How to write Midjourney prompts: 50+ cool examples
We live in an amazing time, witnessing artificial intelligence revolutionizing content creation. Neural networks are rapidly becoming a part of our lives, replacing not only objects in photos but also people in some instances. The first field that has already been impacted by this progress is design. To obtain a unique image, all you need to do is provide a textual description, and the neural network will handle the rest. Here, we are primarily talking about Midjourney.
In our article, we explain how to input Midjourney prompts, what to pay attention to, and what non-obvious features are available.
What are Midjourney prompts and why are they needed?
A prompt is a textual description that helps the neural network understand what is required of it. If you provide something very simple, such as the word “tiger”, the result will be quite predictable:
Such request execution is very similar to one of the Instagram filters, and this is the basic perception of the neural network. To obtain truly unique images, queries need to be as detailed as possible. We will explain how to do this further.
How to write Midjourney prompts?
A proper prompt has a set of characteristics. Include everything you consider important because whatever you don’t describe, the neural network will interpret on its own. And it is not as receptive to implicit information as the human brain.
Here are some other things to pay attention to:
Punctuation. In most cases, Midjourney does not pay attention to punctuation marks. However, we strongly recommend using them to structure your queries, at least in your mind. Here’s what the neural network understands:
- — (double hyphen) – to separate parameters, for example — 16:9 (aspect ratio), — v 5 (neural network version), — no + object that should not be in the image.
- :: – to determine the weight of a fragment in the final image. Let us explain a bit: if a query consists of multiple parts, you can separate them with these double colons followed by numbers. The higher the number, the more important this fragment will be for the final generation.
Unfortunately, we couldn’t remove the wheels. (car —v 5 —ar 16:9 —no wheels).
Here are the stars – completely. (adidas logo in space —v 5 —ar 16:9 —no stars).
Style. Any image can be given a style: an artist’s, an art movement, an era, or animation. For example, a lion in the style of Van Gogh (Van Gogh style painting of a lion —v 5 —ar 16:9).
Photorealism. Midjourney excels at creating images that are practically indistinguishable from photographs. To achieve such an effect, you need to select:
- Camera type and lens: SLR for film cameras or DSLR for digital cameras;
- Lens: 18, 35, 50, or even 250mm;
- Camera model: Canon, Nikon, Sony, and so on.
Subject. In previous examples, there was only one main character. If there are several, it’s better to be specific. Not “several” or “a group”, but “a pack of three lions”, for example. Or “a little girl with a corgi”, as in this image (Photo with a Canon R5, 50mm, DSLR, of a little girl with a corgi —v 5 —ar 3:2 —no camera; “—no camera” is needed so that the neural network does not add the camera itself to the image).
Well, not quite a corgi. And somewhere, not quite a girl.
Environment. It also needs to be described in as much detail as possible. Midjourney gives you a fantastic opportunity to combine the uncombinable. For example, sending a pack of tigers for a walk in Moscow.
And yes, sometimes even the smartest AI struggles with counting to three.
Weather effects. The description of weather and lighting also greatly affects the final result. Here are the same three tigers in the same location on a rainy day (three tigers in Moscow, rainy afternoon —v 5 —ar 3:2).
Other aspects. In general, you can add anything you like to the description. For example, shooting distance, angle, level, and so on. The final result will depend solely on your imagination and how clearly you have formulated your request.
50+ best Midjourney prompts
And now let’s see how all these recommendations will look in practice when creating images from popular categories.
Prompts for portraits and photos
Midjourney can create hyper-realistic portraits and photos that are sometimes impossible to distinguish from shots taken with a real camera. For this, you will need key words:
- eyes/ears/forehead and other body parts you want to focus on;
- dynamic pose;
- inspired by;
- with an emphasis on;
- 4/8/16k resolution.
The prompt “A portrait of a young moody man in a futuristic style, monochromatic, with golden lighting, creating a nostalgic and dreamy atmosphere —v 5 —ar 3:2” – the result is simply amazing.
Always keep in mind what you want to achieve in the end and shape your thoughts accordingly. Combine different combinations, experiment, and gradually you will start getting exactly the portraits and photos you envisioned.
Prompts for photo editing
You can’t directly upload a photo for editing – you’ll need to go through a few steps. The most challenging part is uploading the photo to a Discord channel and copying the link to it. After that, everything is done according to the usual scenario: at the beginning of the prompt, you paste the link and describe what you want to do with the photo. You can use any techniques from other sections described in our article: change the background, adjust lighting and style, add details, make it colorful or black and white, and so on.
Here are some words that can help you:
- and any others — depending on what you want to do.
Here’s what we got from an image using the prompt “monochromatic”:
Not exactly what was intended, but okay nonetheless.
There are neural networks specifically designed for processing photos. You can read about them here.
