--- # System prepended metadata title: midjourney tips for images --- # Midjourney Prompting – Recent Documentation (October 2025) [midjourney-tips-for-videos](/nAck-g-jT3y-PzkH2BTbjQ) ## Understanding Prompts - **What is a prompt?** A prompt is your creative starting point—it's the text or phrase that tells Midjourney what kind of image you want ([Prompt Basics – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32023408776205-Prompt-Basics#:~:text=What%20is%20a%20Prompt%3F)). The more thought you put into your prompt, the more unique and interesting the resulting image can be. Prompts can be a single word or a complete phrase and should guide Midjourney’s creative process ([Prompt Basics – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32023408776205-Prompt-Basics#:~:text=What%20is%20a%20Prompt%3F)). ## Prompting Tips & Tricks - **Keep prompts short and simple.** Clear phrases generally produce better images than long lists or overly detailed instructions ([Prompt Basics – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32023408776205-Prompt-Basics#:~:text=Prompting%20Tips%20%26%20Tricks)). - **Choose the right words.** Use precise synonyms (e.g., *gigantic* instead of *big*) to help Midjourney understand your intent ([Prompt Basics – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32023408776205-Prompt-Basics#:~:text=)). - **Be specific with numbers.** Instead of vague plurals like “cats,” specify “three cats” or use collective nouns such as “a flock of birds” ([Prompt Basics – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32023408776205-Prompt-Basics#:~:text=)). - **Focus on what you want.** Describe desired elements rather than what you don’t want. To explicitly exclude items, use the `--no` parameter ([Prompt Basics – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32023408776205-Prompt-Basics#:~:text=,Want)). - **Prompt length & details.** Short prompts rely on Midjourney's default style; adding details increases control. Consider including elements like **subject**, **medium**, **environment**, **lighting**, **color**, **mood**, and **composition** to refine the output ([Prompt Basics – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32023408776205-Prompt-Basics#:~:text=)). ## Advanced Prompting - **Combine techniques.** After mastering basic prompts, experiment with multiple text phrases, image prompts, style references, and parameters (e.g., aspect ratio) to customize the result ([Prompt Basics – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32023408776205-Prompt-Basics#:~:text=Advanced%20Prompts)). Parameters are appended at the end of the prompt and can control aspects like aspect ratio or quality. - **Explore the community.** The Midjourney community offers a **Prompt Craft** channel on Discord and a **Prompt Craft Room** on the web, where users share ideas and troubleshoot together ([Prompt Basics – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32023408776205-Prompt-Basics#:~:text=More%20Prompt%20Help)). ## Recent Changes in Midjourney (Version 7 – October 2025) - **Version 7 overview.** Released on **April 3 2025** and made the default model on **June 17 2025**, Version 7 interprets text and image prompts more precisely and produces images with richer textures, coherent details (like human bodies and hands) and improved aesthetics compared with earlier versions ([Midjourney Update October 2025 : What’s New ](https://multidocgenerator.com/blog/q/midjourney-october-2025-release-key-features-tips-changes-you-must-know#:~:text=Current%20Model%3A%20Version%207%20Overview)). - **Key features:** - **Draft Mode** speeds up image generation for rapid prototyping ([Midjourney Update October 2025 : What’s New ](https://multidocgenerator.com/blog/q/midjourney-october-2025-release-key-features-tips-changes-you-must-know#:~:text=Key%20features%20of%20Version%207,include)). - **Omni Reference** allows you to reference multiple aspects (structure, style, content) of an image for greater control in new outputs ([Midjourney Update October 2025 : What’s New ](https://multidocgenerator.com/blog/q/midjourney-october-2025-release-key-features-tips-changes-you-must-know#:~:text=,greater%20control%20over%20the%20output)). - **Flexible Parameters** support any aspect ratio, offer strong and subtle variations, enable remixing existing outputs and provide personalization via ranking or moodboards ([Midjourney Update October 2025 : What’s New ](https://multidocgenerator.com/blog/q/midjourney-october-2025-release-key-features-tips-changes-you-must-know#:~:text=,through%20ranking%20or%20moodboard%20creation)). - **Differences from Version 6.** Version 7 discontinues character reference and multi‑prompting but introduces improved upscalers, new quality settings (values 1, 2 or 4) and experimental parameters like `--exp` ([Midjourney Update October 2025 : What’s New ](https://multidocgenerator.com/blog/q/midjourney-october-2025-release-key-features-tips-changes-you-must-know#:~:text=Compared%20to%20Version%206%2C%20released,style%20art%2C%20Midjourney%20offers%20Niji)). - **Upcoming releases.** - **Version 7.1** (in testing) aims to improve prompt interpretation, refine Omni Reference and enhance Draft Mode; its release is expected soon ([Midjourney Update October 2025 : What’s New ](https://multidocgenerator.com/blog/q/midjourney-october-2025-release-key-features-tips-changes-you-must-know#:~:text=Upcoming%20Model%20Releases%3A%20V7,V8)). - **Version 8** (targeted for late 2025) will feature a new architecture and larger dataset; plans include an advanced image editor and next‑generation video model ([Midjourney Update October 2025 : What’s New ](https://multidocgenerator.com/blog/q/midjourney-october-2025-release-key-features-tips-changes-you-must-know#:~:text=Upcoming%20Model%20Releases%3A%20V7,V8)). - **Style system enhancements.** On **October 6 2025**, Midjourney expanded the Style Explorer, increasing the number of styles tenfold. Users can search for styles with keywords, rank styles during style ranking events (earning free fast hours) and influence personalization algorithms ([Midjourney Update October 2025 : What’s New ](https://multidocgenerator.com/blog/q/midjourney-october-2025-release-key-features-tips-changes-you-must-know#:~:text=10x%20More%20Styles%20)). - **New & upcoming tools.