Midjourney's main command for creating images from a text prompt.
What the AI does
Midjourney reads the prompt that follows and creates a set of image options.
When to use it
Every time you want Midjourney to make a new image.
Where it works
Midjourney in Discord: type /imagine, then your prompt. On the Midjourney website you type the prompt in the Imagine bar instead. Parameters such as --ar go at the end.
Common mistakes
Putting parameters like --ar in the middle of the prompt.
Expecting /imagine to work in ChatGPT, Gemini or AI Compass; it is a Midjourney command.
Example
/imagine a lighthouse on a cliff at golden hour, soft clouds, cinematic photo --ar 16:9
/describe
Also written: describe, image to prompt
A Midjourney command that looks at an image you upload and suggests text prompts that could describe it.
What the AI does
Midjourney writes several prompt ideas based on the image's subject and style.
When to use it
When you like an image's look but don't know which words describe it.
Where it works
Midjourney: run /describe and upload an image (on the website, use the describe option on an image). The suggested prompts are a starting point, not an exact recipe.
Common mistakes
Expecting the suggested prompt to recreate the same image exactly.
Uploading images you don't have the right to use.
Example
/describe (then upload your image)
/blend
Also written: blend, mix images
A Midjourney command that mixes two to five uploaded images into a new one.
What the AI does
Midjourney combines the subjects, colors and style of the images.
When to use it
When you want a new image that merges the feel of several pictures.
Where it works
Midjourney in Discord: run /blend and upload 2-5 images. It works from images only, without a text prompt.
Common mistakes
Blending images with very different shapes and expecting a clean layout.
Adding long text: /blend works from images.
Example
/blend (then upload two images, for example a city photo and a watercolor painting)
/settings
Also written: settings, default settings
A Midjourney command that shows and changes your default settings.
What the AI does
Opens buttons for defaults such as model version, stylize level and speed mode, so you don't have to type them every time.
When to use it
When you keep adding the same parameters to every prompt.
Where it works
Midjourney in Discord: type /settings. On the website, defaults are in the settings panel next to the Imagine bar.
Common mistakes
Forgetting a default you set, then wondering why results look different.
Typing a parameter in the prompt that conflicts with your defaults.
Example
/settings
/info
Also written: info, account info
A Midjourney command that shows your account and job information.
What the AI does
Shows details such as your plan, remaining fast time and jobs that are running or queued.
When to use it
When you want to check your usage or whether a job is still running.
Where it works
Midjourney in Discord: type /info.
Common mistakes
Expecting it to show information about an image; it shows your account.
Example
/info
/shorten
Also written: shorten, prompt analysis
A Midjourney command that analyzes a prompt and suggests shorter versions.
What the AI does
Shows which words carry the most weight and which ones barely change the result.
When to use it
When a long prompt gives messy results and you want to know which words matter.
Where it works
Midjourney in Discord: type /shorten and paste your prompt.
Common mistakes
Deleting words just because they look weak, without testing the result.
Example
/shorten a beautiful very detailed amazing photo of a red fox sitting in a snowy forest at sunrise, high quality
--ar
Also written: --aspect, --ar 16:9, --ar 9:16
A Midjourney parameter that sets the image shape, like --ar 16:9 for wide or --ar 9:16 for tall.
What the AI does
Tells Midjourney which width-to-height ratio to render.
When to use it
Use it when writing prompts for Midjourney.
Where it works
Midjourney only, at the end of the prompt. Other generators and AI Compass's own generator treat it as ordinary text; use the size setting or describe the shape in words instead.
Common mistakes
Putting --ar in the middle of the prompt.
Using it in AI Compass's image generator and expecting a different shape.
Example
Write a Midjourney prompt for a cozy reading nook by a rainy window, warm lamp light, ending with --ar 4:5.
--no
Also written: --no text, no, exclude
A Midjourney parameter that lists things to leave out, like --no text, people.
What the AI does
Tells Midjourney to avoid the named things.
When to use it
Use it in Midjourney when something unwanted keeps appearing.
Where it works
Midjourney only, at the end of the prompt. AI Compass and plain-language generators ignore the flag; write 'without text' or 'no people' as a normal sentence instead.
Common mistakes
Using --no in AI Compass's generator.
Writing 'no' inside the main description in Midjourney, which can add the item instead.
Example
Write a Midjourney prompt for a calm empty beach at sunrise, ending with --no people, text.
--seed
Also written: seed, seed number
A Midjourney parameter that sets the starting noise number. The same seed with the same prompt and settings gives a very similar image.
What the AI does
Lets you repeat or fine-tune a result instead of getting a random new one.
When to use it
Use it to test small prompt changes or keep a consistent look.
Where it works
Midjourney only, at the end of the prompt: --seed followed by a whole number. Stable Diffusion apps have their own Seed field; ChatGPT, Gemini and AI Compass have no seed control.
Common mistakes
Expecting the same seed to give the same image in a different tool or model.
Changing many settings at once, so the seed can't help compare.
Example
Write a Midjourney prompt for a watercolor fox in a forest using --seed 4242, and explain how to change only the colors in a second try.
--stylize
Also written: --s, stylize, --s 250
A Midjourney parameter that controls how strongly its own artistic style is applied.
What the AI does
Low values follow your prompt closely; high values add more artistic, decorative choices.
