Mastering Midjourney: The ultimate step-by-step guide for beginners

Not to long ago we talked about Midjourney, this absolutely mind-blowing AI art generating tool which took the internet by surprise.

If you’re stepping into the world of AI-powered image creation, you’re in the right place. This comprehensive guide will walk you step-by-step as we head through the beautiful landscape of Midjourney together. With easy-to-understand explanations and examples, we’ll discover everything from prompts to model versions and how to “attack” this AI beast as a beginner.

Using Midjourney requires you to use Discord. This can unlock a unique dimension of creativity, making image generation and sharing a fun and interactive experience with the community. If you’re new to Midjourney or Discord, don’t worry; this guide will have you up and running in no time at all.

Step 1: Invite the Midjourney bot to your server

Midjourney can be used in various modes, but all of them are tied to the Discord application.

Discord can be downloaded as an app for mobile or Windows / Mac, but it can run in your browser without requiring a installation as well. Simply go to www.midjouney.com, sign up and let’s start using Midjourney.

After that, the easiest and most fun way to use Midjourney is to invite the Midjourney bot to your own Discord server. To do that, you need to:

1. Go to Midjourney’s official website or search for ‘Midjourney Discord’ channel.
2. Click on the Midjourney Bot and after click the “Add to Server” button.
3. Choose your server from the dropdown menu and click “Continue”.
4. Accept the permissions requested by the bot, and click “Authorize”.

Voila! The Midjourney bot is now on your own server and you can ask it to create art right away.

Step 2: Prompting Midjourney on Discord

With the Midjourney bot on your server, you can start generating images. All commands should start with / for the bot to recognize them. Let’s explore a few examples.

Keep in mind that all prompts related to image generation in Midjourney start with /imagine.

Single-Word Prompts

Want to see an image inspired by ‘rainbow’? Simply type /imagine rainbow, and wait for the magic to happen.

Artistic Medium Prompts

If you desire a sketch of a dog, your prompt would be /imagine sketch of a dog.

Time Travel Prompts

For an image in the style of the 80s, type /imagine 80s cat illustration.

Emotive Prompts

For an image infused with joy, your command could be /imagine joyful cat.

Colorful Prompts

To see a blue cat, use the prompt /imagine blue cat.

Environmental Prompts

To visualize a cat by the seaside, prompt Midjourney with /imagine beach cat.

Remember, the bot responds to your command in the text channel you’re using, so make sure you’re in the appropriate channel when you generate your images.

Step 3: Blending Images

Blending images with the Midjourney bot is as simple as using the /blend command. You will need to upload two or more images. Here’s how:

1. Type /blend in the chat.
2. Discord will prompt you to upload the first image.
3. Once the first image is uploaded, you’ll be prompted to upload the second image.
4. If you want to add more images (up to five), upload them when prompted.

Blended images default to a 1:1 aspect ratio, but you can adjust this according to your needs.

Step 4: Experimenting with Model Versions

Experimenting with different model versions is as simple as appending --v followed by the model number at the end of your prompt. For example, if you want to use Model Version 5 for a specific image, your command would be /imagine sketch of a cat --v 5.

For the Niji Model, the command will look like /imagine anime style cat --niji 5.

Step 5: Fine-Tuning with Style Parameters

Lastly, you can fine-tune your images using style parameters. For instance, you can adjust the style of Model Version 5.1 by typing /imagine sketch of a cat --v 5.1 --style raw. If you’re using the Niji Model and want a cute style, you’d type /imagine anime style cat --niji 5 --style cute.

Step 6: Understanding Image Versions

As Midjourney continues to evolve and improve, different versions of the AI model are released. Each version has unique characteristics that affect the output images. For example, Version 5.1, released in May 2023, has a stronger default aesthetic, higher coherency, and supports advanced features. However, Version 5 might give you more photographic results, while Version 4, the default from Nov 2022–May 2023, excels with image prompts.

For a touch of anime and illustrative styles, the Niji model is a great choice. It’s a collaboration between Midjourney and Spellbrush, offering a robust understanding of anime aesthetics.

To use a particular version, you just need to add --v or --niji followed by the version number at the end of your prompt. For instance, to use the Niji Model, the command would look like /imagine anime style cat --niji 5.

Step 7: Upscaling Images

For times when you need a higher resolution image, Midjourney offers an upscaling feature. Here’s how you can upscale an image:

1. Generate an image as per the steps above. Let’s say we’re generating an image of a cat in anime style using the Niji Model Version 5. The command will be /imagine anime style cat --niji 5.
2. Once the image is generated, you will see options under the image. Click on ‘upscale’.
3. Midjourney will then start to process and upscale your image. This process may take a few moments.
4. Once the upscaling process is complete, a new, higher-resolution version of your image will be posted in the channel.

Please keep in mind that upscaling might not always be necessary. Midjourney’s default resolution is often sufficient for sharing images online. However, if you plan to print the image or use it for a purpose that requires high resolution, then upscaling is a valuable tool.

Step 8: Understanding Midjourney Parameters

Parameters in Midjourney are a powerful way to control the AI’s image generation. They can modify an image’s aspect ratios, switch between different Midjourney Model versions, change the upscaler used, and much more. You always add parameters to the end of a prompt, and you can add multiple parameters to each prompt.

Let’s talk about some basic parameters:

Aspect Ratios (--aspect or --ar): This allows you to change the aspect ratio of the generated image.