** Midjourney is developing a stand‑alone image editor, a depth reference tool for re‑texturing images, multiple personalization profiles, storytelling tools and **Midjourney TV** (grid views for videos) ([Midjourney Update October 2025 : What’s New ](https://multidocgenerator.com/blog/q/midjourney-october-2025-release-key-features-tips-changes-you-must-know#:~:text=New%20and%20Upcoming%20Features)). ## Using Parameters - The Midjourney documentation provides a **Parameter List** and additional articles about controls like **Chaos / Variety**, **Multi‑Prompts & Weights**, **No**, and **Pan**. These tools help fine‑tune prompts and should be appended after the main text to customize output ([Documentation – Midjourney](https://docs.midjourney.com/hc/en-us/categories/32013335627533-Documentation#:~:text=Midjourney%20Controls)). ## Tips for Effective Prompt Design - Use visual language and art style cues (e.g., “oil painting,” “cinematic lighting”) to guide the model. - Adjust prompt weights to emphasise key elements. - Experiment with synonyms and descriptive adjectives; iterate to refine results. - Explore the Midjourney community (Discord or website) for additional inspiration and support ([Prompt Basics – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32023408776205-Prompt-Basics#:~:text=More%20Prompt%20Help)). ## Midjourney Parameter Reference (Images) The parameters below are special instructions appended to the end of your prompt (with two leading hyphens) that control how Midjourney generates images. Remember to place parameters after the descriptive text and separate them with spaces ([Parameter List – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32859204029709-Parameter-List#:~:text=Parameters%20are%20like%20special%20instructions,or%20using%20a%20Personalization%20profile), [Parameter List – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32859204029709-Parameter-List#:~:text=1,punctuation%20marks%20in%20your%20parameters)). ### Aspect Ratio (`--ar` or `--aspect`) - Specifies the width‑to‑height proportion of the generated image. Midjourney images start as 1:1 squares by default ([Aspect Ratio – Midjourney](https://docs.midjourney.com/hc/en-us/articles/31894244298125-Aspect-Ratio#:~:text=Midjourney%20images%20start%20as%20squares%2C,ar)). - Use `--ar W:H` where W and H are whole numbers (no decimals). Common ratios include 1:1 (square), 4:3 (slightly rectangular), 2:3 (portrait photography), 16:9 (widescreen) and 9:16 (vertical mobile) ([Aspect Ratio – Midjourney](https://docs.midjourney.com/hc/en-us/articles/31894244298125-Aspect-Ratio#:~:text=1%3A1%20%E2%80%93%20Perfect%20square%2C%20popular,mobile%20content%20on%20social%20media)). - The aspect ratio influences composition but is separate from resolution; extremely wide or tall ratios are experimental ([Aspect Ratio – Midjourney](https://docs.midjourney.com/hc/en-us/articles/31894244298125-Aspect-Ratio#:~:text=,and%20may%20produce%20unpredictable%20results)). - Set a default aspect ratio in settings, or append `--ar` to a specific prompt. ### Chaos / Variety (`--chaos` or `--c`) - Adds variation among the four images Midjourney produces. Default chaos is 0; values range from 0‑100 ([Chaos / Variety – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32099348346765-Chaos-Variety#:~:text=What%20is%20Chaos%3F)). - Higher chaos values yield images that diverge more from one another and may not adhere closely to the prompt ([Chaos / Variety – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32099348346765-Chaos-Variety#:~:text=What%20is%20Chaos%3F)). - Add `--c #` at the end of your prompt to set the desired level of variety. ### Exclude Elements (`--no`) - Excludes unwanted elements from your image. List one or more items after `--no` to instruct Midjourney not to include them ([No – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32173351982093-No#:~:text=What%20is%20)). - Separate multiple terms with commas: e.g., `--no fruit, apple, pear` ([No – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32173351982093-No#:~:text=For%20example%2C%20if%20you%27re%20creating,by%20separating%20them%20with%20commas)). - Avoid phrases that could be misinterpreted (e.g., `--no modern clothing` is parsed as “no modern” and “no clothing”); instead specify what you *do* want ([No – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32173351982093-No#:~:text=Midjourney%27s%20moderation%20system%20reads%20every,no)). ### Personalization Profiles (`--profile` or `--p`) - Use personalized moodboards or profiles to apply a custom style. Specify a saved profile name or ID (see Midjourney personalization documentation for details). ### Quality (`--quality` or `--q`) - Controls GPU time spent and level of detail. In Version 7, the default quality is 1; you can set 2 or 4 to spend more GPU time and get more detail ([Quality – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32176522101773-Quality#:~:text=What%20is%20Quality%3F)). - `--q 2` doubles the GPU time; `--q 4` quadruples it. `--q 4` is incompatible with Omni Reference, and `--q 3` is unsupported (it defaults to 4) ([Quality – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32176522101773-Quality#:~:text=Quality%20controls%20how%20much%20GPU,the%20details%20and%20textures%20look)). - The quality setting applies only to the initial set of images (not variations or upscales) ([Quality – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32176522101773-Quality#:~:text=Keep%20in%20mind%2C%20the%20quality,7%2C%20or%20upscales)). ### Repeat (`--repeat` or `--r`) - Generates multiple sets of images from a single prompt. Provide a number to indicate how many times to repeat. Useful for batch generation. ### Seed (`--seed`) - Locks the random noise pattern used to start image generation, enabling reproducible tests. Seeds can be any integer from 0 to 4,294,967,295 ([Seeds – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32604356340877-Seeds#:~:text=What%20are%20Seeds%3F)). - Without a seed, Midjourney chooses a random seed each time, yielding varied results. - Good for testing changes in prompts: keep the same seed while altering other parameters ([Seeds – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32604356340877-Seeds#:~:text=)). - Seeds do not save style or character consistency and may behave unpredictably across sessions ([Seeds – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32604356340877-Seeds#:~:text=Image%3A%20thumb,references%20%2C%20and%20%206)). - Not reliable when using Turbo Mode ([Seeds – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32604356340877-Seeds#:~:text=Image%3A%20thumb,seed%20locking%20is%20not%20reliable)). ### Stealth and Public Modes (`--stealth` / `--public`) - `--stealth` hides your generated images from the public gallery (available on certain plans). - `--public` explicitly publishes