How to Use Midjourney: Discord and Web App Guide
Midjourney turns a line of text into a set of finished images, but getting your first result takes a few setup decisions that the tool never explains up front. There is no free tier, you reach it through two different doors, and the parameters that separate amateur output from polished work are tucked behind cryptic flags. This guide walks you through the whole path: what you need before you begin, how the Discord and web app routes differ, the basic generate-and-refine loop, and the handful of parameters that do most of the heavy lifting.
Before anything else: Midjourney has no free tier and no free trial. You need a paid subscription, starting at $10 per month, before you can generate a single image. We cover the plans below.
Before You Start
Midjourney is subscription-only. Since March 2023 there has been no permanent free tier and no free trial, so the very first requirement is a paid plan. Everything else follows from the access path you choose.
One requirement that no longer applies: Midjourney used to make new users generate 1,000 images on Discord before unlocking the web app. That gate has been removed. New subscribers can go straight to either door.
Discord or Web App: Pick Your Door
Midjourney gives you two ways in, and they share the same account and the same image library. The choice is mostly about workflow comfort. Discord is conversational and fast for rapid-fire prompting; the web app is a more conventional interface with a proper editor.
The Discord path
You can use Midjourney inside Discord three ways: join the official Midjourney server, send a direct message to the Midjourney bot for a private workspace, or invite the bot to a server you control. All three use the same /imagine command. Direct-messaging the bot is the cleanest option for solo work because your prompts and results are not buried in a busy public channel.
The web app path
The web app at midjourney.com consolidates creating and editing into one screen. You type prompts in a create bar, browse your full history, and open any image in an editor that supports pan, zoom, Vary Region, and inpainting. Because the two paths sync, an image you make in Discord shows up on the web, and vice versa.
| Consideration | Discord | Web App |
|---|---|---|
| Prompt entry | /imagine command | Create bar |
| Editing tools | Basic upscale and vary | Pan, zoom, Vary Region, inpainting |
| Browsing history | Scroll the channel | Organized gallery |
| Best for | Fast iterative prompting | Refining and editing |
| Library | Shared | Shared |
The Discord Workflow
The Discord workflow centers on one command. Once you are subscribed and in a channel where the bot lives, you type /imagine, and a prompt field appears. Write a description, send it, and the bot replies with a grid of four image options.
After the grid appears, a row of buttons sits beneath it. The buttons let you act on the four results:
- Upscale turns one of the four small previews into a single larger, finished image.
- Vary generates a fresh grid of four based on the option you liked, nudging the result in a new direction.
- Re-roll runs the same prompt again from scratch for a completely new grid.
This loop, prompt then grid then upscale or vary, is the heart of using Midjourney. You rarely get a perfect image on the first try; you steer toward it by picking the closest option and varying from there.
The Web App Editor
The web app does everything Discord does and adds a real editor on top. When you open an image in the editor, you get tools that let you change part of a picture without regenerating the whole thing.
What the editor gives you
| Tool | What It Does |
|---|---|
| Pan | Extends the image outward in a chosen direction, generating new scenery that matches the existing frame |
| Zoom | Pulls the camera back to reveal more context around your current image |
| Vary Region | Lets you select an area and regenerate only that part of the image |
| Inpainting | Paints a masked region with new content based on a fresh prompt, leaving the rest untouched |
Vary Region and inpainting are the tools you reach for when an image is 90 percent right but has one distracting flaw, an extra finger, an awkward object, a color that clashes. Instead of re-rolling and losing everything you liked, you mask the problem area and regenerate just that patch.
Why the editor matters: Re-rolling a whole prompt is a gamble; you might lose the parts that worked. Region-based editing keeps your good results and fixes only the bad spots, which is how most finished work actually gets made.
Your First Image, Step by Step
Here is the basic loop end to end. The same five steps apply whether you start in Discord or the web app, because both share the same engine and the same library.
Step 1: Subscribe
Pick a plan and subscribe. The Basic plan at $10 per month is enough to learn the workflow; you can upgrade later for more generation time.
Step 2: Choose your door
Open Discord and find the bot, or go to midjourney.com. Either path works with the same account.
Step 3: Write a prompt
In Discord, type /imagine and describe what you want. In the web app, type the same description in the create bar. Be specific about subject, style, and mood.
Step 4: Review the grid
Midjourney returns four options. Look for the one whose composition and feel are closest to your goal, even if the details are not perfect yet.
Step 5: Upscale or refine
Upscale the best option for a finished file, or vary it to explore alternatives. In the web app, open it in the editor to pan, zoom, or fix a specific region.
Key Parameters Worth Knowing
Parameters are flags you add to the end of a prompt to control how Midjourney interprets it. You do not need many to get good results. These four cover most of what beginners reach for, and they all attach the same way, as a double-dash flag after your text.
| Parameter | What It Controls |
|---|---|
| --sref | Style Reference: point at an image so Midjourney borrows its palette, texture, and atmosphere |
| --cref | Character Reference: keep the same character consistent across multiple images |
| Image Weight | How strongly an attached image influences the result versus your text prompt |
| --stylize | How heavily Midjourney's default aesthetic is applied to your prompt |
A related editing parameter, Vary (Region) combined with Remix, has been available since version 5.2 and underpins the region editing you reach through the web app. The exact numeric ranges for stylize and image weight are documented in the official guides, and because they shift between versions it is worth checking the current docs rather than memorizing values.
Limits and Gotchas
A few things trip up new users, and most of them come down to how Midjourney charges for compute and which plan you are on.
Troubleshooting
These are the issues newcomers run into most often, with the fix for each.