To prompt Midjourney, this sheet’s template describes what you want first, puts settings such as --ar and --v after it, and excludes things by adding --no followed by a comma-separated list. Midjourney’s documentation says --no is a weight, not a ban. Midjourney documents --no directly; this sheet quotes that documentation, and drops two claims this page used to make without a source.
- V8.2 is the default model. Midjourney: “V8.2 released as the default version on July 24, 2026.”
--nois how you exclude things. Add it “to the end of your prompt”, followed by what you do not want.- List several items with commas. Midjourney: “You can even list multiple elements by separating them with commas.”
- It is a weight, not a ban. Midjourney says
--no“is the same as weighing part of a multi-prompt to "-0.5"”. - Do not write "please don't add fruit" in the prompt. Midjourney lists that style as a bad example and uses
--no fruitinstead. - Every parameter, with its range and default, is in the table below, taken from Midjourney’s parameter list on 12 Oct 2026.
::weights are a V6-era tool. Midjourney’s chart marks multi-prompting unavailable on V7 and V8.
Every rule on this sheet is quoted from the vendor's own published documentation, linked where it is used.
The formula
Midjourney, No parameter, read in a browser 16 Sep 2026: “Add --no to the end of your prompt in the Imagine bar, followed by the thing or list of things you don't want in your image.”
Template:
[subject] [action] [setting] [lighting] [style] --ar [ratio] --no [thing, thing, thing]
Worked example:
A ceramicist centring clay, north window light, documentary photography, muted palette --ar 3:2 --no text, watermark
Breakdown — what each part does
Prompt body
Describe what you do want. This is the part that steers the image.
Parameters
Settings such as aspect ratio (--ar) and version (--v) go after the description. Midjourney: “Add --v # to the end of your prompt in the Imagine bar.”
--no
Midjourney: “The --no parameter is a tool you can use to tell Midjourney what you don't want in your image.” Separate several items with commas.
Every parameter Midjourney lists
The plain point: a parameter is a setting typed after your description, starting with two dashes. Midjourney’s rules for typing them: put them at the end, leave a space before the dashes, and use no commas or full stops inside them. 1
The table follows Midjourney’s own parameter list as it stood on 12 Oct 2026, grouped by what each one controls. Ranges and defaults are quoted from each parameter’s page; where a page only describes an older version, the row says so.
| Parameter | What it does | Range and default |
|---|---|---|
| THE IMAGE ITSELF | ||
--ar or --aspect | Does: sets the shape: width to height. | Range and default: default 1:1; whole numbers only (write 139:100, not 1.39:1); up to 14:1 on V6 to V8 (4:1 for HD images). 2 3 |
--chaos or --c | Does: how different the four images in a set are from each other (called Variety on the website). | Range and default: 0 to 100; default 0. 4 |
--stylize or --s | Does: how much of Midjourney’s own artistic flair is added. Low follows the prompt closely; high takes more liberties. | Range and default: 0 to 1000; default 100. 5 |
--weird or --w | Does: pushes towards quirky, unconventional images. Experimental. | Range and default: 0 to 3000; default 0. 6 |
--raw | Does: switches off Midjourney’s automatic styling, so simple prompts look more photographic and detailed prompts are followed more literally. | Range and default: on or off; V5.1 and later. 7 |
--quality or --q | Does: how much GPU time goes into the first set of four images. | Range and default: V7: default 1, or 2 or 4. The Version chart marks it not available on V8.1 and V8.2. 8 3 |
--seed | Does: fixes the starting noise, so you can change one word and compare. It does not save a style or a character. | Range and default: any whole number from 0 to 4294967295; random if you leave it out. 9 |
--no | Does: pushes something out of the image. | Range and default: works like a weight of -0.5, not a ban (see below). 10 |
--tile | Does: makes an image whose edges join up, for repeating patterns. | Range and default: on or off. Not compatible with the Edit Model. 11 |
--hd / --sd | Does: higher or standard resolution. | Range and default: V8.1: HD 2048px, SD 1024px. 1 |
| MODEL AND MODE | ||
--version or --v | Does: picks the model version. | Range and default: default V8.2. 3 |
--niji | Does: the separate model for anime and Eastern illustration styles. | Range and default: current version Niji 7, launched 9 Jan 2026. 1 3 |