your images. ### Raw Mode (`--raw`) - Turns off Midjourney’s automatic stylistic adjustments to give you more control. Produces more realistic, photo-like results for simple prompts ([Raw Mode – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32634113811853-Raw-Mode#:~:text=What%20is%20Raw%20Mode%3F)). - Compatible with versions 5.1 and later ([Raw Mode – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32634113811853-Raw-Mode#:~:text=What%20is%20Raw%20Mode%3F)). - Add `--raw` to the end of your prompt or enable Raw Mode in settings ([Raw Mode – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32634113811853-Raw-Mode#:~:text=Using%20Raw%20Mode)). ### Stylize (`--stylize` or `--s`) - Adjusts how literal or artistic the output is. Low values keep images close to your prompt; high values allow more creative interpretation ([Stylize – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32196176868109-Stylize#:~:text=Think%20of%20stylize%20as%20a,exact%20details%20of%20your%20prompt)). - Default stylize is 100; range is 0–1000 ([Stylize – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32196176868109-Stylize#:~:text=So%2C%20stylize%20lets%20you%20choose,or%20more%20creative%20and%20artistic)). - Use `--s #` to set the stylize value. You can also choose a default stylization in settings ([Stylize – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32196176868109-Stylize#:~:text=%2A%20Add%20%60,of%20your%20prompt%20in%20Discord)). ### Style Reference (`--sref`) and Style Weight (`--sw`) - Use `--sref` followed by an image URL to apply the look and feel of another image to your prompt ([Style Reference – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32180011136653-Style-Reference#:~:text=What%20is%20a%20Style%20Reference%3F), [Style Reference – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32180011136653-Style-Reference#:~:text=Using%20a%20Style%20Reference)). - Combine multiple image URLs separated by spaces to blend styles. - The `--sw` parameter (0‑1000, default 100) controls how strongly the style influences your new image ([Style Reference – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32180011136653-Style-Reference#:~:text=Style%20Weight)). - You can request a random style by using `--sref random`; Midjourney will convert it to a style code which you can reuse ([Style Reference – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32180011136653-Style-Reference#:~:text=The%20Style%20Reference%20feature%20has,or%20switch%20to%20V6)). - Use `--sv` to select style reference versions; the default is 6, while `--sv 4` uses the old V7 model ([Style Reference – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32180011136653-Style-Reference#:~:text=The%20Style%20Reference%20feature%20has,or%20switch%20to%20V6)). ### Tile (`--tile`) - Creates a seamless repeating pattern by ensuring that the edges of the image match up with themselves ([Tile – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32197978340109-Tile#:~:text=What%20is%20)). - Useful for wallpaper or fabric designs. Adds `--tile` to your prompt; note that upscaling often breaks the seamless pattern ([Tile – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32197978340109-Tile#:~:text=When%20you%20use%20%60,how%20multiple%20tiles%20fit%20together)). ### Version (`--version` or `--v`) - Selects the Midjourney model version to use. Each version has different capabilities and aesthetics. - Current default is Version 7 (released April 3 2025, default since June 17 2025) ([Version – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32199405667853-Version#:~:text=Version%207)). - Add `--v #` to the end of your prompt to specify a version ([Version – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32199405667853-Version#:~:text=Setting%20a%20Version)). ### Draft Mode (`--draft`) - Generates lower-fidelity images 10× faster using half the GPU cost ([Draft & Conversational Modes – Midjourney](https://docs.midjourney.com/hc/en-us/articles/35577175650957-Draft-Conversational-Modes#:~:text=What%20is%20Draft%20Mode%3F)). - Activate Draft Mode in the UI or by appending `--draft` to a prompt ([Draft & Conversational Modes – Midjourney](https://docs.midjourney.com/hc/en-us/articles/35577175650957-Draft-Conversational-Modes#:~:text=Using%20Draft%20Mode)). - After drafting, you can “enhance” an image to regenerate it with a higher quality setting ([Draft & Conversational Modes – Midjourney](https://docs.midjourney.com/hc/en-us/articles/35577175650957-Draft-Conversational-Modes#:~:text=Enhancing%20Your%20Drafts)). ### Weird (`--weird` or `--w`) - Adds experimental or quirky elements. Default is 0; values up to 3,000 are allowed ([Weird – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32390120435085-Weird#:~:text=What%20is%20Weird%3F)). - Use higher values for more unconventional results. Weird is experimental and may change over time; it is not compatible with seeds ([Weird – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32390120435085-Weird#:~:text=By%20default%2C%20weird%20is%20set,not%20fully%20compatible%20with%20seeds)). ### GPU Speed Parameters (`--fast`, `--relax`, `--turbo`) - Override your default generation speed. `--fast` uses your Fast time for quicker image generation; `--relax` uses unlimited Relax time (slower but unlimited); `--turbo` uses high‑speed GPUs to generate images up to four times faster but consumes double Fast time ([GPU Speed (Fast, Relax, Turbo) – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32016412137741-GPU-Speed-Fast-Relax-Turbo#:~:text=GPU%20Speed%20), [GPU Speed (Fast, Relax, Turbo) – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32016412137741-GPU-Speed-Fast-Relax-Turbo#:~:text=Relax%20Mode)). - These speeds affect how quickly your prompt is processed but do not change the artistic output. ### Image Weight (`--iw`) - Controls how strongly an image prompt influences the final image. Default weight is 1; range is 0‑3 in Version 7 ([Image Prompts – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32040250122381-Image-Prompts#:~:text=Image%20Weight)). - Use `--iw #` along with an image prompt to increase or decrease its influence. ### Additional Parameters - **Repeat (`--repeat`)**: specify the number of image sets to generate. - **Seed (`--sameseed`)**: lock seeds across all images in a grid. - **Stop (`--stop`)**: set the percentage at which image generation stops (e.g., 50 stops midway). - **Permutations / Multi-Prompt & Weights**: advanced features to mix prompts. - **Style Explorer**: use style codes discovered via Midjourney’s Style Explorer to