Midjourney Tutorial: The Complete Beginner's Guide to AI Image Generation

Midjourney turns plain English into professional-quality images in under a minute. Whether you want to mock up a product, generate concept art, or simply explore what AI can do with a creative prompt, this tutorial walks you through everything — from creating your account to writing advanced prompts that get you exactly what you're picturing.

No design experience required.

What Is Midjourney and Why Is It the Leading AI Image Generator?

Midjourney is an AI image-generation tool developed by the independent research lab Midjourney, Inc. You describe an image in plain text — a "prompt" — and the model produces four image options in roughly 30–60 seconds.

What sets Midjourney apart from competing tools like DALL·E 3 or Stable Diffusion:

• Aesthetic quality out of the box. Midjourney's default output looks polished without heavy prompt engineering. Its training leans toward cinematic, painterly, and editorial aesthetics that are immediately usable.
• Active community. Millions of images are generated daily and shared publicly (on free plans), so you can study what other prompts produce before writing your own.
• Consistent updates. The team ships major model versions frequently, with each version delivering measurable leaps in realism, coherence, and text rendering.
• No local hardware required. The model runs entirely in the cloud. A mid-range laptop is all you need.

Midjourney is the go-to choice for designers, marketers, indie game developers, and content creators who need high-quality visuals fast.

Getting Started: Discord Setup, Joining the Bot, and Subscription Plans

Midjourney runs inside Discord, the chat platform originally built for gamers. If you don't have a Discord account, the setup takes about three minutes.

Step 1 — Create a Discord Account

Go to discord.com and sign up with your email address. Verify your email. That's it.

Step 2 — Join the Midjourney Server

Visit midjourney.com and click "Join the Beta." This drops you into the official Midjourney Discord server, where you can explore public generation channels and community showcases.

Step 3 — Add the Midjourney Bot to Your Own Server (Recommended)

Public channels move fast and can be distracting. For a cleaner workspace:

1. Create a new Discord server (the "+" icon in your server list).
2. In the Midjourney server, click on the Midjourney Bot in the member list.
3. Click "Add to Server" and select your personal server.
4. Accept the permissions prompt.

Now you can generate images privately in your own server.

Step 4 — Subscribe to a Plan

Midjourney ended its free trial tier in 2023. A paid subscription is required to generate images.

Relaxed mode generates the same quality images but joins a shared queue, which typically adds 1–5 minutes of wait time per image. For most beginners, the Standard plan hits the best balance of speed and volume.

Subscribe at midjourney.com/account.

Your First /imagine Command — Step by Step

Once you're in a Discord channel with the Midjourney Bot active:

1. Click the message bar and type /imagine
2. A prompt field appears — type your description into it
3. Press Enter

Midjourney generates a 2×2 grid of four images. The process takes 30–60 seconds on fast mode.

Your first prompt to try:

/imagine prompt: a golden retriever sitting in a field of sunflowers, warm afternoon light, shallow depth of field, DSLR photography

After the grid appears, you'll see eight buttons below it:

• U1 U2 U3 U4 — Upscale the corresponding image (larger, more detailed version)
• V1 V2 V3 V4 — Create four new variations based on that image
• 🔄 — Re-run the same prompt for a fresh set of four images

Click U2 to upscale the second image, for example. That gives you a single high-resolution image you can download.

Anatomy of a Midjourney Prompt

A prompt is not a sentence. It is a structured list of visual instructions. The more deliberately you construct it, the more control you have over the output.

The Core Elements

Subject — What is the main focus of the image?

"a weathered lighthouse"

Style or Medium — How should it look? Photography? Oil painting? Illustration?

"vintage travel poster illustration"

Mood or Atmosphere — What feeling should the image convey?

"dramatic, stormy, moody"

Technical Parameters — Aspect ratio, version, quality, and other settings (covered in detail below).

A fully structured prompt looks like this:

/imagine prompt: a weathered lighthouse on a rocky coast, vintage travel poster illustration, stormy sky, dramatic lighting, bold colors --ar 2:3 --v 6

You don't need all four elements every time. Even a two-word prompt works. But the more specific and intentional your prompt, the closer the output will match your vision.

