AI tools · Image Generation

Midjourney: setup, settings and troubleshooting

Generates original images from text prompts with a distinctive, art-directed style.

What is Midjourney?

Midjourney is an AI image generation tool that turns text prompts into original artwork, illustrations, and photorealistic images. It has no traditional installable app — it runs through its web app (alpha.midjourney.com) and, historically, through Discord bot commands inside the Midjourney Discord server. It's especially known for a strong, painterly default aesthetic that many designers and artists favor over more literal image models.

  • Text-to-image generation with a distinct, high-quality default artistic style
  • Parameter flags like --ar (aspect ratio), --v (model version), --stylize, and --chaos to control output precisely
  • Image prompting and image weight (using a reference image alongside text) plus Style Reference (--sref) for consistent looks
  • Vary, Upscale, Pan, Zoom Out, and Remix tools to refine a generated image instead of starting over
  • Web-based image editor for inpainting/outpainting and retexturing parts of an image
  • Community and personal galleries where every generation (on standard plans) is visible and searchable for prompt ideas

Create an account

  1. Go to midjourney.com and click 'Sign Up' or 'Join the Beta'.
  2. Create an account with an email address (or Google/Discord login) and verify via the confirmation email.
  3. Choose and pay for a subscription plan (Basic, Standard, Pro, or Mega) — Midjourney no longer offers a free trial, so a paid plan is required to generate images.

Install the app

  1. There is no desktop or mobile app to install — Midjourney runs entirely at alpha.midjourney.com in a web browser.
  2. Optionally join the official Midjourney Discord server (link on the website) if you prefer generating images with /imagine slash commands in chat channels or DMs with the bot.
  3. Bookmark alpha.midjourney.com and, if desired, add it to your phone's home screen for quick browser-based access on mobile.

Recommended first settings

  1. Set a default aspect ratio and model version in your web app profile settings so you don't have to type --ar and --v on every prompt.
  2. Review the Community Feed for prompt examples in the style you want before writing your own.
  3. If privacy matters, note that Basic/Standard plans generate publicly by default — Pro and Mega plans include Stealth Mode for private generations.
  4. Create folders in the web app to organize generations by project.

How to use Midjourney

Type a descriptive prompt — subject, style, lighting, composition — into the Imagine bar on the web app (or after /imagine in Discord) and Midjourney returns a grid of image variations within about a minute. From there you pick a result to Vary (subtle or strong), Upscale for higher resolution, or Remix by editing the prompt and regenerating from that image as a starting point.

  • Front-load the most important subject and style words early in the prompt — earlier terms carry more weight.
  • Use --sref with a reference image URL to keep a consistent visual style across multiple separate prompts.
  • Generate a batch, pick the strongest result, then use Vary (Subtle) to iterate instead of rewriting the whole prompt from scratch.

Common problems and fixes

Can't sign in / account says subscription required to generate.
Confirm an active paid plan under Manage Subscription — free trials are no longer offered, so an expired card or lapsed subscription will block generation. Update billing details, then refresh the web app.
Discord bot doesn't respond to /imagine commands.
Make sure the Midjourney bot is actually present in the server/channel you're using and that you have an active subscription linked to that Discord account. Try DMing the Midjourney Bot directly, or switch to the web app at alpha.midjourney.com, which doesn't depend on Discord at all.
Generated images look nothing like the prompt, or quality is inconsistent.
Simplify the prompt to its core subject and style first, then add modifiers one at a time to see what's shifting the result. Check that --v is set to the latest model version, and consider lowering --chaos if outputs are too random.
Web app is slow or stuck generating ('Waiting to start').
This usually means the queue is busy at peak hours; higher-tier plans (Pro/Mega) get priority (Fast/Turbo) GPU time. Try again in a few minutes, or switch from Relax mode to Fast mode if your plan includes it.

Best practices

  • Write prompts as a clear sentence describing subject, setting, style, and lighting rather than a pile of disconnected keywords.
  • Use parameters (--ar, --v, --stylize, --chaos) deliberately instead of copy-pasting them from someone else's prompt without understanding what they do.
  • Iterate with Vary and Remix on a promising result instead of regenerating from scratch every time.
  • Turn on Stealth Mode (Pro/Mega) for client or unreleased work you don't want appearing in the public feed.

Frequently asked questions

Do I need Discord to use Midjourney?

No. Midjourney originally only worked through Discord, but it now has its own web app at alpha.midjourney.com where you can prompt, edit, and manage images without touching Discord.

Is there a free version of Midjourney?

Not currently — Midjourney requires a paid monthly or annual subscription (Basic, Standard, Pro, or Mega) to generate images; there is no ongoing free tier.

Who owns the images Midjourney generates?

Paying subscribers generally own the images they create, subject to Midjourney's Terms of Service; company-size subscribers above a revenue threshold are required to be on Pro or Mega plans. Always review the current Terms of Service for specifics.

What does --v actually control?

It selects which underlying Midjourney model version generates the image; newer versions typically improve realism, prompt accuracy, and detail, while older versions can still be used for their distinct look.

Other AI tool guides