craft new looks. - **Profile (`--profile`)**: apply personalization profiles or moodboards created in Midjourney. - **Niji (`--niji`)**: switch to Midjourney’s anime‑focused model (Niji 6). - **Public Mode (`--public`)**: explicitly publish your creations. These parameters provide granular control over Midjourney’s image generation. You can combine them (e.g., `a neon cityscape --ar 16:9 --s 200 --c 30 --v 7`) to fine‑tune outputs for your workflows. ## Complete Parameter Reference for Image Generation (Version 7) Below is an exhaustive reference of all parameters currently supported for Midjourney image generation in Version 7 (October 2025). Append parameters to the end of a prompt separated by spaces and avoid punctuation ([Parameter List – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32859204029709-Parameter-List#:~:text=Parameters%20are%20like%20special%20instructions,or%20using%20a%20Personalization%20profile)). ### Aspect Ratio (`--ar` or `--aspect`) - Sets the width:height ratio of the canvas. The default ratio is 1:1, but you can choose ratios like 4:3, 2:3, 16:9, or 9:16 ([Aspect Ratio – Midjourney](https://docs.midjourney.com/hc/en-us/articles/31894244298125-Aspect-Ratio#:~:text=Midjourney%20images%20start%20as%20squares%2C,ar), [Aspect Ratio – Midjourney](https://docs.midjourney.com/hc/en-us/articles/31894244298125-Aspect-Ratio#:~:text=1%3A1%20%E2%80%93%20Perfect%20square%2C%20popular,mobile%20content%20on%20social%20media)). - Use whole‑number ratios (e.g., `--ar 5:4`) and avoid decimals. Extremely wide or tall ratios are experimental and may change when upscaling ([Aspect Ratio – Midjourney](https://docs.midjourney.com/hc/en-us/articles/31894244298125-Aspect-Ratio#:~:text=,and%20may%20produce%20unpredictable%20results)). ### Chaos (`--chaos` or `--c`) - Controls the variety among the four generated images. The default is 0; values range from 0 – 100. Higher chaos produces more varied and unexpected results ([Chaos / Variety – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32099348346765-Chaos-Variety#:~:text=What%20is%20Chaos%3F)). ### Omni Reference (`--oref`) - Inserts a specific person or object from a reference image into your prompt. Upload a single reference image and add `--oref `. The feature uses more GPU and is only available on Version 7 ([Omni Reference – Midjourney](https://docs.midjourney.com/hc/en-us/articles/36285124473997-Omni-Reference#:~:text=What%20is%20an%20Omni%20Reference%3F), [Omni Reference – Midjourney](https://docs.midjourney.com/hc/en-us/articles/36285124473997-Omni-Reference#:~:text=Using%20an%20Omni%20Reference)). - When using Omni Reference, provide a descriptive text prompt as well; use style references for consistent look; only one Omni reference can be used per prompt. Not compatible with inpainting/outpainting, variations, pan or zoom, or with fast, draft or quality 4 settings ([Omni Reference – Midjourney](https://docs.midjourney.com/hc/en-us/articles/36285124473997-Omni-Reference#:~:text=Using%20an%20Omni%20Reference)). - Omni Reference weight (`--ow`) controls the influence of the reference. It accepts values from 1 – 1000 and defaults to 100 ([Omni Reference – Midjourney](https://docs.midjourney.com/hc/en-us/articles/36285124473997-Omni-Reference#:~:text=Omni%20Reference%20Weight)). ### No Parameter (`--no`) - Excludes unwanted elements from the image. Use `--no item1, item2` after your prompt to tell Midjourney what to leave out ([No – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32173351982093-No#:~:text=What%20is%20)). - Be cautious with negative phrases: the moderation system reads each word separately, so instead of saying "not winter," describe what you want, such as "summer landscape" ([No – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32173351982093-No#:~:text=Midjourney%27s%20moderation%20system%20reads%20every,no)). ### Personalization (`--p`) - Lets you apply a trained personalization profile or mood‑board to influence the aesthetic of the output. Profiles are unlocked by ranking image pairs in the app. Use `--p` followed by a profile name or code to apply that style ([Personalization – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32433330574221-Personalization#:~:text=What%20is%20Personalization%3F), [Personalization – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32433330574221-Personalization#:~:text=Using%20Personalization)). - Profiles and mood‑boards can be managed in the Midjourney app. Only one personalization can be used per prompt. ### Quality (`--quality` or `--q`) - Adjusts GPU time and detail level for the initial image grid. Default quality is 1; valid values in Version 7 are 0.5 (where available), 1, 2 and 4. Higher values produce more detailed images and consume more GPU time ([Quality – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32176522101773-Quality#:~:text=What%20is%20Quality%3F), [Quality – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32176522101773-Quality#:~:text=Quality%20controls%20how%20much%20GPU,the%20details%20and%20textures%20look)). - Quality affects only the initial four images. Variations and upscaling always use quality 1. `--q 4` is not compatible with Omni Reference and `--q 3` is not available ([Quality – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32176522101773-Quality#:~:text=Quality%20controls%20how%20much%20GPU,the%20details%20and%20textures%20look)). ### Repeat (`--repeat` or `--r`) - Generates the same prompt multiple times in one command. Each repeat uses a full set of GPU minutes. Allowed repeat counts depend on your subscription: Basic (2–4), Standard (2–10) and Pro/Mega (2–40) ([Repeat – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32757107922061-Repeat#:~:text=What%20is%20Repeat%3F)). - Works only in Fast or Turbo modes; `--repeat` is stripped from the prompt before saving, so you must add it again to re‑run ([Repeat – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32757107922061-Repeat#:~:text=run%20your%20prompt%20multiple%20times,the%20end%20of%20your%20prompt)). ### Seed (`--seed`) and Same Seed (`--sameseed`) - `--seed` locks the random noise used to start image generation. Use numbers between 0 and 4 294 967 295; using the same seed with the same prompt produces similar outputs ([Seeds – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32604356340877-Seeds#:~:text=What%20are%20Seeds%3F), [Seeds – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32604356340877-Seeds#:~:text=)). - Seeds do not store stylistic settings and may vary if Midjourney models change. Seeds are not reliable in Turbo Mode ([Seeds – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32604356340877-Seeds#:~:text=Image%3A%20thumb,references%20%2C%20and%20%206), [Seeds – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32604356340877-Seeds#:~:text=Image%3A%20thumb,seed%20locking%20is%20not%20reliable)). - `--sameseed` forces all four images in a grid to share the same seed, producing consistent variations across the set. ### Stealth / Public Mode (`--stealth` and `--public`) - Controls the visibility of your jobs. `--stealth` keeps images private (only available on Pro or Mega plans); `--public` publishes them to the community feed and galleries ([Stealth Mode – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32019750070669-Stealth-Mode#:~:text=Important%3A%20Even%20with%20Stealth%20Mode,7%20while%20using%20Stealth%20Mode)). - Even in stealth mode, images posted in public Discord channels remain visible, so use direct messages for private generation ([Stealth Mode – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32019750070669-Stealth-Mode#:~:text=Important%3A%20Even%20with%20Stealth%20Mode,7%20while%20using%20Stealth%20Mode)). ### Raw Mode (`--raw`) - Disables Midjourney’s default creative style and produces a more photorealistic, literal interpretation of your prompt. Available on Version 5.1 and newer including Version 7. Use `--raw` after your prompt to activate ([Raw Mode – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32634113811853-Raw-Mode#:~:text=What%20is%20Raw%20Mode%3F), [Raw Mode – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32634113811853-Raw-Mode#:~:text=Using%20Raw%20Mode)). ### Stylize (`--stylize` or `--s`) - Adjusts how strongly Midjourney’s default artistic style influences the result. Low values (e.g., 0–50) produce more literal, prompt‑focused images; high values (e.g., 500‑1000) produce more creative and artistic interpretations ([Stylize – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32196176868109-Stylize#:~:text=Think%20of%20stylize%20as%20a,exact%20details%20of%20your%20prompt)). - Valid range is 0‑1000. The default stylize value is 100 unless changed in your settings ([Stylize – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32196176868109-Stylize#:~:text=%2A%20Add%20%60,of%20your%20prompt%20in%20Discord)). ### Style Reference (`--sref`) - Matches the look and feel of one or more reference images. Upload an image and append `--sref ` to have Midjourney adopt its style (colour palette, textures and mood) without replicating objects ([Style Reference – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32180011136653-Style-Reference#h_01JD5G8C3RY38GSHNSFH6H7H31#:~:text=What%20is%20a%20Style%20Reference%3F)). - You can include multiple style references separated by spaces; each contributes to the final aesthetic. Use with descriptive prompts. - Style Weight (`--sw`) controls the influence of the style reference from 0 – 1000 (default 100). Higher values make the style reference more dominant ([Style Reference – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32180011136653-Style-Reference#h_01JD5G8C3RY38GSHNSFH6H7H31#:~:text=Style%20Weight)). - Style Version (`--sv`) selects which style‑version algorithm to use. Version 6 is default; version 4 replicates the v7 style explorer styles. You can also use `--sref random` to pick a random style code ([Style Reference – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32180011136653-Style-Reference#h_01JD5G8C3RY38GSHNSFH6H7H31#:~:text=produce%20the%20same%20styles%20anymore,or%20switch%20to%20V6), [Style Reference – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32180011136653-Style-Reference#h_01JD5G8C3RY38GSHNSFH6H7H31#:~:text=,parameter%20to%20choose%20between%20them)). ### Tile (`--tile`) - Creates a seamless repeating pattern where the edges of the image line up, perfect for wallpapers and textures. Not recommended for upscaling because the pattern can break ([Tile – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32197978340109-Tile#:~:text=What%20is%20), [Tile – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32197978340109-Tile#:~:text=When%20you%20use%20%60,how%20multiple%20tiles%20fit%20together)). ### Version (`--version` or `--v`) - Selects the Midjourney model version. The default is Version 7. You can switch to previous versions (e.g., `--v 6`) when available ([Version – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32199405667853-Version#:~:text=Setting%20a%20Version), [Version – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32199405667853-Version#:~:text=Version%207)). ### Draft Mode (`--draft`) - Generates a low‑fidelity draft up to 10× faster using half the GPU time. Draft images can later be enhanced to full quality via the “Enhance” button. Use `--draft` to enable this mode or select it in the /settings menu ([Draft & Conversational Modes – Midjourney](https://docs.midjourney.com/hc/en-us/articles/35577175650957-Draft-Conversational-Modes#:~:text=What%20is%20Draft%20Mode%3F), [Draft & Conversational Modes – Midjourney](https://docs.midjourney.com/hc/en-us/articles/35577175650957-Draft-Conversational-Modes#:~:text=Using%20Draft%20Mode), [Draft & Conversational Modes – Midjourney](https://docs.midjourney.com/hc/en-us/articles/35577175650957-Draft-Conversational-Modes#:~:text=Enhancing%20Your%20Drafts)). ### Weird (`--weird` or `--w`) - Adds quirky and unconventional elements to your output. Default weirdness is 0; values up to 3000 produce increasingly odd results. Not fully compatible with seeds ([Weird – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32390120435085-Weird#:~:text=What%20is%20Weird%3F), [Weird – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32390120435085-Weird#:~:text=By%20default%2C%20weird%20is%20set,not%20fully%20compatible%20with%20seeds)). ### GPU Speed Modes (`--fast`, `--relax`, `--turbo`) - **Fast Mode** uses your monthly allotment of fast GPU minutes to render images quickly. - **Relax Mode** offers unlimited rendering at slower speeds; available to Standard, Pro and Mega plans. Note that relax mode does not support permutations or repeat parameters ([GPU Speed (Fast, Relax, Turbo) – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32016412137741-GPU-Speed-Fast-Relax-Turbo#:~:text=Relax%20Mode)). - **Turbo Mode** is the fastest but consumes double the fast GPU minutes; only available on Version 6 and 7 and not compatible with Omni Reference or some features ([GPU Speed (Fast, Relax, Turbo) – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32016412137741-GPU-Speed-Fast-Relax-Turbo#:~:text=Turbo%20Mode)). - You can