If you wanted to generate an image with an aspect ratio of 16:9, you would use the --aspect parameter like this: a beautiful purple dolphin --aspect 16:9.

Chaos (--chaos <number 0–100>): This parameter changes how varied the results will be. Higher values produce more unusual and unexpected generations.

To generate an image with high variation, you could use the --chaos parameter like this: a beautiful beach at night --chaos 80.

Image Weight (--iw <0–2>): This sets the image prompt weight relative to the text weight. The default value is 1.

If you wanted the AI to prioritize the image prompt over the text, you could increase the image weight like this: a mountain at sunset --iw 1.5.

No (--no): This is for negative prompting. For example, --no plants would try to remove plants from the image.

Prompt example: a tropical beach --no palm trees.

Quality (--quality <.25, .5, or 1>, or --q <.25, .5, or 1>): This parameter allows you to control the rendering quality time. The default value is 1.

Higher values use more GPU minutes, while lower values use less.

Prompt example: a bustling cityscape --q 0.5.

Repeat (--repeat <1–40>, or --r <1–40>): This parameter is useful for quickly rerunning a job multiple times, thus creating multiple images from a single prompt.

Prompt example: a colorful farm --repeat 5.

Seed (--seed <integer between 0–4294967295>): The seed number is a starting point to generate the initial image grids. Using the same seed number and prompt will produce similar ending images.

Prompt example: a forest stream --seed 123456789.

Stop (--stop <integer between 10–100>): This parameter is used to finish a job partway through the process. Stopping a job at an earlier percentage can create blurrier, less detailed results.

Prompt example: a busy marketplace --stop 50.

Style (--style <raw, 4a, 4b, 4c, cute, expressive, original, or scenic>): This allows you to switch between different versions of the Midjourney Model and Niji Model.

Example: a fantasy castle --style 4c or to use a style from the Niji model: a fantasy castle --style cute.

Stylize (--stylize <number>, or --s <number>): This parameter influences how strongly Midjourney’s default aesthetic style is applied to the images.

Example: an epic space battle --stylize 500.

Tile (--tile): This parameter generates images that can be used as repeating tiles to create seamless patterns.

Example: a floral pattern --tile.

Remember, you can mix and match parameters as needed, like so: a vibrant sunset --aspect 16:9 --chaos 50 --repeat 3. This would generate three images of a vibrant sunset, with a 16:9 aspect ratio and medium variation.

Step 9: Using different Midjourney model versions

Midjourney routinely releases new model versions to improve efficiency, coherency, and quality. Different models excel at different types of images.

Here are some examples of Midjourney models:

1. Midjourney Model Version 5.1: This is the default and most advanced model as of May 4th, 2023. It excels at producing high-quality images with a strong default aesthetic. You can use it by default without specifying any parameters.
2. Midjourney Model Version 5: This model produces more photographic generations and closely matches the prompt. It requires longer prompts to achieve the desired aesthetic. To use this model, add the --v 5 parameter to your prompt.
3. Midjourney Model Version 4: This model features an entirely new codebase and AI architecture designed by Midjourney. It has increased knowledge of creatures, places, and objects compared to previous models. To use this model, add the --v 4 parameter to your prompt.
4. Niji Model Version 5: The Niji model is a collaboration between Midjourney and Spellbrush, specifically focused on anime-style images. It excels at dynamic and action shots, as well as character-focused compositions. To use this model, add the --niji 5 parameter to your prompt.

To use the Niji model, for instance, add --niji at the end of your prompt.

If you want to use a specific version of the Midjourney algorithm, you can do so by adding --version <1, 2, 3, 4, or 5> or --v <1, 2, 3, 4, or 5> at the end of your prompt.

Step 10: Understanding Compatibility

Not all parameters are compatible with all Midjourney model versions. For example, the ‘Image Weight’ parameter is applicable for versions 4 and 5, but not for Niji 5. The ‘Tile’ parameter affects the initial generation and variations of the images in versions 4 and 5, but not in the Niji 5 version. Knowing the compatibility of parameters with different versions helps you use them effectively.

Understanding and effectively utilizing these parameters will allow you to have finer control over your image generation. Enjoy the added depth of creativity that these tools offer as you continue your journey with Midjourney on Discord!

Wrapping thing up

In summary, Midjourney is an exciting AI-powered tool that empowers users to create unique and captivating images from simple text prompts. With its intuitive interface and powerful image generation capabilities, Midjourney opens up a world of creative possibilities for both beginners and experienced artists.

We also discussed the notion of image variations, where distinct Midjourney models excel at generating specific categories of images. From the default Model Version 5.1 with its robust inherent style to Model Version 4 and the specialized Niji Model Version 5 for anime-inspired creations, each model offers distinctive capabilities to cater to diverse artistic preferences.

Additionally, we delved into Midjourney parameters that enable users to fine-tune their image generation process. From modifying aspect ratios and levels of unpredictability to managing image prominence, quality, and recurring patterns, these parameters grant flexibility and command over the creative outcome.

Bear in mind the importance of respect and adhering to the Midjourney’s Code of Conduct, when writing prompts. Make sure that the produced images align with community guidelines and foster a positive atmosphere for all.

With Midjourney, the possibilities are boundless. Whether you’re a digital artist, a content creator, or simply seeking to embark on a journey of creativity, Midjourney provides a formidable platform to bring your imagination to fruition.

So, immerse yourself in the paradise of Midjourney, unleash your ingenuity, and allow the bot to metamorphose your text prompts into breath-taking visual masterpieces.