--draft | Does: fast, low-cost drafts for trying ideas. | Range and default: V7: four images. V8.1 and V8.2: a batch of 24 small images, on midjourney.com only. Midjourney’s own pages disagree here: the Version chart marks Draft Mode not available on V8.1 and V8.2. 12 |
--repeat or --r | Does: runs the same prompt several times. | Range and default: 2 up to a plan limit (2 to 40 on the largest plans); Fast and Turbo modes only. 13 |
--fast, --relax, --turbo | Does: sets GPU speed for this job. | Range and default: Turbo is not supported on V8.1. 14 |
--stealth / --public | Does: makes the result private or public on the Midjourney website. | Range and default: on or off. 1 |
| YOUR OWN IMAGES AND TASTE | ||
--sref | Does: style reference: borrow the look of an image or a style code (explained below). | Range and default: V6 and later. 15 |
--sw | Does: style weight: how strongly the style reference applies. | Range and default: 0 to 1000; default 100. Not compatible with Moodboards. 15 |
--sv | Does: picks which version of the style reference system to use. | Range and default: V7: six versions, default --sv 6; V6: four, default --sv 4. The page does not describe V8. 15 |
--p or --profile | Does: Personalization: applies a taste profile built from images you picked. | Range and default: V6 and later; you must unlock your profile first. 16 |
--iw | Does: image weight: how much an image prompt counts against your words. | Range and default: V8.1 and V7: 0 to 3, default 1; Niji 7: 0 to 2, default 1. 17 |
--edit | Does: Edit Model: change an image with written instructions, or build a new one from up to four reference images. | Range and default: V8.1 and V8.2. 1 18 |
--oref and --ow | Does: Omni Reference: put a person or object from one image into a new one; --ow sets how closely. | Range and default: V7 only; replaced by the Edit Model on V8. --ow 1 to 1000, default 100. 1 19 |
| VIDEO | ||
--video, --motion, --loop, --end, --bs | Does: video jobs: --video starts one in Discord; --motion low or high sets how much moves; --loop ends on the first frame; --end sets a different end frame; --bs sets how many videos. | Range and default: motion: low is the default; batch size 1, 2 or 4. Video jobs take only these parameters plus --raw. 20 |
Not on the list: the Version chart also shows an --exp parameter for V7 and V8, but the parameter list gives it no entry, so this sheet does not describe it. --cref (Character Reference) now sits under “Legacy Parameters”; on V8 the Edit Model does that job. 21
Most prompts need two or three of these: --ar, maybe --s or --raw, and --no. Check the version column before you copy a setting from an old prompt.
Style references and style codes, in plain words
A style reference copies the look of a picture, not what is in it. You give Midjourney an image, and it borrows the colours, medium, textures and lighting for your new image, without copying the people or objects. 15
A style code is a number that stands for a look. Instead of an image, you write --sref followed by a code from Midjourney’s internal style library. --sref random picks one for you and shows you the code afterwards, so you can reuse it. You cannot turn your own uploaded image into a code. 15
Keep the words simple when you use one. Midjourney’s advice is to describe what you want to see, not how to change the reference, and to avoid style words that fight the reference. Its bad example is “copy this style and make a bunny”; its good one is “ballpoint pen sketch of a bunny”. Midjourney’s Style Explorer, on the Explore page, lets you browse codes. This sheet does not list codes; the Style Reference page is the place to start. 15
A personalization profile (--p) is different again: it learns your taste from images you pick, rather than copying one image. Plain-English background: what a style reference is.
What changed from V6 to V7 and V8
Midjourney’s Version page carries a comparison chart for V6, V7 and V8.1 & V8.2. The changes that matter for prompts:
- Multi-prompts and
::weights are V6 and earlier. The chart marks multi-prompting available on V6 and not on V7 or V8, and the multi-prompt page lists versions only up to 6.1. 10 Midjourney still explains--noas a -0.5 multi-prompt weight, but the chart marks--noavailable on all three, so on current versions use--norather than writing::-0.5. - Omni Reference arrived in V7 and left in V8. On V8.1 and V8.2 the Edit Model replaces it, along with Character Reference and the Retexture tool. 3
- Quality values changed. V6 took 0.5, 1 or 2; V7 takes 1, 2 or 4; the chart lists no quality setting for V8.