switch modes via the /settings UI or by adding `--fast`, `--relax` or `--turbo` to your prompt ([GPU Speed (Fast, Relax, Turbo) – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32016412137741-GPU-Speed-Fast-Relax-Turbo#:~:text=Switching%20Speeds)). ### Image Weight (`--iw`) - Controls how strongly an image prompt influences the result when mixing images and text. In Version 7 the weight range is 0 – 3 (default 1). Higher values give the image more influence relative to the text ([Image Prompts – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32040250122381-Image-Prompts#:~:text=Image%20Weight)). ### Niji (`--niji`) - Switches to Midjourney’s anime‑focused model (Niji 6) for stylised, Japanese animation‑inspired art. ### Permutations / Multi‑Prompt & Weights - Use curly braces `{}` to create permutations within a prompt. For example, `{red|blue|green} dragon` produces three prompts (“red dragon”, “blue dragon”, “green dragon”). Each set is limited based on your subscription: Basic 4, Standard 10, Pro/Mega 40 variants ([Permutations – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32761322355597-Permutations#:~:text=What%20are%20Permutations%3F)). - You can also vary parameters by putting them in braces (e.g., `{--ar 16:9|--ar 9:16}`). Permutations require Fast or Turbo mode and use GPU minutes per permutation ([Permutations – Midjourney](https://docs.midjourney.com/hc/en-us/articles/32761322355597-Permutations#:~:text=What%20are%20Permutations%3F)). ### Stop (`--stop`) - Stops image generation early, at a percentage of completion. Accepts values from 1–100; smaller values result in blurrier, less developed images, while 100 represents a fully rendered image ([✋ What Is The --Stop Parameter & How To Use It](https://www.dotslinesandpaper.com/what-is-the-stop-parameter-how-to-use-it/#:~:text=What%20Is%20The%20Stop%20Parameter%3F)). - Creative uses include stopping around 50 for dreamy, painterly looks or low values for sketch‑like outputs ([✋ What Is The --Stop Parameter & How To Use It](https://www.dotslinesandpaper.com/what-is-the-stop-parameter-how-to-use-it/#:~:text=Here%20you%20can%20see%20the,90)). `--stop` cannot be used during upscaling ([✋ What Is The --Stop Parameter & How To Use It](https://www.dotslinesandpaper.com/what-is-the-stop-parameter-how-to-use-it/#:~:text=FYI%20,the%20stop%20parameter%20when%20upscaling)). --- You can combine multiple parameters together to fine‑tune your prompts for different workflows—for example: ``` a neon cityscape at night --ar 16:9 --s 200 --c 30 --seed 1234 --repeat 3 --v 7 ``` This reference provides a comprehensive overview of all image‑specific parameters available in Midjourney as of October 2025. Midjourney Prompting – Recent Documentation (October 2025) Understanding Prompts What is a prompt? A prompt is your creative starting point—it's the text or phrase that tells Midjourney what kind of image you want (Prompt Basics – Midjourney). The more thought you put into your prompt, the more unique and interesting the resulting image can be. Prompts can be a single word or a complete phrase and should guide Midjourney’s creative process (Prompt Basics – Midjourney). Prompting Tips & Tricks Keep prompts short and simple. Clear phrases generally produce better images than long lists or overly detailed instructions (Prompt Basics – Midjourney). Choose the right words. Use precise synonyms (e.g., gigantic instead of big) to help Midjourney understand your intent (Prompt Basics – Midjourney). Be specific with numbers. Instead of vague plurals like “cats,” specify “three cats” or use collective nouns such as “a flock of birds” (Prompt Basics – Midjourney). Focus on what you want. Describe desired elements rather than what you don’t want. To explicitly exclude items, use the --no parameter (Prompt Basics – Midjourney). Prompt length & details. Short prompts rely on Midjourney's default style; adding details increases control. Consider including elements like subject, medium, environment, lighting, color, mood, and composition to refine the output (Prompt Basics – Midjourney). Advanced Prompting Combine techniques. After mastering basic prompts, experiment with multiple text phrases, image prompts, style references, and parameters (e.g., aspect ratio) to customize the result (Prompt Basics – Midjourney). Parameters are appended at the end of the prompt and can control aspects like aspect ratio or quality. Explore the community. The Midjourney community offers a Prompt Craft channel on Discord and a Prompt Craft Room on the web, where users share ideas and troubleshoot together (Prompt Basics – Midjourney). Recent Changes in Midjourney (Version 7 – October 2025) Version 7 overview. Released on April 3 2025 and made the default model on June 17 2025, Version 7 interprets text and image prompts more precisely and produces images with richer textures, coherent details (like human bodies and hands) and improved aesthetics compared with earlier versions (Midjourney Update October 2025 : What’s New ). Key features: Draft Mode speeds up image generation for rapid prototyping (Midjourney Update October 2025 : What’s New ). Omni Reference allows you to reference multiple aspects (structure, style, content) of an image for greater control in new outputs (Midjourney Update October 2025 : What’s New ). Flexible Parameters support any aspect ratio, offer strong and subtle variations, enable remixing existing outputs and provide personalization via ranking or moodboards (Midjourney Update October 2025 : What’s New ). Differences from Version 6. Version 7 discontinues character reference and multi‑prompting but introduces improved upscalers, new quality settings (values 1, 2 or 4) and experimental parameters like --exp (Midjourney Update October 2025 : What’s New ). Upcoming releases. Version 7.1 (in testing) aims to improve prompt interpretation, refine Omni Reference and enhance Draft Mode; its release is expected soon (Midjourney Update October 2025 : What’s New ). Version 8 (targeted for late 2025) will feature a new architecture and larger dataset; plans include an advanced image editor and next‑generation video model (Midjourney Update October 2025 : What’s New ). Style system enhancements. On October 6 2025, Midjourney expanded the Style Explorer, increasing the number of styles tenfold. Users can search for styles with keywords, rank styles during style ranking events (earning free fast hours) and influence personalization algorithms (Midjourney Update October 2025 : What’s New ). New & upcoming tools. Midjourney is developing a stand‑alone image