Midjourney Prompt Guide: Settings, Parameters and Examples

Creating advanced prompts in Midjourney can be simpler than you might think.

By adding simple tweaks to your Midjourney prompts you can see significant boosts to the quality of your generated images.

In this ultimate Midjourney prompt guide, I’ve taken a deep dive into creating amazing images in the Midjourney web app at midjourney.com.

We’ll cover everything from aspect ratios to image chaos settings and everything in between. But first…

---

What is Midjourney?

Have you ever tried to sit down and create an amazing piece of artwork only to be less than satisfied with your results?

Well, Midjourney allows you to create the world-class visuals you’ve always dreamed of!

Using a powerful AI model that has been trained on millions of images, Midjourney takes your text description, known as a prompt, and brings it to life before your eyes.

The San Francisco company founded by David Holz rose to fame after its release to the public in July 2022, and its Discord server passed 20 million members in 2024.

You can still join the Midjourney Discord to hang out with other creators or make images with the Midjourney Bot.

Midjourney is commonly regarded as one of the best AI image generators around, and it keeps shipping new models. V8.2 has been the default since July 24, 2026.

Powered by a proprietary machine learning model designed for generating high-quality AI images from text descriptions, Midjourney has made AI art accessible to the world.

Getting Started with Midjourney

To get started using Midjourney, head to midjourney.com and click Sign Up.

The site looks different from this screenshot now, but the steps are the same.

You can sign in with Google or Discord. If you’ve made Midjourney images on Discord before, log in with that Discord account so your old images come across with you.

There’s no free trial on the website, so you’ll need to pick a plan before you can create. Basic is $10 a month, Standard is $30, Pro is $60 and Mega is $120, with 20% off if you pay yearly (Midjourney’s plan comparison).

Once you’re in, download my Midjourney cheat sheet for later, and head to the Create page.

Generating Your First Image

To generate your first image in Midjourney, it’s as simple as typing what you want to generate into the Imagine bar at the top of the Create page.

It can be anything you want! I tried out the super simple prompt A cartoon hamster.

Hit Enter (or the send button on your phone) and Midjourney makes a set of four images.

Amazing right?!

Adjusting Basic Settings

Click the settings button in the Imagine bar. This opens the basic image settings.

The panel has had a facelift since I took this screenshot. These days you’ll see four sections: Aspect Ratio, Model, Aesthetics and More Options.

These settings are sticky – meaning the settings you adjust here will stay on for future generations until you change them.

You can overwrite the settings using specific syntax in your text prompt, but we’ll get to that later!

Aspect Ratio

This is where you set the shape of your image. Simply pick the aspect ratio you want and it’ll apply to everything you make.

If you need specific aspect ratios, or ratios the panel doesn’t offer, append the following syntax to your text prompt:

--ar [your aspect ratio]

For example, if I wanted to generate a banner with a 6:1 ratio, simply add the following to the end of your prompt:

--ar 6:1

V6, V7 and V8 handle ratios all the way out to 14:1 (4:1 for HD images). Stick to whole numbers: 139:100 works, 1.39:1 doesn’t.

Aesthetics

The aesthetics panel is where things get interesting. There are 3 sliders you can adjust in the Aesthetics panel:

Stylization

Influences how strongly the Midjourney aesthetic is applied. Low stylization values produce images that closely match the prompt but are less artistic. High stylization values create images that are very artistic but less connected to the prompt.

Stylization values can be from 0 to 1000, the default setting is 100.

To control stylization via your text prompt, use the following syntax:

--s [a value between 0 and 1000]

Weirdness

Introduces quirky qualities to your generated images, resulting in unique and unexpected outcomes.

Weirdness values can be from 0 to 3000, the default setting is 0.

To control weirdness via your text prompt, use the following syntax:

--w [a value between 0 and 3000]

Variety

Influences how varied the images are. High values will produce more unusual and unexpected results and compositions. Lower values have more reliable, repeatable results.