Prompt Weight Syntax

You can tell Midjourney to emphasize certain words using :: followed by a number.

/imagine prompt: dragon::2 castle::1 forest

This makes the dragon twice as prominent as the castle in the composition. Negative weights (dragon::-1) remove elements — though the --no parameter (below) is more reliable for exclusions.

Key Parameters Explained

Parameters are appended to the end of your prompt after --. They modify how Midjourney interprets and renders the image.

--ar — Aspect Ratio

Controls the width-to-height ratio of the output.

• --ar 1:1 — Square (good for social media profile images)
• --ar 16:9 — Widescreen (presentations, wallpapers, YouTube thumbnails)
• --ar 9:16 — Vertical (Instagram Stories, TikTok, mobile wallpapers)
• --ar 2:3 — Portrait (book covers, posters, editorial)
• --ar 3:2 — Landscape (standard photography ratio)

--v — Model Version

Specifies which version of the Midjourney model to use.

• --v 6 — The current flagship model (best realism, best text rendering)
• --v 5.2 — Previous flagship; still excellent for stylized work
• --niji 6 — Optimized for anime and illustrated styles

--q — Quality

Controls how much GPU time is spent rendering. Higher quality = more detail, but costs more fast-GPU minutes.

• --q .25 — Fast, rough draft
• --q 1 — Default (good balance)
• --q 2 — Maximum detail (costs 2x the GPU time)

--style — Style Preset

Available in certain versions:

• --style raw — Reduces the model's default aesthetic bias, giving you a more neutral, photographic result. Useful when you want less "AI art" look.

--no — Negative Prompting

Excludes elements from the image.

/imagine prompt: a bowl of pasta on a wooden table --no hands, fork, utensils

--seed — Reproducibility

Every image is generated from a random seed number. If you like an output and want to generate something similar, you can reuse the seed.

To find a seed: React to your generated image with the ✉️ emoji. Midjourney Bot DMs you the seed number.

/imagine prompt: futuristic city skyline at dusk --seed 84729

--iw — Image Weight (for Image Prompts)

Controls how strongly a reference image influences the output (covered in depth below).

8 Prompt Formulas with Real Examples

1. Portrait Photography

/imagine prompt: close-up portrait of a 35-year-old woman with natural curly hair, soft window light, shallow depth of field, shot on Canon EOS R5, editorial fashion photography --ar 4:5 --v 6

2. Concept Art

/imagine prompt: abandoned space station interior, overgrown with alien bioluminescent plants, eerie green light, sci-fi concept art, matte painting style --ar 16:9 --v 6

3. Product Mockup

/imagine prompt: minimalist glass perfume bottle on white marble surface, soft diffused studio lighting, luxury cosmetics product photography, clean background --ar 1:1 --v 6 --style raw

4. Landscape

/imagine prompt: misty mountain valley at sunrise, pine forest reflection in a still lake, golden hour light, landscape photography, National Geographic style --ar 16:9 --v 6

5. Logo / Icon Design

/imagine prompt: flat vector logo of a fox head, geometric shapes, terracotta and cream color palette, clean lines, simple icon design, white background --ar 1:1 --v 6

6. Anime / Illustrated Style

/imagine prompt: a young girl reading a book under a giant glowing mushroom in a magical forest, Studio Ghibli-inspired illustration, soft watercolor textures, warm lighting --niji 6 --ar 3:4

7. Architectural Render

/imagine prompt: modern Scandinavian house surrounded by birch trees, snow on the ground, warm interior light visible through floor-to-ceiling windows, architectural visualization, golden hour --ar 16:9 --v 6

8. Abstract Art

/imagine prompt: abstract fluid art, deep indigo and copper swirls, metallic sheen, high contrast, macro photography style, textured canvas --ar 1:1 --v 6

How to Use Reference Images

Midjourney can use an image you provide as a style or composition reference.

How to Add an Image Prompt

1. Upload your image to Discord (drag and drop into the message bar, then press Enter to send it alone).
2. Right-click the image → "Copy Link."
3. Paste that URL at the beginning of your /imagine prompt, before the text.