editor, a depth reference tool for re‑texturing images, multiple personalization profiles, storytelling tools and Midjourney TV (grid views for videos) (Midjourney Update October 2025 : What’s New ). Using Parameters The Midjourney documentation provides a Parameter List and additional articles about controls like Chaos / Variety, Multi‑Prompts & Weights, No, and Pan. These tools help fine‑tune prompts and should be appended after the main text to customize output (Documentation – Midjourney). Tips for Effective Prompt Design Use visual language and art style cues (e.g., “oil painting,” “cinematic lighting”) to guide the model. Adjust prompt weights to emphasise key elements. Experiment with synonyms and descriptive adjectives; iterate to refine results. Explore the Midjourney community (Discord or website) for additional inspiration and support (Prompt Basics – Midjourney). Midjourney Parameter Reference (Images) The parameters below are special instructions appended to the end of your prompt (with two leading hyphens) that control how Midjourney generates images. Remember to place parameters after the descriptive text and separate them with spaces (Parameter List – Midjourney, Parameter List – Midjourney). Aspect Ratio (--ar or --aspect) Specifies the width‑to‑height proportion of the generated image. Midjourney images start as 1:1 squares by default (Aspect Ratio – Midjourney). Use --ar W:H where W and H are whole numbers (no decimals). Common ratios include 1:1 (square), 4:3 (slightly rectangular), 2:3 (portrait photography), 16:9 (widescreen) and 9:16 (vertical mobile) (Aspect Ratio – Midjourney). The aspect ratio influences composition but is separate from resolution; extremely wide or tall ratios are experimental (Aspect Ratio – Midjourney). Set a default aspect ratio in settings, or append --ar to a specific prompt. Chaos / Variety (--chaos or --c) Adds variation among the four images Midjourney produces. Default chaos is 0; values range from 0‑100 (Chaos / Variety – Midjourney). Higher chaos values yield images that diverge more from one another and may not adhere closely to the prompt (Chaos / Variety – Midjourney). Add --c # at the end of your prompt to set the desired level of variety. Exclude Elements (--no) Excludes unwanted elements from your image. List one or more items after --no to instruct Midjourney not to include them (No – Midjourney). Separate multiple terms with commas: e.g., --no fruit, apple, pear (No – Midjourney). Avoid phrases that could be misinterpreted (e.g., --no modern clothing is parsed as “no modern” and “no clothing”); instead specify what you do want (No – Midjourney). Personalization Profiles (--profile or --p) Use personalized moodboards or profiles to apply a custom style. Specify a saved profile name or ID (see Midjourney personalization documentation for details). Quality (--quality or --q) Controls GPU time spent and level of detail. In Version 7, the default quality is 1; you can set 2 or 4 to spend more GPU time and get more detail (Quality – Midjourney). --q 2 doubles the GPU time; --q 4 quadruples it. --q 4 is incompatible with Omni Reference, and --q 3 is unsupported (it defaults to 4) (Quality – Midjourney). The quality setting applies only to the initial set of images (not variations or upscales) (Quality – Midjourney). Repeat (--repeat or --r) Generates multiple sets of images from a single prompt. Provide a number to indicate how many times to repeat. Useful for batch generation. Seed (--seed) Locks the random noise pattern used to start image generation, enabling reproducible tests. Seeds can be any integer from 0 to 4,294,967,295 (Seeds – Midjourney). Without a seed, Midjourney chooses a random seed each time, yielding varied results. Good for testing changes in prompts: keep the same seed while altering other parameters (Seeds – Midjourney). Seeds do not save style or character consistency and may behave unpredictably across sessions (Seeds – Midjourney). Not reliable when using Turbo Mode (Seeds – Midjourney). Stealth and Public Modes (--stealth / --public) --stealth hides your generated images from the public gallery (available on certain plans). --public explicitly publishes your images. Raw Mode (--raw) Turns off Midjourney’s automatic stylistic adjustments to give you more control. Produces more realistic, photo-like results for simple prompts (Raw Mode – Midjourney). Compatible with versions 5.1 and later (Raw Mode – Midjourney). Add --raw to the end of your prompt or enable Raw Mode in settings (Raw Mode – Midjourney). Stylize (--stylize or --s) Adjusts how literal or artistic the output is. Low values keep images close to your prompt; high values allow more creative interpretation (Stylize – Midjourney). Default stylize is 100; range is 0–1000 (Stylize – Midjourney). Use --s # to set the stylize value. You can also choose a default stylization in settings (Stylize – Midjourney). Style Reference (--sref) and Style Weight (--sw) Use --sref followed by an image URL to apply the look and feel of another image to your prompt (Style Reference – Midjourney, Style Reference – Midjourney). Combine multiple image URLs separated by spaces to blend styles. The --sw parameter (0‑1000, default 100) controls how strongly the style influences your new image (Style Reference – Midjourney). You can request a random style by using --sref random; Midjourney will convert it to a style code which you can reuse (Style Reference – Midjourney). Use --sv to select style reference versions; the default is 6, while --sv 4 uses the old V7 model (Style Reference – Midjourney). Tile (--tile) Creates a seamless repeating pattern by ensuring that the edges of the image match up with themselves (Tile – Midjourney). Useful for wallpaper or fabric designs. Adds --tile to your prompt; note that upscaling often breaks the seamless pattern (Tile – Midjourney). Version (--version or --v) Selects the Midjourney model version to use. Each version has different capabilities and aesthetics. Current default is Version 7 (released April 3 2025, default since June 17 2025) (Version – Midjourney). Add --v # to the end of your prompt to specify a version (Version – Midjourney). Draft Mode (--draft) Generates lower-fidelity images 10× faster using half the GPU cost (Draft & Conversational Modes – Midjourney). Activate Draft Mode in the UI or by appending --draft to a prompt (Draft & Conversational Modes – Midjourney). After drafting, you can “enhance” an image to regenerate it with a higher quality setting (Draft & Conversational Modes – Midjourney). Weird (--weird or --w) Adds