← Back to Blog
["midjourney""AI art""prompts"]

"The Complete Midjourney Prompt Guide: From Beginner to Pro"

7/2/2026

# The Complete Midjourney Prompt Guide: From Beginner to Pro

Midjourney has transformed digital art creation. With the right prompt, you can generate stunning, professional-quality images in seconds. But the gap between a beginner's first attempt and a pro's polished output is entirely about prompt engineering. This guide takes you from your first Midjourney prompt to advanced techniques that produce gallery-quality AI art.

## What Is Midjourney and How It Works

Midjourney is an AI image generation model that converts text prompts into images. You describe what you want in natural language, and the model interprets your description to create a visual. Midjourney is known for its artistic, painterly aesthetic — it tends to produce more stylized, visually striking images compared to photorealistic models.

The key to great Midjourney output is understanding that your prompt is a set of instructions interpreted by the model. Every word matters. Learning how to structure those words is the core skill of Midjourney prompt engineering.

## The Anatomy of a Midjourney Prompt

A well-crafted Midjourney prompt has several components. You don't need all of them every time, but understanding each part helps you control your results.

### 1. Subject

The subject is what the image is about. Be specific:

- **Weak**: "a house"
- **Better**: "a weatherbeaten fisherman's cottage on a rocky cliff"
- **Best**: "a weathered two-story fisherman's cottage with a sagging roof and a single lit window, perched on a rain-soaked rocky cliff"

The more specific and evocative your subject description, the more control you have over the final image.

### 2. Style

Style tells Midjourney how the image should look. You can reference art movements, specific artists, aesthetics, or visual qualities:

```
Style keywords to try:
- photorealistic, hyperrealistic
- oil painting, watercolor, gouache, ink wash
- impressionist, art nouveau, brutalist
- cinematic, editorial, documentary photography
- isometric illustration, flat design, pixel art
- cyberpunk, steampunk, vaporwave
```

Example: `a forest at dawn, oil painting in the style of Thomas Cole, luminist lighting`

### 3. Composition and Camera

You can direct the composition like a photographer or cinematographer:

```
- close-up, medium shot, wide shot, aerial view, worm's-eye view
- golden ratio composition, rule of thirds, centered composition
- shallow depth of field, deep focus
- shot on 85mm lens, shot on 35mm, GoPro footage, drone shot
- Dutch angle, over-the-shoulder, tracking shot
```

Example: `a lone figure on a mountain ridge at golden hour, wide shot, aerial view, shot on 24mm lens, dramatic clouds`

### 4. Lighting

Lighting dramatically changes the mood and quality of your image:

```
- golden hour, blue hour, overcast, harsh midday sun
- volumetric lighting, god rays, dappled light
- studio lighting, rim lighting, three-point lighting
- neon glow, candlelight, chiaroscuro
- backlighting, silhouette, lens flare
```

Example: `a jazz club interior, volumetric lighting, smoky atmosphere, warm amber glow from pendant lamps`

### 5. Mood and Atmosphere

Words that convey emotion help Midjourney set the right tone:

```
- serene, contemplative, melancholic
- tense, foreboding, ominous
- whimsical, playful, dreamlike
- gritty, raw, dystopian
- ethereal, otherworldly, transcendent
```

### 6. Parameters

Midjourney uses parameter flags (added at the end of your prompt) to control technical aspects:

```
--ar 16:9       Aspect ratio (try 1:1, 16:9, 3:2, 9:16, 21:9)
--chaos 0-100   Variation in results (higher = more unexpected)
--stylize 0-1000  How much Midjourney's aesthetic is applied (default 100)
--quality 0.25-2   Render quality/time (default 1, higher = more detail)
--niji 6         Niji model for anime/manga style
--seed 12345     Reproducible results with the same seed
--no [text]      Negative prompts — exclude elements
--tile           Creates seamless tileable images
```

## Beginner Prompts: Getting Started

Start with simple, descriptive prompts and gradually add layers of detail. Here's the progression:

### Level 1: Basic Subject

```
a cup of coffee on a table
```

Result: A basic, somewhat generic image. It works but lacks personality.