Variety (also known as Chaos) values can be from 0 to 100, the default setting is 0.

To control Chaos via your text prompt, use the following syntax:

--c [a value between 0 and 100]

Here’s an example of what the Aesthetics settings can do. The image on the left uses the default settings and on the right is an image using the same prompt, but with Stylize, Weirdness, and Variety all set to the maximum values!

a puppy happy with excitement, cartoon realism, Disney animation

Model

The model panel allows you to select which Midjourney model to use for your image generations, plus a couple of extras.

• Version lets you pick your default model from the dropdown. V8.2 is the default, and you can still choose older versions. Version 4 was about where things started to get really good with Midjourney, but it can be fun to look at the earlier ones.

• Raw switches on raw mode, which turns off Midjourney’s default styling. Using it can help you create more photo-realistic images, cinematic scenes, or cuter characters. I’ve found that raw mode can also improve prompt adherence too. The prompt version is --raw.

• SD or HD sets the resolution. V8.1 and V8.2 can render HD images at 2048px without upscaling. An HD prompt costs about 1.3 minutes of GPU time versus 0.8 for SD, and you can add --hd or --sd to a prompt instead.

More Options

• Speed dictates how quickly Midjourney generates your images. Fast uses your monthly GPU hours, Relax is unlimited but slower (Standard plan and up), and Turbo is quicker again at double the GPU cost. Turbo isn’t currently supported on V8.1 or V8.2.

• Stealth is only available to Pro and Mega subscribers. Enabling it means your generations won’t be publicly available on the Midjourney Explore page.

• Video resolution and batch size live here too.

Personalize has moved out of the settings panel. It now has its own P button next to the Imagine bar (more on that below).

Crafting Great Text Prompts in Midjourney

So far we’ve covered the basic Midjourney settings you can use to get started creating stunning images.

Now it’s time to take a closer look at how to craft the perfect text prompts for Midjourney.

By following a few simple rules when creating text prompts, you’ll give yourself more control over the generated content and at the end of the day become a stronger user of Midjourney.

Let’s learn how to craft the perfect text prompts. 👇

1. Start Simple

Midjourney has become so good now, that in many cases you can get what you need with a basic prompt.

No fancy tricks, just start with a clear and simplistic description of your vision. Imagine trying to describe your image to another person. Be sure to include any specific details, colors, art styles, or other features.

Here’s 3 great examples of what a simple, yet well crafted prompt can do in Midjourney:

A realistic photograph of a pretty woman listening to music over a blue background --ar 16:9

Magazine minimalist product photography of high end men's aftershave

Food photography of a freshly picked green apple

2. Specify Lighting

Specifying the lighting in your prompt is an excellent way to control the mood of your image.

Terms like natural lighting, soft lighting, afternoon light, colored lighting, and cinematic lighting can make an impactful difference to your image prompts.

Use lighting commands in your prompt to control the mood and tones of your AI generated artwork.

3. Leverage Keywords

I have a few keywords committed to memory that I’ll often pull out to improve the aesthetics of an image I’m working on. Here’s a list of some high-impact keywords you can leverage to realize your creative vision:

• High contrast

• Cinematic

• Minimalist

• Line art

• Vintage

• Neon theme

• Grunge

• Retro

• Industrial

• Futuristic

• Pastel

• Organic

• Dystopian

• Gothic

The list goes on! Depending on the image generation you’re trying to create, you should identify some of these keywords and keep them handy somewhere!

4. Find Your Style

Everyone has their own unique style when it comes to generating AI art.

Some creators are drawn to high contrast, bright, and colorful generations while others like low light medieval styles.

Probably the best way to find your style in Midjourney is to honestly just use Midjourney.

The more generation hours you get under your belt, the more tricks, tweaks, and settings you’ll learn to use to get your AI generated art exactly the way you want it.

