How to Use Midjourney for AI Art

Midjourney is a powerful and increasingly popular AI art generator that allows users to create stunning images from text descriptions (prompts). Unlike some other AI art tools, Midjourney primarily operates through a Discord bot, which can be a bit different for newcomers.

Here’s a step-by-step guide on how to use Midjourney for AI art:

Step 1: Get Access to Midjourney

Midjourney is accessed via Discord.

1. Join the Midjourney Discord Server:
   • Go to Midjourney’s official website.
   • Click on “Join the Beta” (or a similar button) to get an invite link to their Discord server.
   • You’ll need a Discord account. If you don’t have one, you’ll be prompted to create one.
2. Understand Subscription Tiers:
   • Midjourney used to offer a free trial, but this is often unavailable due to high demand. You will likely need to subscribe to a paid plan to generate images.
   • Check their website or the Discord server’s #announcements or #billing channels for current pricing and plans. Paid plans offer more “fast GPU time” (which means quicker image generation), commercial rights, and other features.

Step 2: Navigate the Midjourney Discord Server

Once you’re in the server:

1. Find the “Newbie” Channels:
   • On the left sidebar, look for channels named #newbies-XX (e.g., #newbies-1, #newbies-42). These are dedicated channels for new users to generate images and learn.
   • You can generate images in any of these channels. The public feed can be very fast-moving, so your creations might get buried quickly, but it’s a great way to see what others are creating and learn from their prompts.
2. Alternatively, use Midjourney with a Private Bot or Your Own Server:
   • Once you have a paid subscription, you can invite the Midjourney bot to your own Discord server. This allows you to generate images privately without your creations getting lost in the public feed.
   • To do this, find the Midjourney Bot in the Midjourney Discord server’s user list (usually on the right), click its profile, and select “Add to Server.”

Step 3: Start Generating Images with Your First Prompt

This is where the magic happens!

1. Use the /imagine Command:
   • In a #newbies-XX channel (or your private server), type /imagine and then press Space or Enter.
   • A prompt: field will appear. This is where you describe the image you want to create.
2. Write Your Prompt:
   • Be Descriptive: The more detail you provide, the closer the AI will get to your vision.
   • Keywords & Phrases: Use specific nouns, adjectives, verbs, and artistic styles.
   • Structure: Think about:
     • Subject: What is the main thing? (e.g., “A majestic dragon”)
     • Action/Scene: What is it doing or where is it? (e.g., “flying over a misty mountain range”)
     • Style/Artist: What artistic style? (e.g., “digital painting, fantasy art, by Frank Frazetta”)
     • Lighting/Atmosphere: (e.g., “golden hour, ethereal glow”)
     • Details: (e.g., “intricate scales, glowing eyes”)
   • Example Prompt: /imagine prompt: A majestic dragon flying over a misty mountain range, digital painting, fantasy art, golden hour, ethereal glow, intricate scales, glowing eyes --ar 16:9 --v 5.2
3. Add Parameters (Optional but Recommended):
   Parameters are modifiers you add to the end of your prompt to control specific aspects of the image generation. They start with two hyphens (--).
   • --ar <width>:<height> (Aspect Ratio): Defines the image’s width-to-height ratio.
     • --ar 1:1 (Square, default)
     • --ar 16:9 (Widescreen, like YouTube)
     • --ar 9:16 (Portrait, like phone wallpapers)
     • --ar 3:2 (Common photo ratio)
   • --v <version> (Model Version): Specifies which Midjourney model to use. Newer versions often have better results.
     • --v 5.2 (Current default and recommended for most uses)
     • --v 4 (Older, distinct style)
     • --niji (Anime/illustration focused model, often used with --style expressive)
   • --seed <number>: Uses a specific seed number to generate similar images. Good for iterating on a concept.
   • --s <number> (Stylize): Controls how “artistic” or opinionated the AI is. Lower values are more literal, higher values are more creative (default varies by version).
   • --chaos <number>: Influences how varied the initial grid of images is. Higher values result in more diverse options.
   • --no <something>: Tells the AI to try and avoid certain elements.
     • --no text, words, blurry
   • --cref <URL> (Character Reference – NEW!): Use an image URL to reference a character’s features. (Check for latest feature rollouts as Midjourney constantly updates).

Step 4: Interpret and Refine Your Results

After submitting your prompt, Midjourney will generate four initial images in a grid.

1. Understand the Buttons:
   • U Buttons (U1, U2, U3, U4): “Upscale” a selected image. This creates a larger, more detailed version of that specific image. (U1 is top-left, U2 top-right, U3 bottom-left, U4 bottom-right).
   • V Buttons (V1, V2, V3, V4): “Variations” of a selected image. This creates a new grid of four images that are similar in style and composition to the chosen image, allowing you to explore different iterations.
   • Redo/Reroll Button (circular arrow): Reruns the original prompt to generate an entirely new set of four images.
2. Iterate and Upscale:
   • If you like one of the images, click its corresponding U button to upscale it.
   • If you like the idea of an image but want to explore variations, click its V button.
   • Once an image is upscaled, you’ll see more options for it, such as:
     • Vary (Strong) / Vary (Subtle): Create more dramatic or slight variations of the upscaled image.
     • Zoom Out 2x / Zoom Out 1.5x: Expands the canvas, filling in the new areas with AI-generated content that matches the existing image.
     • Custom Zoom: Allows you to input a custom prompt for the zoomed-out areas.
     • Pan (Left, Right, Up, Down): Extends the image in a specific direction.
     • Web: Opens the image in your Midjourney web gallery for better viewing and downloading.

Step 5: Save and Manage Your Art

1. Save Images:
   • Once an image is upscaled, click on it to open it fully.
   • Right-click (on desktop) or long-press (on mobile) and select “Save Image” or “Download Image.”
   • You can also find all your generated images in your Midjourney web gallery (https://www.midjourney.com/app/).
2. Organize and Favorite:
   • In your web gallery, you can “favorite” images, view your prompt history, and organize your creations.

Advanced Tips for Better Midjourney Art:

• Be Specific but Concise: Use powerful keywords. Avoid conversational language.
• Emphasize with Weighting: You can give more importance to certain words by adding :: after them (e.g., magical::3 forest::1).
• Learn from Others: Spend time in the public #newbies or #general channels. When you see an image you like, click on it, then click the “Web” button below it to see the prompt used. This is invaluable for learning prompt engineering.
• Experiment with Blending: Use the /blend command to combine 2-5 images into a single new creation.
• Use Image Prompts: Include URLs of images in your prompt to influence the style or composition. /imagine prompt: <image-url> a futuristic city
• Master the Parameters: Experiment with --ar, --s, --chaos, and --v to get different results.
• Join the Community: The Midjourney community on Discord is active and helpful. Don’t be afraid to ask questions.
• Read the Documentation: Midjourney has extensive documentation (often linked from their Discord server) that explains all commands and parameters in detail.

Midjourney offers a truly creative experience. With practice and experimentation, you’ll be able to generate incredible AI art that matches your vision.