### Level 2: Add Style

```
a cup of coffee on a wooden table, watercolor painting, soft lighting
```

Result: Much more artistic. The watercolor style gives it a distinctive look.

### Level 3: Add Composition and Lighting

```
a cup of coffee on a rustic wooden table, watercolor painting,
close-up shot, soft morning light through a window, steam rising,
warm and cozy atmosphere --ar 3:2 --stylize 150
```

Result: A polished, specific image with a clear mood and composition.

### Level 4: Add Atmosphere and Parameters

```
a steaming cup of black coffee on a weathered oak table by a
rain-streaked window, watercolor painting with loose brushstrokes,
close-up shot, soft diffused morning light, melancholic and
contemplative atmosphere, muted earthy tones --ar 3:2 --stylize 200 --chaos 10
```

Result: A gallery-quality image that looks intentional and evocative.

## Intermediate Techniques

### Using Artist References

Naming artists helps Midjourney match a specific aesthetic:

```
a desert landscape with mesas and a winding river, in the style of
Albert Bierstadt, luminist painting, epic scale, warm golden light --ar 16:9
```

Popular artist references to explore:

- **Greg Rutkowski** — fantasy illustrations, dramatic lighting
- **Wes Anderson** — symmetrical composition, pastel palette
- **Zaha Hadid** — flowing architectural forms
- **Hayao Miyazaki** — lush, hand-painted fantasy landscapes
- **Annie Leibovitz** — dramatic portrait photography
- **HR Giger** — biomechanical surrealism
- **Alphonse Mucha** — art nouveau, ornamental style

### Combining Multiple Styles

You can blend styles for unique results:

```
a futuristic city skyline, blend of cyberpunk and art deco,
neon lights reflecting on wet streets, geometric skyscrapers
with organic curves, moody atmosphere --ar 21:9 --stylize 250
```

### Negative Prompts

Use `--no` to exclude elements you don't want:

```
a minimalist living room with a single sofa, natural light,
clean lines --no people, clutter, plants, text, watermark
```

This is especially useful for removing common artifacts like text or watermarks that the model might add.

### Using Image Prompts

You can use reference images as a starting point. Paste an image URL at the beginning of your prompt:

```
[https://example.com/reference-image.jpg] a mountain landscape
in the style of this image, oil painting, dramatic lighting --ar 16:9
```

The model uses the reference image as an influence on style, color, and composition.

## Advanced Techniques

### Multi-Prompt Blending

You can blend separate concepts using the `::` separator:

```
a robot::2 a flower::1 in a surreal landscape --ar 3:2
```

The numbers after `::` are weights — the robot is weighted twice as much as the flower.

### Permutation Prompts

Generate multiple variations in one command using curly braces:

```
/a imagine a portrait of a {young, old, weathered} {man, woman} in {golden hour, blue hour, neon light}
```

This generates 18 variations (3 ages × 2 genders × 3 lightings), letting you explore quickly.

### Character Consistency with `--cref`

For consistent characters across images, use the character reference parameter:

```
a knight in a tavern, storytelling scene --cref [character image URL] --cw 100
```

`--cw` controls how much of the character is referenced (100 = full including clothing and style; 0 = face only).

### Style Reference with `--sref`

To match a visual style across multiple images:

```
a mountain village at dawn --sref [style reference image URL] --sw 1000
```

`--sw` controls how strongly the style is applied (0-1000).

### Controlling Stylization

The `--stylize` parameter is one of the most powerful and underused tools:

- **`--stylize 0`**: Literal interpretation. Your prompt is followed very closely, with less artistic flair from Midjourney.
- **`--stylize 100`** (default): Balanced artistic enhancement.
- **`--stylize 500`**: More artistic, more "Midjourney-looking."
- **`--stylize 1000`**: Highly stylized, dramatic, painterly.

Experiment with this parameter — it dramatically changes results and is rarely used by beginners.

