Midjourney guide: getting started and what paid plans add

Midjourney is an AI image generator made by an independent research lab of the same name. You describe what you want in a text prompt, and the model produces a set of images that you can then refine, vary, upscale or edit. It is known for a strong default aesthetic, which makes it popular with designers, illustrators, marketers and hobbyists who want striking visuals without a long production process.

The service changes often. New model versions arrive with different strengths, features are added or renamed, and plans are adjusted. This guide sticks to the concepts that have stayed stable, so check Midjourney's official documentation for the current details of any specific feature.

Generating images generally requires a paid subscription; free trials have only been offered from time to time. The public gallery on Midjourney's website shows images shared by other users together with their prompts, and studying it is one of the best ways to learn what the model does well.

Midjourney began as a bot on Discord, and that route still works. You join the official Midjourney server, open a channel where the bot is available or send it a direct message, and type the imagine command followed by your prompt. You can also add the Midjourney bot to a Discord server you manage, which gives a team a quieter shared space. Commands for settings, account information and your subscription work the same way.

The website offers a more comfortable workspace. You sign in, type your prompt into the prompt bar, and results appear in a feed that you can organize into folders, filter and search. Editing tools, the public gallery and personalization options are easier to reach there, and many people now use the website for everyday work while keeping Discord for community channels.

Your subscription belongs to your Midjourney account rather than to one interface, so you can switch between the two. If you sign in with Discord, protect that Discord account with a strong password and two-step verification, since whoever controls it can also reach your Midjourney images and plan. rStock is not affiliated with Midjourney or Discord, so follow the terms of use of both services.

A good prompt is clear rather than long. Start with the subject, then add the details that matter: the medium, such as a photograph, watercolor, 3D render or line drawing, followed by the setting, lighting, color palette, mood and composition. A prompt like ceramic teapot on a wooden table by a rainy window, soft morning light, muted greens, close-up product photo gives the model far more to work with than a nice teapot.

Midjourney does not follow instructions the way a chatbot does, so describe what should appear in the image rather than writing commands. Words near the start of a prompt tend to carry more weight, and piling up synonyms such as beautiful, stunning and gorgeous adds little. If an unwanted element keeps appearing, describe the alternative you want or use the exclusion parameter covered below.

You can also guide results with images. An image prompt uses a picture as a starting point for content and composition, while reference features let you borrow a visual style or keep a character or object consistent across several images. Newer versions can also learn your taste through personalization, based on images you rate or like. Change one thing at a time between attempts so you can see what actually made the difference.

Parameters are short options added at the end of a prompt, each starting with two hyphens, and many of them also appear as switches in the website's settings panel. They control how an image is generated rather than what it shows.

The ones most people use early on are --ar for aspect ratio, followed by a ratio such as 16:9 for a wide image or 2:3 for a portrait; --no, which lists things to leave out; --stylize, which controls how strongly Midjourney applies its own artistic taste; --chaos, which makes the first set of images more varied; --seed, which helps reproduce similar results; and --v, which selects a model version. A raw style option produces more literal, less stylized images.

Accepted values and defaults vary between model versions, and some parameters exist only in certain versions. When a parameter does not behave as you expect, check which version you are using and compare it with the current parameter list in the official documentation.

Each job produces a small grid of images. From there you can upscale a favorite to get a larger and more detailed version, create subtle or strong variations and, depending on the model version, zoom out to extend the scene or pan to add space on one side. Remix lets you change the prompt when you make variations, so you can alter details while keeping the composition of an image you already like.

The editor lets you select part of an image and regenerate only that area with a new description, or expand the canvas in any direction. Newer releases have also added video features, such as animating a still image. As with everything else, what you can use depends on your plan and the current version, so treat the official feature list as the reference.

Midjourney sells several subscription tiers that can be billed monthly or yearly. They share the same core models and tools; the differences are mostly about how much you can generate and under what conditions.

The main variable is fast generation time, the amount of processing time used for jobs that run at full speed. Higher tiers include more of it and allow more jobs to run at the same time. Some tiers also include a relaxed mode for images, where jobs wait in a slower queue without using up fast time, which suits people who create a lot and are not in a hurry. Extra fast time can usually be bought if you run out.

The top tiers add a stealth option that keeps your creations out of the public gallery; on other plans, images are generally visible to other users on Midjourney's website. Plan choice also matters for business use, because Midjourney's terms require larger companies to be on a higher tier. Choose based on how often you generate, whether privacy matters for client work and who you are creating for, and check the current plan comparison before you subscribe.

Midjourney's terms of service explain who owns the images you create and under what conditions. In general, paying subscribers own their images to the extent the law allows, with exceptions described in the terms, such as the rule for larger companies. Read the current version before using images for a client, a product or advertising, and follow Midjourney's terms of use and community guidelines at all times.

Owning an image under Midjourney's terms is not the same as having full copyright protection. In some countries, including the United States, authorities have taken the position that material generated entirely by AI without enough human creative input cannot be registered for copyright. That can matter if you need exclusive rights to a logo or another key brand asset, so get legal advice for important commercial work.

Use the tool with care. Avoid prompts meant to copy a living artist's distinctive work or to reproduce trademarked characters and logos for commercial use, and do not create misleading images of real people. When AI images appear in advertising or in news-like contexts, labelling them as AI-generated is good practice and is required on some platforms.