- New on V8: HD images (2048px) and the Edit Model. Gone on V8, per the chart: Turbo mode and Draft mode (though the Draft page describes a V8 draft mode on the website, as noted in the table).
- Personalization carries over. Your V7 global profile works on V8.1 and V8.2; there is no global V8 profile yet. 16
An old V6 prompt with :: weights or --cref will not behave the same on V8. Rewrite weights as plain description and use the Edit Model for characters.
Does it take a negative prompt?
--no parameter.YESMidjourney, No parameter, read in a browser 16 Sep 2026, re-read directly at source 13 Oct 2026: “Using the --no parameter is the same as weighing part of a multi-prompt to "-0.5"”.
Because it works as a negative weight, treat --no as a strong push away rather than a guarantee. If something must not appear, also describe a scene that leaves no room for it.
--no is a weight, not a ban. When something must not appear, describe a scene with no room for it as well.--no lowers the odds of something appearing. It does not promise it will not appear.
What not to do
--no as a hard ban. Midjourney equates it to a -0.5 weight.✕An earlier version of this sheet was labelled secondary sourcing and said Midjourney's documentation could not be retrieved. It can: every quote above comes from docs.midjourney.com. Two claims that had no source were removed — that a second --no does not stack, and that each word after --no is read as a separate exclusion.
Free or paid? What the free plan allows, who owns the output, and when you need to pay: free vs paid AI image, video and music tools.
Where this model sits among the others: which models take a negative prompt, and what each vendor publishes as its formula.
Why a photoreal prompt still looks like an advert is on why AI output looks fake.
The checklist
Before you send it:
--no, separated by commas?□Midjourney, No parameter · Midjourney, Version All read at source on 16 Sep 2026. Added 12 Oct 2026, read in a browser: Parameter List and the page for each parameter linked in the table, Multi-Prompts & Weights, Style Reference, Legacy Features. The Version page was re-read the same day: V8.2 is still the default.
Model versions and vendor documentation both move. Re-read the source before relying on a specific number.
Every other sheet in this set: all model cheat sheets.
- 1Midjourney, Parameter List, read at source 12 Oct 2026 in a browser“Parameters always go at the end of your text prompt, after you've described what you want to see.”; “Don't use commas, periods, or other punctuation marks in your parameters.” ↑a“Generate V8.1 images at higher resolution (2048px) using --hd or in standard resolution (1024px) using --sd”. ↑b“Use our model focused on anime and Eastern aesthetics”. ↑c“Make your creations private on the Midjourney website”. ↑d“Create and modify images using written instructions and up to four reference images using --edit”. ↑eOmni Reference “(replaced by the Edit Model in V8.X)”. ↑f
- 2Midjourney, Aspect Ratio, read at source 12 Oct 2026 in a browser“The default aspect ratio is 1:1.”; “--ar cannot contain decimals.” ↑
- 3Midjourney, Version, read at source 12 Oct 2026 in a browsercomparison chart, “Max. Aspect Ratio”, V8.1 & V8.2: “14:1 (4:1 for HD)”. ↑acomparison chart, “Quality Parameter”: V6 “(0.5, 1, 2)”, V7 “(1, 2, 4)”, V8.1 & V8.2 marked not available. ↑b“The current default Midjourney version is V8.2.” ↑c“Niji 7, launched on January 9, 2026”. ↑dV8.2 “also features the new Edit Model, which replaces Omni Reference, Character Reference, and the Retexture tool.” ↑e
- 4Midjourney, Chaos / Variety, read at source 12 Oct 2026 in a browser“By default chaos is set to 0”; “You can set the chaos value anywhere between 0 and 100”. ↑
- 5Midjourney, Stylize, read at source 12 Oct 2026 in a browser“The default value for stylize is 100, and you can adjust it anywhere between 0 and 1000 with the latest Midjourney versions.” ↑