5. Use a Prompt Marketplace

Prompt marketplaces like PromptBase allow you to scroll through hundreds of unique prompts created by expert prompt engineers in Midjourney and other AI apps. You can sell your own there too, which I cover in my guide to making money with Midjourney.

No matter how good you get, it’s always worth checking into prompt marketplaces from time to time and seeing what’s possible!

To this day I still buy prompts just to see how other high-level creators are using Midjourney!

6. Put Text in Double Quotes

Want words in your image? Wrap them in double quotation marks, like a neon sign that says "open late". Midjourney V6 and newer will try to spell out whatever’s inside the quotes. Short words and phrases work best, and single quotes won’t do the trick.

Advanced Prompt Techniques for Midjourney

Negative Prompts

Negative prompts in Midjourney allow you to remove unwanted items, colors, or themes from your different art styles.

Using negative prompts in Midjourney give an extra level of control to the advanced prompt engineer.

The easiest way to use negative prompts in Midjourney is as follows:

--no [items you want removed]

Here’s a great example of how negative prompting helped me create this pencil sketch prompt.

Notice each image included pencils and hands? We can fix that using negative prompts.

Imperfect silhouette outline pencil hand sketch of a pretty girl on white paper background with graphite stroke shading --no pencil, pencils, hand, hands

Permutation Prompts

Permutation prompts are legit one of my favorite things in Midjourney.

The amount of time permutation prompts have saved me would be in the tens, if not hundreds of hours.

Permutation prompts allow you to run a series of unique prompts in the single original prompt. Here’s what I mean:

A fresh {green, red, blue} apple

This single prompt is equivalent to three individual prompts: A fresh green apple, A fresh red apple, and A fresh blue apple.

There’s heaps of ways to use permutation prompts but my personal favorite is the following:

[Your text prompt] --s {100, 200, 600, 1000} --c {1, 50, 100}

This permutation takes your prompt and applies 12 unique aesthetic combinations to generate a total of 48 images to choose from!

Just know that 12 prompts is over the Standard plan’s limit of 10 per permutation (Basic caps out at 4), so this one needs Pro or Mega. Permutations also only run in Fast or Turbo mode, not Relax.

You can see how much time permutation prompts can save when used appropriately.

Reference Images

Another great way to create captivating visuals with Midjourney is by using a reference image.

Reference images allow you to influence the model with existing images you might have from magazines, social media, or even other AI art creations.

To use one, click the image icon in the Imagine bar, upload your image, then drag it into one of these slots:

• Image Prompt guides the content, composition and colors. Control how much it counts with --iw (0 to 3, default 1).

• Style Reference copies the look (colors, medium, lighting) without copying the objects. Dial it up or down with --sw (0 to 1000, default 100), or try --sref random for a surprise style code.

• Attach to prompt uses the Edit model on V8.1 and V8.2. It takes up to four images to keep a character or product consistent, mash references together, or edit an image from plain instructions. It replaced Omni Reference (--oref, V7 only) and Character Reference (--cref, V6 only).

• Animate uses the image as the first frame of a video.

Then type your prompt as you normally would. For Image Prompts and Style References, describe what you want to see in the final image rather than giving instructions like “copy this style”.

Describe

What if you have an existing image that you love but you’re not sure how you could create a prompt to describe it?

Describe is kind of like Midjourney in reverse – upload an image and Midjourney will give you a text description! It’s an awesome tool for leveling up your prompt skills.

To use Describe on the website, drag an image onto the Imagine bar and drop it on the Drop image to describe panel, or right-click any image on your Create or Organize page and choose Describe Image.

Midjourney then writes four prompts based on your image. Click Use Prompt to drop one into the Imagine bar, or Run all prompts to try all four. I go through it step by step in my guide to using Describe in the Midjourney web app.

The Editor (Inpainting and Outpainting)

The Editor is Midjourney’s generative fill/remove tool, similar to how you might make edits in Photoshop. It used to be called Repaint on the website (and Vary Region before that), and Discord still calls it Vary Region.