When to use it
Lower it for exact results, raise it for more artistic images.
Where it works
Midjourney only (--stylize or --s with a number). Other generators and AI Compass ignore it; describe the look in words instead, like 'simple' or 'highly artistic'.
Common mistakes
Setting it very high when you need a precise product image.
Using it outside Midjourney.
Example
Write two Midjourney prompts for a minimalist mountain poster, one with a low --s value and one with a high --s value, and explain the difference.
--chaos
Also written: --c, chaos, variety
A Midjourney parameter that controls how different the starting images in one batch are from each other.
What the AI does
Higher values give more unusual, varied results; low values give similar ones.
When to use it
Raise it when exploring ideas; keep it low when you know what you want.
Where it works
Midjourney only (--chaos or --c with a number). Not used by other generators or AI Compass.
Common mistakes
Using high chaos for final, polished images.
Confusing chaos with stylize.
Example
Write a Midjourney prompt to explore logo ideas for a bakery, using --chaos 50 to get varied directions.
--quality
Also written: --q, quality parameter
A Midjourney parameter that changes how much time is spent rendering the image.
What the AI does
Higher values spend more time on detail; lower values are faster and use less of your plan.
When to use it
Lower it for quick drafts, keep the default for finished images.
Where it works
Midjourney only (--quality or --q). Other generators set quality in their own settings; AI Compass sets image quality itself.
Common mistakes
Thinking it raises the pixel size; it doesn't.
Using it in tools that don't read flags.
Example
Write a Midjourney prompt for quick draft sketches of a sci-fi city with a low --q value, then a final version with the default quality.
--v
Also written: --version, version, model version
A Midjourney parameter that picks which of its model versions to use, written --v followed by a number.
What the AI does
Switches the underlying model, which changes the look and how well the prompt is followed.
When to use it
Use it when you need the look of a specific version or to compare versions.
Where it works
Midjourney only. Version numbers change over time, so check Midjourney's current documentation. AI Compass's generator ignores it.
Common mistakes
Copying an old version number from an old prompt without checking it still exists.
Using --v and --niji together.
Example
Write a Midjourney prompt for a portrait of an astronaut in a sunflower field and explain how to choose a model version with --v.
--niji
Also written: niji, niji mode
A Midjourney parameter that switches to its anime and illustration focused model.
What the AI does
Produces images in anime and manga-like styles more reliably.
When to use it
Use it in Midjourney for anime, manga and stylized character art.
Where it works
Midjourney only. In AI Compass and other generators, write 'anime style' in words instead.
Common mistakes
Using --niji for photorealistic images.
Using it with --v at the same time.
Example
Write a Midjourney prompt using --niji for an anime girl reading under a cherry blossom tree, soft pastel colors.
--style raw
Also written: style raw, raw mode
A Midjourney parameter that reduces its default beautifying style.
What the AI does
Makes results follow your prompt more literally, often more photographic and less decorated.
When to use it
Use it for realistic photos or when Midjourney adds too much of its own flair.
Where it works
Midjourney only. AI Compass's Prompt Builder may add it to image prompts meant for pasting into Midjourney; AI Compass's own generator doesn't use it.
Common mistakes
Expecting it to make images more artistic.
Using it outside Midjourney.
Example
Write a Midjourney prompt for a natural candid photo of a street musician at dusk, using --style raw and --ar 3:2.
--iw
Also written: image weight, --iw 2
A Midjourney parameter that sets how much an image you add to the prompt influences the result, compared with the text.
What the AI does
Higher values copy more from the reference image; lower values follow the text more.
When to use it
Use it when you start a Midjourney prompt with a reference image link.
Where it works
Midjourney only, used together with an image link at the start of the prompt. Not used by AI Compass.
Common mistakes
Setting --iw without adding an image.
Using a very high value and then wondering why the text was ignored.
Example
Explain how to write a Midjourney prompt that uses a reference photo of my living room with --iw to redesign it in Scandinavian style.
--tile
Also written: tile, seamless pattern
A Midjourney parameter that makes images which repeat seamlessly, like wallpaper or fabric patterns.
What the AI does
Midjourney makes the edges match, so copies of the image fit together without visible seams.
When to use it
For backgrounds, textures, wrapping paper and fabric designs.
Where it works
Midjourney only, at the end of the prompt. Other generators and AI Compass ignore it; write 'seamless repeating pattern' in words instead (results may not tile perfectly).
Common mistakes
Using it for scenes with one main subject; it suits patterns.
Not testing the tile by placing copies side by side.
Example
Write a Midjourney prompt for a seamless pattern of small lemons and leaves on a cream background, ending with --tile.
--weird
Also written: weird, --w
A Midjourney parameter that adds unusual, offbeat qualities to an image.
What the AI does
Higher values push results toward strange and unexpected looks.
When to use it
When exploring surprising, experimental ideas.
Where it works
Midjourney only, at the end of the prompt with a number. Not used by other generators or AI Compass.
Common mistakes
Using high values for product or business images.
Confusing it with --chaos, which varies the batch instead.
Example
Write a Midjourney prompt for a surreal tea party in a forest, using a moderate --weird value, and explain what it changes.
Every guide and dictionary card is also in the AI Compass app, with saved items, search and “Try it” buttons that open chat, image or video with the example filled in. Reading is free; creating uses credits, and the price is shown before you start.