Welcome to this tutorial on Midjourney, an AI image generator created by David Holz in 2022.

If you're new to AI, this guide will walk you through everything you need to know about using Midjourney effectively.

Key Objectives:

• Understand what Midjourney is and how it was developed.
• Learn how to get started with Midjourney.
• Explore basic usage of Midjourney.
• Discover practical applications and best practices for using Midjourney.

What's Midjourney?

Midjourney is an AI image generator that enables anyone to create images, visuals, icons... from text. Co-founded by David Holz, Midjourney launched its open beta in July 2022 and has been rolling out updates to enhance its product ever since.

You can interact with Midjourney through a Discord bot by typing things like "/imagine a futuristic city skyline".

Recently, Midjourney made things even more user-friendly with a web app, offering an interactive platform for generating and tweaking AI images.

Although it’s a proprietary tool, Midjourney’s AI-generated art has captured significant attention, making its way into major publications like The Economist and sparking conversations about the ethics and copyright of AI-created artwork. To access this powerful service, users can subscribe, supporting the company’s goal of making innovative image generation accessible to all.

Midjourney offers four paid subscription plans:

1. Basic Plan: $10 per month or $96 annually
   • Includes 3.3 fast GPU hours per month (around 200 image generations)
   • No access to relaxed/unlimited mode
2. Standard Plan: $30 per month or $288 annually
   • 15 fast GPU hours per month
   • Unlimited image generations via relaxed mode
3. Pro Plan: $60 per month or $576 annually
   • 30 fast GPU hours per month
   • Unlimited relaxed mode
   • Stealth mode to keep images private
4. Mega Plan: $120 per month or $1,152 annually
   • 60 fast GPU hours per month
   • Unlimited relaxed mode
   • Stealth mode

In this tutorial we will use the basic plan, which offers enough for most use cases (except if you plan to generate images every day).

Getting started with Midjourney

Getting Started with Midjourney on Discord

Sign up here to Discord if you don't already have an account. (Discord is required to access and interact with the Midjourney bot).

Once it's done, sign up or login to Midjourney.

If you don't have a paid plan yet, you can subscribe here by choosing a subscription tier that suits your needs (remember that the $10/month plan is enough for most use cases).

Then, join the Midjourney Discord server by opening Discord.

Navigate to a #general or #newbie channel in the Midjourney server. These are the channels where you can interact with the Midjourney bot.

Use the /imagine command to generate images by typing "/imagine" followed by a text description or prompt of what you want to create.

For example: "/imagine a futuristic city skyline"

‍

Getting Started with the Midjourney Web App

At the time when I write this article, you need to have generated over 1,000 images on Discord to get early access to the web app.

First, go here and log in with your Discord account credentials.

Use the text input box at the top to enter image prompts, similar to the Discord /imagine command. Remember that the /image command is not needed on the web app.

Main features

If you use the Discord bot, these features will appear when available just under the generated pictures as buttons.

Rerun (🔄)

The rerun or re-roll button (🔄) allows you to generate a new set of images based on the original prompt, without modifying it. This is useful when you want to explore more variations of the same concept.

Upscale (U1, U2, U3, U4)

The U buttons (U1, U2, U3, U4) upscale the corresponding image to a higher resolution (1024px by default). This increases the detail and clarity of the selected image.

Vary (V1, V2, V3, V4)

The V buttons (V1, V2, V3, V4) generate new variations of the corresponding image grid, while attempting to maintain the overall composition and style. This allows you to explore subtle changes to the original image.

Vary (Subtle) and Vary (Strong)

After upscaling an image, you can use "Vary (Subtle)" to create minor variations that closely match the starting image, or "Vary (Strong)" for more distinct variations.

Vary (Region)

The "Vary (Region)" option allows you to select a specific area of an upscaled image and generate variations focused on changing that region, while keeping the rest of the image relatively unchanged.

Reframe

The reframe options like "Zoom Out" and directional arrows (Pan) enable you to extend the boundaries of an upscaled image, revealing more of the scene around it.

Remix Mode (/prefer remix)

Remix Mode allows you to edit the original prompt text, model version, or parameters when creating variations or using other refinement options. This gives you more control over how the image evolves with each iteration.

Some basic use cases

Although we will cover the art of prompting with Midjourney, which is very different from ChatGPT, in another part, here are a few basic use cases that you could start with.

Generate a logo

Creating a logo with Midjourney requires a clear understanding of the desired graphic elements and how to articulate them. Using artistic references and avoiding certain details can result in minimalist and effective designs.

Prompt:

‍/imagine Vector graphic logo of frog, simple minimal, by Rob Janoff --no realistic photo details

Generate a Portrait

Generating a realistic portrait with Midjourney involves detailed descriptions to capture the essence and mood of the subject. Ensuring clarity and avoiding contradictory elements is crucial for high-quality renders.

Prompt:

‍/imagine Blonde beauty, bathroom, soap in hand, by the window, shelf on the bathtub, bright sunlight, fresh colors, minimalist style, beautiful image, smiling, happy, comfortable state of mind, scene image, simple, bokeh effect, portrait fixed focal length lens shot, photographic texture, 8k HD,::1

Generate an Architecture Photo

Midjourney can create stunning photorealistic architectural images by specifying unique design features and settings. Including details like lighting and structure can enhance the realism and impact of the image.

Prompt:

‍/imagine exterior of a cutting-edge sports arena that features dynamic LED lighting and a retractable roof, convenient for some of the world's biggest sports events --ar 16:9

Generate a Web Design

Designing a website with Midjourney involves outlining the layout and style elements clearly. Specifying features such as call-to-action buttons, image quality, and design aesthetics helps create a modern and engaging web interface.

Prompt:

/imagine Design of an e-commerce website with a modern and edgy design. The home page features a full-screen hero image with a call-to-action button, and the product pages are designed with clean lines and high-quality images to showcase the items.

Generate an Anime-style image

Creating an anime-style image with Midjourney involves detailed descriptions of characters, settings, and stylistic influences. This approach ensures that the final image captures the essence of anime art, complete with vibrant scenes and dynamic compositions.

Prompt:

‍/imagine Still frame of a short but piloted humanoid mecha with squat legs standing by a car, residential American neighborhood, daytime, full view, anime, 1999