## Prompt Templates by Category

### Portraits

```
Portrait of [subject], [age] [gender], [expression], wearing [clothing],
[setting/background], shot on [lens] at [aperture], [lighting style],
[mood word] --ar [ratio] --stylize [value] --no text, watermark
```

Example:
```
Portrait of a blacksmith with soot-stained hands, 50-year-old man,
tired but proud expression, wearing a leather apron, standing in
a dim forge with glowing embers, shot on 85mm at f/1.8, dramatic
chiaroscuro lighting, dignified and stoic mood --ar 4:5 --stylize 150
```

### Landscapes

```
[landscape subject], [time of day], [season/weather], [art style],
[composition], [lighting], [mood], [color palette] --ar [ratio] --stylize [value]
```

Example:
```
a misty valley with terraced rice fields at dawn, autumn morning fog,
Chinese ink wash painting style, wide aerial view, soft volumetric
light, serene and timeless mood, monochrome with subtle green tones
--ar 16:9 --stylize 200
```

### Product Photography

```
[product] on [surface/background], [lighting setup], [lens/aperture],
commercial product photography, [color palette], [mood] --ar [ratio] --no text
```

Example:
```
a minimalist wristwatch with a black dial on a concrete surface,
softbox studio lighting with a single highlights, shot on 100mm
macro at f/8, commercial product photography, monochromatic palette,
luxurious and precise mood --ar 1:1 --stylize 50 --no text, watermark
```

### Abstract Art

```
Abstract image evoking [theme/emotion], [art style or technique],
[color palette], [composition], [texture] --ar [ratio] --stylize [value]
```

Example:
```
abstract image evoking the feeling of falling through space,
liquid acrylic painting technique, deep blues with gold accents,
swirling radial composition, smooth glossy texture --ar 1:1 --stylize 800
```

## Troubleshooting Common Issues

### Problem: Hands Look Wrong

This is a known issue with AI image models. Mitigations:

- Frame the shot so hands aren't visible (close-up on face, wide shot)
- Use `--no hands` if hands aren't needed
- Specify hands in natural positions: "hands resting on the table, fingers interlaced"

### Problem: Too Much Text in Image

Use `--no text, words, letters, watermark, signature` in your prompt.

### Problem: Results Are Too Generic

Add more specific style keywords, artist names, or technical photography terms. The more precise your vocabulary, the less generic the output.

### Problem: Can't Get a Consistent Style Across Images

Use `--sref` with a style reference image, or save a seed value with `--seed` to reproduce similar results.

## Building a Prompt Library for Midjourney

Professional Midjourney users don't write from scratch every time. They build libraries of template prompts with placeholders:

```
[SUBJECT] in [STYLE] style, [COMPOSITION], [LIGHTING],
[MOOD], [COLOR PALETTE] --ar [RATIO] --stylize [VALUE]
```

Fill in the variables for each new image, and keep notes on which combinations work best. Over time, you'll discover your signature aesthetic — the specific combinations of style, lighting, and composition that produce consistently great results.

## Organizing Your Midjourney Workflow

Creating great AI art is iterative. A typical workflow looks like:

1. **Draft a prompt** using a template from your library.
2. **Generate 4 variations** and study the results.
3. **Identify what works** — lighting, composition, colors.
4. **Refine the prompt** based on what you see, adding or removing keywords.
5. **Vary strong candidates** to explore nearby options.
6. **Upscale the winner** for final use.

Keep notes on which prompt patterns produce the best results. This turns ad-hoc experiments into a repeatable system.

## Conclusion

Great Midjourney prompts come down to understanding how words translate to visual qualities and building a repeatable system for crafting them. Start with a strong subject description, layer in style, composition, lighting, and mood, and use parameters to control the technical details. The gap between a beginner and a pro is mostly about specificity, iteration, and building a prompt library you can reuse.

If you want to organize your Midjourney prompt templates with variables, keep track of which variations work best, and build a reusable system for your AI art workflow, [try PromptWright free](https://promptwright.net/signup). It's built to turn prompt experimentation into a structured, repeatable process.

Enjoyed This Article?

Get more prompt engineering tips delivered weekly. Free, no spam.

Join 500+ prompt engineers. Unsubscribe anytime.

Ready to build better prompts?

Try PromptWright free — structured prompt editor with multi-model testing.

Get Started Free →