Badge Webflow Award Winner 2023

Complete Guide 2023: Use Midjourney AI to create impressive images

Our complete 2023 guide to mastering Midjourney AI and generating exceptional images. Become an expert in this revolutionary tool by following the tips and advanced techniques in this documentation.

Published on 
3/7/2023
-
Amended on 
10/7/2023
-
5 min
Several images generated by Midjourney with the AI logo

In the world of visual creation, images play an essential role. They are present in many fields, including graphic design, advertising, fashion, art and, of course, web design. However, generating quality images can sometimes be a complex and time-consuming challenge.

That's where Midjourney comes in. Midjourney is an innovative platform that uses artificial intelligence to generate images automatically. Whether you're a creative professional or an enthusiast looking to bring your ideas to life, Midjourney offers a powerful and efficient way to create attractive, high-quality images.

The advantages of using Midjourney are numerous. As well as saving you precious time, MidJourney lets you explore endless creative possibilities. You can create realistic, stylized or whimsical images, manipulating various attributes such as color, texture, shape and much more. Midjourney also features a gallery of examples that serve as a real source of inspiration, allowing you to discover new ideas and stimulate your creativity.

At first glance, the tool may seem simple to use: you write a sentence and an image is created, but it actually hides dozens of features that will enable you to go even further in image generation. In this guide, we'll take you through the main features of the tool, enabling you to generate even more precise and advanced images. You'll learn how to exploit pre-trained templates, customize the attributes of generated images and fine-tune results for unique compositions.

1. Quick links

The heart of Midjourney's use lies in the creation of prompts. A prompt is simply a description that guides the artificial intelligence model in the desired direction. The Midjourney bot breaks down the words and phrases of a prompt into smaller elements, called "tokens", which can be compared with its training data and used to generate an image. Prompts therefore play a fundamental role in the generation of quality images , and with Midjourney you'll be able to create more or less advanced ones.

1.1 Simple prompt

image of a simple prompt on midjourney

Simple prompts are a basic approach to guiding the model through image generation. They usually consist of a sentence or short description that describes the general idea of the image you want to generate. For example, if you want to create an image of a seascape, your simple prompt might be "A beautiful sunset over the ocean with crashing waves."

To create an image using a prompt :

  1. add Midjourney Bot in private message on Discord
  2. enter the command "/imagine
  3. add your prompt
💡 Tips: the Midjourney bot works best with simple, short sentences that describe what you want to see.

1.2 Advanced Prompt

Image of an advanced prompt on midjourney

Advanced prompts offer greater flexibility and more precise control over the image generation process. They allow you to fine-tune the specific attributes and characteristics of the generated image. For example, you can add an image to your prompt, add separators to the same prompt to create "multi-phrases" or add parameters (see point 6. Using parameters) to adjust ratios, aspects, quality, etc.

💡 Tips : intégrer un style, une couleur, une période, un lieux, etc. dans vos prompts en utilisant les symbole : <>. Ex : /image <pink> style cat

1.3 How to write a prompt

images of several pink cats
  • Prompt length: Prompts can be simple, but a more descriptive prompt is preferable to achieve a unique rendering. Focus on the main concepts you want to create.
  • Grammar: The Midjourney bot doesn't understand grammar, sentence structure or words the way a human would. Word choice is also important. More specific synonyms usually work better. Use commas, parentheses and dashes to organize your ideas, but be aware that the Midjourney bot won't interpret them reliably. The Midjourney Version 4 model is, however, slightly better at interpreting traditional sentence structure.
  • Focus on what you want : It's better to describe what you want than what you don't want. For example, if you ask for a "cake-free" party, your image will probably include a cake. To exclude a specific object from the final image, you can use the "--no" parameter when generating advanced prompts.
  • Think about the important details: Anything left unspecified may surprise you. Be as specific or vague as you like, but note that anything you don't mention will be random. Be clear about the context or details that are important to you, such as the subject (: person, animal, character, place, object, etc.), the medium (photography, a painting, illustration, etc.), the environment (café, forest, cityscape, beach, etc.), the lighting (warm light, sunlight, etc.), the location (location, location, etc.).), lighting (warm light, dark atmosphere, bright light, etc.), colors (bright, vivid, monochromatic, etc.), mood (calm, serene, animated, joyful, etc.) and composition (facial expression, wide shot on landscapes, etc.).
  • Use collective nouns: Plural words leave something to chance. Try using specific numbers. For example, "three cats" is more precise than "cats". Collective nouns also work well, like "a group of birds" instead of "birds".
💡 Tips: use the /describe command to ask MidJounrey to generate sample prompts based on one or more images.

2. Prompt + Image

prompt & image midjourney

Midjourney lets you manage an image from another image. So you can integrate an image into your prompts.

  1. Upload an image to Midjourney and send it as a message to the bot
  2. Run the /imagine command
  3. Retrieve the URL of this image by dragging and dropping it into your message.
  4. Write a prompt after the URL
💡 Tips: you can adjust the degree of importance of the image in relation to the text in your prompt using the"--iw" parameter. Add a ratio to the parameter to give more importance to either the image or the text. Ex: --iw 0.75, --iw 1, --iw 1.25, etc.

3. Image + Image

blend functionality in midjourney with two images

It is also possible to combine several images to merge them together. To do this, use the /blend function rather than /imagine.