Open any image, click Edit under Creation Actions, paint over the area you want to change with the Erase brush (or use Smart Select), type what you want there instead and hit Submit Edit. You can also resize the canvas to add more around your image, and upload your own photos in the Edit tab.

On V8.1 and V8.2 the Editor runs on the new Edit model, which also understands plain instructions like “turn this into a pencil sketch”.

It’s a great alternative to using negative prompts, or for adding specific traits or features to your images.

Getting good at the Editor can save you hundreds of hours and or dollars getting good at Photoshop or hiring someone to do the edits for you.

Prompt Weights

Prompt weights allow you to communicate the importance of specific words to the AI, giving you a significant level of control.

You can also assign weights to words in your prompt to increase the presence of a specific feature or remove it entirely.

Heads up: multi-prompts and weights only work on V6.1 and older models (and Niji 6). V7 and V8 don’t support them, so add --v 6.1 to the end of your prompt if you want to play with weights.

Writing prompts with prompt weights can be done by adding a double colon after a chunk of your prompt, followed by the weight. For example:

[A leopard]::3 in the [dense Amazon]::7

This gives the term A leopard a 30% weight, and the term dense Amazon a 70% weight. You don’t actually need the square brackets: Midjourney splits the prompt at each ::, so everything before a double colon counts as one chunk. Just don’t put a space before the ::.

Alternatively if we ran:

[A leopard]::7 in the [dense Amazon]::3

It does the opposite. Here are those two generations so you can see the difference:

Personalize

Midjourney’s Personalization feature learns your personal artistic preferences to better match your tastes!

To unlock it, head to the Personalize page on midjourney.com and click the images you like from the grid. Keep picking until the progress bar fills and your Global Profile unlocks. (It used to take 200 pair rankings, and it’s quicker now.)

Once it’s unlocked, click the P button next to the Imagine bar to switch it on, or add --p to the end of your prompt. Liking images on the Explore page also feeds your profile, and your Global V7 Profile works with V8.1 and V8.2.

If anyone wants to try out my personalize code from 2024, add this to the end of your prompt. It was made on V6, so run it with V6.1:

--p prhx95i --v 6.1

Draft Mode

Draft mode is the cheap and cheerful way to explore ideas. Click the Draft mode button next to the Imagine bar (or add --draft to your prompt), and on V8.1 and V8.2 each prompt spits out a batch of 24 low-res images for about 0.4 minutes of GPU time. Click Vary on the ones you like to render them at full quality. Draft mode on V8 is website only, not Discord.

Video

Midjourney animates images too. Open any image and click one of the Animate buttons, or drop your own image into the Animate slot, and you get four 5-second clips. You can extend a clip up to 21 seconds, pick low or high motion (--motion low or --motion high), make it loop with --loop, or add --raw for more control. A batch of four SD videos costs about 8 minutes of GPU time.

9 Midjourney Prompt Examples

Here’s a library of 9 Midjourney prompt examples (I made these back on V6) you can use to get your creative juices flowing and level up your skills as a prompt engineer!

Food shot, ultra wide angle, 3 delicate lemons against a bright blue sky, morning dew, fresh fruit, realistic, ultra fine detail, depth of field, high resolution

An American Pitbull terroir is sitting in the center of the frame in front of a fireplace, looking around curiously. The background is bright and cozy.

Embossed minimalist R on white paper

cross-section CAD drawing of a brutalist train station

High contrast thick paint color explosion

White coffee mug front lens, black lid, placed on the water, with a little smoke around it

A beautiful vampire woman with pale skin, long white hair and red eyes wearing black robes sitting on an altar in the style of Magali Villeneuve street photography

Chibi dark soul character design of with white and teal color theme, dark background, glowing eyes and black helmet, simple style in the style of ch32

Worldly landscape post The singularity