/imagine prompt: https://your-image-url.jpg a woman in a flowing red dress, cinematic lighting --iw 1.5 --ar 2:3

--iw Controls the Blend

• --iw 0.5 — The reference image has mild influence; text prompt dominates
• --iw 1 — Balanced between image and text
• --iw 2 — The reference image dominates; text prompt plays a supporting role

You can use two image URLs to blend them:

/imagine prompt: https://image1-url.jpg https://image2-url.jpg blend of both styles --iw 1

This is useful for style transfer: upload a painting you love, add a text description of your subject, and let Midjourney merge the two.

Upscaling and Variations: U and V Buttons Explained

After every generation, you get a 2×2 grid of four images (numbered 1–4, left to right, top to bottom).

U Buttons — Upscale

Clicking U1, U2, U3, or U4 upscales the corresponding image. Upscaling in V6 produces a single, larger image with enhanced detail. This is the image you'll typically download and use.

After upscaling, you get additional options:

• Vary (Subtle) — Small, controlled variations while keeping the core composition
• Vary (Strong) — Bigger variations, useful if you like the concept but want to explore different directions
• Zoom Out 2x / 1.5x — Expands the canvas outward, generating more scene around the original image
• Pan (arrows) — Extends the image in a specific direction

V Buttons — Variations

Clicking V1, V2, V3, or V4 generates four new images based on the composition and style of that grid item — without upscaling. Use V buttons when you like the direction of one image but want to explore iterations before committing to an upscale.

Style Modifiers: Lighting, Camera Types, and Art Movements

Adding specific visual vocabulary to your prompts dramatically improves consistency and quality.

Camera and Lens Types

• shot on 35mm film — grain, warmth, analog feel
• shot on Hasselblad medium format — crisp, wide tonal range, fashion/portrait look
• fisheye lens — distorted wide-angle perspective
• macro photography — extreme close-up, fine detail
• tilt-shift lens — miniature model effect

Lighting Descriptors

• golden hour lighting — warm, directional, cinematic
• Rembrandt lighting — dramatic shadow with a triangle of light on the cheek
• backlit — subject silhouetted or haloed with rim light
• neon-lit — colorful urban night scenes
• overcast diffused light — soft, even, no harsh shadows (great for portraits)

Art Movements and Styles

Rather than referencing living artists (which raises ethical concerns), draw on art movements, historical periods, and medium types:

• Impressionist painting — loose brushwork, light and color over line
• Art Deco illustration — geometric elegance, gold and black palettes
• Brutalist architecture — raw concrete, imposing scale
• Ukiyo-e woodblock print — flat color, bold outlines, Japanese aesthetic
• Bauhaus design — functional, geometric, primary colors
• Baroque chiaroscuro — extreme light/dark contrast, oil painting drama

Combining art movements with subject and technical parameters gives you a huge range of distinct visual styles without guessing what any individual artist's output looks like.

Common Mistakes and How to Fix Them

Mistake 1: Prompts That Are Too Vague

Weak: /imagine prompt: a cool building Better: /imagine prompt: a futuristic glass skyscraper at night, blue and white LED lighting, reflections on wet pavement, architectural photography --ar 9:16 --v 6

Vague prompts produce generic output. Specific prompts produce intentional output.

Mistake 2: Overloading the Prompt

Packing 30 adjectives into a prompt dilutes each one's influence. Midjourney works best with 10–20 meaningful tokens. Prioritize the details that matter most.

Mistake 3: Ignoring Aspect Ratio

The default is square (1:1). If you're generating a phone wallpaper, banner, or poster, set --ar from the start. Cropping an upscaled image after the fact loses resolution and often clips important content.

Mistake 4: Not Iterating

Your first generation is a starting point, not a final product. Use V buttons to iterate on promising directions. Use seed numbers to build on outputs you like. The best results come after 3–5 rounds of refinement.

Mistake 5: Forgetting --style raw for Realistic Photography

V6's default leans slightly painterly and aesthetic. Adding --style raw neutralizes that bias and gets you closer to genuine photographic realism. Essential for product photography and headshot-style portraits.

Midjourney V6 vs. Previous Versions

Midjourney V6 (released late 2023, continuously improved through 2024–2025) is a meaningful leap over V5.2 and earlier models.