- 6Midjourney, Weird, read at source 12 Oct 2026 in a browser“By default, weird is set to 0, but you can add values up to 3000.” ↑
- 7Midjourney, Raw, read at source 12 Oct 2026 in a browser“Raw is compatible with Midjourney versions 5.1 and later.” ↑
- 8Midjourney, Quality, read at source 12 Oct 2026 in a browser“In Midjourney version 7 the default setting for quality is 1. You can also set values to 2 or 4”. ↑
- 9Midjourney, Seeds, read at source 12 Oct 2026 in a browser“You can lock a seed by plugging in a whole number between 0 and 4294967295 in your prompt.”; “Seeds are 99% identical in V8.X.” ↑
- 10Midjourney, Multi-Prompts & Weights, read at source 12 Oct 2026 in a browser“Using the no parameter is the same as using a -0.5 weight.” ↑a“Multi-prompts work with Midjourney versions 1, 2, 3, 4, Niji 4, 5, Niji 5, 6, Niji 6, and 6.1.” ↑b
- 11Midjourney, Tile, read at source 12 Oct 2026 in a browser“The --tile parameter makes sure that the edges of your image match up perfectly with the same image when you place them next to each other.” ↑
- 12Midjourney, Draft & Conversational Modes, read at source 12 Oct 2026 in a browser“A single Draft mode prompt in V8.1 or V8.2 will generate a big batch of 24 images.”; “Draft mode in V8.1 and V8.2 is only available on midjourney.com”. ↑
- 13Midjourney, Repeat, read at source 12 Oct 2026 in a browser“Pro and Mega plans offer 2 to 40”; “the repeat parameter only works in Fast and Turbo Modes. It won't work in Relax Mode.” ↑
- 14Midjourney, GPU Speed (Fast, Relax, Turbo), read at source 12 Oct 2026 in a browser“Turbo mode is not currently supported in V8.1.” ↑
- 15Midjourney, Style Reference, read at source 12 Oct 2026 in a browser“Style References are compatible with Midjourney versions 6 and later.” ↑a“You can set this parameter to any value between 0 and 1000, with the default being --sw 100.” ↑b“There are six versions of the Style Reference feature in Midjourney version 7 when using images. Use the --sv parameter to choose between them. --sv 6 is default.” ↑c“It doesn't copy objects or people, just the overall style—like colors, medium, textures, or lighting”. ↑d“Each numerical code produces a unique visual style, and can be used by adding --sref <code> to the end of your prompt.”; “You cannot create a style code based on an uploaded image.” ↑e“Focus on content, not instructions - Use your text prompt to describe what you want to see, not how Midjourney should modify the reference image.” ↑f
- 16Midjourney, Personalization, read at source 12 Oct 2026 in a browser“Personalization is compatible with Midjourney versions 6 and later.” ↑a“Your Global V7 Profile works in V8.1 and V8.2.” ↑b
- 17Midjourney, Image Prompts, read at source 12 Oct 2026 in a browserchart for Version 8.1, Version 7 and Niji 7: “Image Weight Default 1 1 1”; “Image Weight Range 0 - 3 0 - 3 0 - 2”. ↑
- 18Midjourney, Edit Model, read at source 12 Oct 2026 in a browser“The Edit model is compatible with Midjourney versions 8.1 and 8.2.” ↑
- 19Midjourney, Omni Reference, read at source 12 Oct 2026 in a browser“Omni Reference can only be used with Midjourney version 7.”; “You can set this parameter to any value between 1 and 1,000, with the default being --ow 100.” ↑
- 20Midjourney, Video, read at source 12 Oct 2026 in a browser“Video generations are only compatible with video-specific parameters: --motion low, --motion high, --raw, --loop, --end, and --bs #.”; “You can use 1, 2, or 4.” ↑
- 21Midjourney, Legacy Features, read at source 12 Oct 2026 in a browserlists “Character Reference --cref --cw” under “Legacy Parameters”. ↑
Free printable AI cheat sheets, and AI news every week.
Sixteen one-page AI guides and checklists in one free PDF, no sign-up. And if you want to keep up, the newsletter covers what changed in AI each week, in plain English, with the sources.
SUBSCRIBE FREE ↗ GET THE CHEAT SHEET PACK