by Pieter Bruegel The Elder, by Pre-Raphaelite Brotherhood, by Raffaello Sanzio, by Ralph Steadman, by Remedios Varo Uranga, by Rene Magritte, by Richard Dadd, by Roy Liechtestein, by Salvador Dali, by Sonia Delaunay, by Takashi Murakami, by Tivadar Csontváry Kosztka, by Tony Diterlizzi, by Victto Ngai, by Wanda Gág, by Wangechi Mutu, Unreal Engine, by Weta Digital, by Wētā FX, by WLOP, by Yaoy Kusama, by Yoshitaka Amano, by Zeng Fanzhi, --v 5.1 --q 5 --s 800 --chaos 29 --tile.
Prompt: a banana world by salvador dali
Prompt: creepy skeleton with little creepy hearts and things making a big heart that he’s standing with. Crude and childish. Abstract and really iconic in the artist style of Gus Fink and Banksy.
Prompt: Collage of Alan Turing by David Hockney: David Hockney creates a vibrant and playful collage that combines elements of Turing's life and achievements. The artwork incorporates images of Enigma machines, mathematical formulas, and colourful geometric patterns, symbolizing the intersection of art and science. --v 5.1
Prompt: a poster of free apes in the style: Karl Lagerfeld, Prada, James Rosenquist – Pop Art Daniel Arsham – Architectural Art Liu Bolin – Invisible Art Ai Weiwei – Political Art Banksy – Street Art Jenny Holzer – Conceptual Art --up beta --q 2 --s 750
Prompt: ball in the style of de stijl and Jenny Holzer
Oftentimes, you might find yourself in the position where you have a clear idea of the image you want to create but you just can't find the words... or prompt, to generate it.
That’s where prompt generator tools come in.
These handy applications can save you from staring into space by suggesting prompts for Midjourney and other AI image-generation models.
Simply enter your text into the field and the website will generate a prompt for you to suit Midjourney's bot.
AIPRM is actually a Chrome Extension that runs off the power of ChatGPT. With it, you can access a library of templates and pre-made prompts. Just type a phrase or keyword and AIPRM will use ChatGPT to generate four Midjourney prompt suggestions.
Write your ideas and thoughts into the main text field and Midjourney Prompt Helper will generate a prompt to paste into Midjourney. To help refine your prompt, you can select from multiple categories such as style, lighting, depth of field, camera and more.
.
As you can see, Midjourney is a mind-blowing AI text-to-image tool that lets you bring your wildest creations to life.
Its applications are many, from casual users to professionals working in creative and design fields.
Models like Midjourney are making art more accessible for people without the necessary artistic skills and for those that do, they are able to produce digital art in a fraction of the time and flex their expertise and knowledge.
To get started, register for Discord and use some of the tips we’ve covered in this blog to generate exciting content!
How does Midjourney work?
Midjourey works by taking your text prompt and transforming it into digital art. You can describe what you would like the computer model to generate, and it will identify keywords and phrases within your text to produce something as close to your request as possible. You can also submit images to further support your request.
How can I write a Midjourney prompt?
There are a few tips for this.
1. Use ChatGPT! Or another AI chatbot to generate a prompt for Midjourney.
2. Use simple and concise language.
3. Use photography terms in relation to lighting and atmosphere, for example, "dramatic, cinematic, vibrant etc”.
4. Describe the overall scene and environment. Imagine it like a novel. Keep it simple, but be descriptive.
Is Midjourney free?
No. You can subscribe to monthly plans ranging between $10-$60 per month. If you choose an annual plan, Midjourney does offer 20% off your subscription fee.
*sources