  1. Add the /blend command
  2. Upload at least 2 images
  3. If required, add other images with the Image3, Image4, Image5 options
  4. The default ratio is (1:1), i.e. a square image. It is possible to adjust this ratio using the "dimension" option and the ratios (2:3) for a portrait format or (3:2) for a landscape format image.
several mages of pink cats and ocean with waves

4. Using Discord

Direct message

direct messages in MidJourney

Newbies-104, Newbies-105, ... Stop generating your images from the public chat channels, and open a private chat with the bot. This way, you can easily keep track of your images and find them again.

  1. Click on"Member list" at top right
  2. Click on MIDJOURNEY BOT
  3. Send a message from the dedicated field
  4. Find your discussion in your private messages (All your private messages are accessible from the "Private messages" button at the top left of your Discord interface).

Image grid

image grid in Midjourney

The /imagine command generates a grid of 4 image options. You can then choose the image that best matches your request using the buttons below:

  • U1 U2 U3 U4 : the U buttons for "Upscale" allow you to select the image you like to obtain a final, larger, higher-quality version.
  • V1 V2 V3 V4 : the V buttons for "Variation" generate 4 new images based on one of the 4 latest versions on offer.
  • Redo icon: the Redo icon lets you restart the generation of 4 new images from the initial prompt.

Feedback with Emoji

React with emojis on your images to trigger certain actions. For example:

  • ❌ : Delete an image
  • ✉️: send a direct message with image details

5. Order list

  • /ask: Get an answer to a question.
  • /blend : Enables you to easily merge two images.
  • /daily_theme: Toggles notification pings for #daily-theme channel updates.
  • /docs: Used in the official Midjourney Discord server to quickly generate a link to the topics covered in the Midjourney doc.
  • /describe: Writes four sample Prompts based on an image you download.
  • /faq: Used in the official Midjourney Discord server to quickly generate a link to the FAQ.
  • /fast: Switches to fast mode.
  • /help: Displays useful basic information and tips about the Midjourney bot.
  • /imagine: Generates an image using a prompt
  • /info: Displays information about your account and pending or current jobs.
  • /stealth: For Pro plan subscribers: switch to stealth mode
  • /public: For Pro plan subscribers: switch to public mode
  • /subscribe: Generate a personal link to a user's account page.
  • /settings: View and adjust Midjourney Bot settings
  • /prefer option: Create or manage a custom option.
  • /prefer option list: View your current custom options.
  • /prefer suffix: Specify a suffix to be added to the end of each prompt.
  • /show: Use an image of the job ID to regenerate the job in Discord.
  • /relax : Switch to Relax mode.
  • /remix: Allows you to activate Remix mode and create variations of a basic image, via a new "Make Variations" button.

6. Use of parameters

parameters in mijourney

In Midjourney, parameters are used to specify certain image-related elements, such as styles, quality, ratio and so on. Parameters are always added at the end of prompts and begin with a double dash --.

  • Ratios --ar : allows you to change the ratio of an image, which by default is 1:1. Ex: --ar 16:9, --ar 7:4, --ar 4:3, etc.
  • Chaos --c : adjusts the level of similarity between the 4 images generated in a grid. The lower the Chaos level (close to or equal to 0), the more similar the images in the same grid will be. The higher the Chaos level (e.g. 80), the more different the images generated by the same prompt will be. Ex: --c 0, --c 20, --c 30, etc.
  • No --no : allows you to exclude elements you don't want to appear when creating the image. Ex: --no fruit, --no blue, etc.
  • Quality --q : defines the time spent to generate an image, and therefore the quality level of this image. The more time the AI spends, the higher the quality of the image, but the more Midjourney credits will be used. In some cases, therefore, you may wish to reduce the image quality level. The default maximum time is 1. e.g. --q .25, --q .5, --q. 1
  • Repeat --r : repeats a command several times. For example, generate several image grids at once. This parameter can be used with values between 2 and 10 for Standard packages and between 2 and 40 for Pro packages. Ex: --repeat 2, --repeat 3, etc.
  • Stop --stop : stops image generation before the end. As the entire image generation path is based on 100, you could, for example, use a --stop 50 to stop image generation at 50% of the path.
  • Stylize --s: adjusts the artistic level of an image. On a scale from 0 to 1000, the --s parameter allows you to either get closer to the prompt (close to 0) or to give yourself more artistic freedom (close to 1000). Ex: --s 200, --s 750, etc.
  • Style --style: defines a style parameter for the image grid generated by Midjourney. Ex: --style original, --style expressive, --style scenic, --style cute, etc.
  • Tile --tile : generates images with "repeating tiles". That is, the image will be an assembly of a single image that has been repeated several times. This is very useful, for example, for wallpaper patterns, textures, fabrics, etc.
  • Version --v : Midjourney continues to evolve and release new versions of the product constantly. By default, you always use the latest version of Midjourney, but you can revert to previous versions by using the --v parameter. Ex: --v 5.2, --v 5.1, --v 5, etc.
  • Video --video: Yes, you read that right, VIDEOS! In addition to images, Midjourney is able to generate short videos in mp4 format, thanks to the --video parameter.
  • Weird --weird : this parameter lets you explore a darker, more unexpected mode, or at any rate one that's less usual in terms of aesthetics. The ratio is between 0 and 3000. 0 being the default value, and 3000 the most unexpected. Ex: --weird 0, --weird 250, --weird 500, --weird 1000
💡 Tips : Orders can be combined with each other

Ready to take your website to the next level?

Improve your online visibility thanks to Digidop's experience, excellence and reactivity!

Stay in touch with Digidop news!