What V6 does better:

• Text rendering — V6 can reliably render short text strings within images (signs, labels, logos). Earlier versions were nearly useless at this.
• Prompt coherence — V6 follows complex, multi-element prompts more faithfully. You can describe a scene with several distinct components and expect all of them to appear.
• Realism — Skin texture, fabric, environmental lighting, and material properties are significantly more convincing.
• Composition control — Camera framing instructions ("close-up," "wide shot," "bird's-eye view") are followed more consistently.

When you might still use V5.2:

Some users prefer V5.2's slightly more stylized, "dreamy" output for certain creative applications — illustrated portraits, fantasy art, and painterly landscapes where hyper-realism isn't the goal. The --v 5.2 parameter keeps that model available.

Niji 6 remains the best choice for anime, manga, and highly illustrated outputs. It's a separate model fine-tuned specifically for those aesthetics.

Take Your Skills Further with a Structured Course

Reading a tutorial is a great start. But the fastest way to internalize Midjourney — the parameter combinations that actually work, the prompt patterns that save you hours of trial and error, the workflow from rough concept to polished asset — is structured, hands-on practice.

NextoolAcademy's Midjourney course is built in a Duolingo-style format: short, practical lessons that build on each other, with real prompt exercises and immediate feedback. No fluff, no filler — just the skills that translate directly to creative output.

Start the Midjourney Course at NextoolAcademy →

Frequently Asked Questions

What is Midjourney used for?

Midjourney is used to generate images from text descriptions. Common use cases include concept art for games and films, marketing and advertising visuals, product mockups, book and album cover design, social media content, architectural visualization, and personal creative projects. It is one of the most widely used AI image generators because of its consistently high aesthetic output.

Do I need art or design experience to use Midjourney?

No. Midjourney is designed to be accessible to complete beginners. You describe what you want in plain English, and the model handles all the visual decision-making. Learning a handful of prompt structures and parameters — covered in this guide — is enough to produce professional-quality results without any prior design background.

How much does Midjourney cost?

Midjourney requires a paid subscription. Plans start at $10/month (Basic), with the most popular option being the Standard plan at $30/month, which includes 15 hours of fast GPU time and unlimited relaxed-mode generations. There is currently no free tier available.

What is the difference between fast mode and relaxed mode?

Fast mode uses dedicated GPU time and generates images in 30–60 seconds. Relaxed mode joins a shared queue and may take 1–10 minutes per generation depending on server load, but it does not consume your monthly fast-GPU allocation. Standard plan and above subscribers have access to unlimited relaxed-mode generations.

Can I use Midjourney images commercially?

On paid plans, you generally own the images you generate and can use them commercially. However, Midjourney's terms of service include specific conditions: on the Basic plan, you must check the terms for commercial use restrictions. Pro plan subscribers get full commercial rights and stealth mode (private generations). Always read the current terms at midjourney.com before using images in commercial work, as policies are updated periodically.

What is the best Midjourney version to use?

As of 2025, Midjourney V6 is the best general-purpose model for photorealistic and high-quality stylized images. For anime, manga, and illustrated styles, Niji 6 is the better choice. V5.2 remains useful for users who prefer its specific aesthetic character. Append --v 6 or --niji 6 to your prompts to ensure you are using the model you intend.

Why does Midjourney keep generating the wrong number of hands or fingers?

Hands and fingers are a known challenge for all AI image generators, including Midjourney. V6 has improved significantly over earlier versions, but issues still arise. Workarounds include: cropping the composition so hands are not in frame, using --no hands if hands are not needed, generating at higher quality (--q 2), and using the Vary (Subtle) function to regenerate just the problematic area. For images where accurate hands are critical, using an inpainting tool or manual editing after generation is the most reliable solution.

How do I make Midjourney generate consistent characters across multiple images?

Character consistency across images is one of Midjourney's harder challenges. The most effective methods are: using the --seed parameter to anchor the random generation, including a detailed physical description in every prompt (hair color, facial structure, clothing), and using image prompts (--iw) with your best result as a reference for subsequent generations. Midjourney has also introduced a "Character Reference" feature (--cref) in recent updates, which is specifically designed to maintain character consistency across a series.