Prompts for architecture and interior design
Midjourney has become a real discovery for architects and interior designers. They can get a multitude of ideas for creating their own masterpieces in a particular style with just one sentence. To do this, it’s enough to formulate a request correctly using the following words:
- interior/exterior design;
- lit up at night;
- barn hous.
Would you like to work in such a creative space? The prompt refers to “an isometric spiral-shaped office building with a Helix design, by Frank Lloyd Wright”.
In fact, there’s a huge scope for creativity here. Just think about how many styles you can blend and see the results with minimal investments. And don’t forget about specific details like a solar panel roof or a black leather sofa. You can also draw inspiration from well-known architects whose styles you’d like to incorporate.
Prompts for creating logos and graphic design
Designers who create graphic works at Midjourney have also made life easier. Especially during moments of creative decadence. To pass through it faster, here is a list of important words:
- vibrant colours;
- circular vortex;
- vector art;
The prompt “minimalist geometric logo of red arrow vector graphic” would be suitable for both gaming apps and delivery services.
Here you can truly unleash your creativity and create truly unique logos and designs. Describing them here is easier than for portraits or interiors, simply because the level of detail is much lower.
Prompts for animation and pixel art
In the life of almost every adult, there comes a moment when they want to set a cartoon character as their avatar. The only thing that could stop them is that such images were usually not unique. Or take a cartoonist who’s tired of inventing new characters. Yes, they probably know how they should look, but they can’t come up with anything. Neural networks can help here as well. And to make it easier for you, we’ve prepared a list of words for art and animations:
- old games;
- spooky environment;
- low polygon;
- game sprite;
- Unity/Unreal Engine.
Would you play as this character? The prompt refers to “steampunk samurai, pixel art, style of old games —v 5”.
As in any creative process, you can combine absolutely everything here: modern game engines with low-polygon models and Disney with anime.
3 tips on creating cool and unique prompts for Midjourney
It often happens that you’ve either tried everything or nothing comes to mind. For such situations, we’ve prepared three tips to help your inspiration break through the deadlock.
Exploring thematic sources for inspiration
In the age of the internet and unrestricted access to information, there are numerous places where people share their works. Check out Instagram, Pinterest, and similar platforms: combine different ideas and try different prompts.
If you’re feeling lazy, you can always borrow from the greats, as everything has already been in this world at some point. There are various libraries for this purpose, like Phraser, where people share generated illustrations along with the prompts they used.
Combining styles for unconventional ideas
Incompatible styles may seem incompatible at first glance. However, the end result can surprise everyone and even go viral. For example, the combination of Renaissance and anime.
Just a dream, not a waifu from the prompt “Mona Lisa anime style”.
Or an abstract chair.
It all depends on what you like and what you want to achieve in the end. The neural network will take care of the rest.
Using random and unexpected combinations
Marvel has a comic series called “What If…”, where they place familiar characters in completely unfamiliar circumstances. Midjourney can do the same, but you don’t even have to leave your home for your favorite Star Wars character to end up in the Middle Ages. Or Harry Potter and his friends from Rowling’s books could move to Dostoevsky’s world.
Despite the fact that sometimes the results may not be exactly what we expect, neural networks are constantly improving. They incorporate new algorithms, settings, and parameters that allow them to better satisfy user requests. What the future holds remains uncertain. It’s possible that artificial intelligence will replace many professions, as it has with fast-food cashiers. However, a true revolution will only be said to have occurred when it replaces accountants, a profession that has been predicted to decline for about three decades.
How to use Midjourney prompts to create visual content?
Using Modjourney prompts is very straightforward: all you need to do is enter a textual description of what you want as the end result in the input field. The final output will depend entirely on the detail of your description.
How to get a description of a liked image from Midjourney?
There’s a special command for this: /describe. After entering it, a button will appear for uploading the image you like. Then, the neural network will automatically offer you four descriptions of what it sees in the picture. While it might not be a perfect match, the generated image will closely resemble the original.
How to write Midjourney prompts using ChatGPT?
To do this, select a subject for your prompt, such as a car, and ask ChatGPT to identify its key components through the command “Break the composition of [car] photography into key elements and write them separated by commas, without explanation. Make near 10 key elements”. After that, fill in these components with the values you want to depict through Midjourney and complete the prompt.
How to bypass restrictions in Midjourney?
Bypassing restrictions can be challenging because as of March 31st, there are no more free or trial periods. So, the simplest option is to either rent an account or subscribe yourself. Telegram bots usually ask users to subscribe to channels and almost immediately start demanding payment for generating content. It’s nearly impossible to verify whether the final result was generated by Midjourney or another neural network.
How can you make money with Midjourney?
If you have your own account in the neural network, you can rent it out or generate images for money. In other cases, you can modify what the neural network creates and sell it, or use it, for example, to save on advertising logos.
How to upload your photo to Midjourney?
To upload your photo for processing with specified parameters, you need to upload it to hosting (direct uploading is not possible), then copy the link, and specify what you want to get as the result.