experimental or quirky elements. Default is 0; values up to 3,000 are allowed (Weird – Midjourney). Use higher values for more unconventional results. Weird is experimental and may change over time; it is not compatible with seeds (Weird – Midjourney). GPU Speed Parameters (--fast, --relax, --turbo) Override your default generation speed. --fast uses your Fast time for quicker image generation; --relax uses unlimited Relax time (slower but unlimited); --turbo uses high‑speed GPUs to generate images up to four times faster but consumes double Fast time (GPU Speed (Fast, Relax, Turbo) – Midjourney, GPU Speed (Fast, Relax, Turbo) – Midjourney). These speeds affect how quickly your prompt is processed but do not change the artistic output. Image Weight (--iw) Controls how strongly an image prompt influences the final image. Default weight is 1; range is 0‑3 in Version 7 (Image Prompts – Midjourney). Use --iw # along with an image prompt to increase or decrease its influence. Additional Parameters Repeat (--repeat): specify the number of image sets to generate. Seed (--sameseed): lock seeds across all images in a grid. Stop (--stop): set the percentage at which image generation stops (e.g., 50 stops midway). Permutations / Multi-Prompt & Weights: advanced features to mix prompts. Style Explorer: use style codes discovered via Midjourney’s Style Explorer to craft new looks. Profile (--profile): apply personalization profiles or moodboards created in Midjourney. Niji (--niji): switch to Midjourney’s anime‑focused model (Niji 6). Public Mode (--public): explicitly publish your creations. These parameters provide granular control over Midjourney’s image generation. You can combine them (e.g., a neon cityscape --ar 16:9 --s 200 --c 30 --v 7) to fine‑tune outputs for your workflows. Complete Parameter Reference for Image Generation (Version 7) Below is an exhaustive reference of all parameters currently supported for Midjourney image generation in Version 7 (October 2025). Append parameters to the end of a prompt separated by spaces and avoid punctuation (Parameter List – Midjourney). Aspect Ratio (--ar or --aspect) Sets the width:height ratio of the canvas. The default ratio is 1:1, but you can choose ratios like 4:3, 2:3, 16:9, or 9:16 (Aspect Ratio – Midjourney, Aspect Ratio – Midjourney). Use whole‑number ratios (e.g., --ar 5:4) and avoid decimals. Extremely wide or tall ratios are experimental and may change when upscaling (Aspect Ratio – Midjourney). Chaos (--chaos or --c) Controls the variety among the four generated images. The default is 0; values range from 0 – 100. Higher chaos produces more varied and unexpected results (Chaos / Variety – Midjourney). Omni Reference (--oref) Inserts a specific person or object from a reference image into your prompt. Upload a single reference image and add --oref . The feature uses more GPU and is only available on Version 7 (Omni Reference – Midjourney, Omni Reference – Midjourney). When using Omni Reference, provide a descriptive text prompt as well; use style references for consistent look; only one Omni reference can be used per prompt. Not compatible with inpainting/outpainting, variations, pan or zoom, or with fast, draft or quality 4 settings (Omni Reference – Midjourney). Omni Reference weight (--ow) controls the influence of the reference. It accepts values from 1 – 1000 and defaults to 100 (Omni Reference – Midjourney). No Parameter (--no) Excludes unwanted elements from the image. Use --no item1, item2 after your prompt to tell Midjourney what to leave out (No – Midjourney). Be cautious with negative phrases: the moderation system reads each word separately, so instead of saying "not winter," describe what you want, such as "summer landscape" (No – Midjourney). Personalization (--p) Lets you apply a trained personalization profile or mood‑board to influence the aesthetic of the output. Profiles are unlocked by ranking image pairs in the app. Use --p followed by a profile name or code to apply that style (Personalization – Midjourney, Personalization – Midjourney). Profiles and mood‑boards can be managed in the Midjourney app. Only one personalization can be used per prompt. Quality (--quality or --q) Adjusts GPU time and detail level for the initial image grid. Default quality is 1; valid values in Version 7 are 0.5 (where available), 1, 2 and 4. Higher values produce more detailed images and consume more GPU time (Quality – Midjourney, Quality – Midjourney). Quality affects only the initial four images. Variations and upscaling always use quality 1. --q 4 is not compatible with Omni Reference and --q 3 is not available (Quality – Midjourney). Repeat (--repeat or --r) Generates the same prompt multiple times in one command. Each repeat uses a full set of GPU minutes. Allowed repeat counts depend on your subscription: Basic (2–4), Standard (2–10) and Pro/Mega (2–40) (Repeat – Midjourney). Works only in Fast or Turbo modes; --repeat is stripped from the prompt before saving, so you must add it again to re‑run (Repeat – Midjourney). Seed (--seed) and Same Seed (--sameseed) --seed locks the random noise used to start image generation. Use numbers between 0 and 4 294 967 295; using the same seed with the same prompt produces similar outputs (Seeds – Midjourney, Seeds – Midjourney). Seeds do not store stylistic settings and may vary if Midjourney models change. Seeds are not reliable in Turbo Mode (Seeds – Midjourney, Seeds – Midjourney). --sameseed forces all four images in a grid to share the same seed, producing consistent variations across the set. Stealth / Public Mode (--stealth and --public) Controls the visibility of your jobs. --stealth keeps images private (only available on Pro or Mega plans); --public publishes them to the community feed and galleries (Stealth Mode – Midjourney). Even in stealth mode, images posted in public Discord channels remain visible, so use direct messages for private generation (Stealth Mode – Midjourney). Raw Mode (--raw) Disables Midjourney’s default creative style and produces a more photorealistic, literal interpretation of your prompt. Available on Version 5.1 and newer including Version 7. Use --raw after your prompt to activate (Raw Mode – Midjourney, Raw Mode – Midjourney). Stylize (--stylize or --s) Adjusts how strongly Midjourney’s default artistic style influences the result. Low values (e.g., 0–50) produce more literal, prompt‑focused images; high values (e.g., 500‑1000) produce more creative and artistic interpretations (Stylize – Midjourney). Valid range is 0‑1000. The default stylize value is 100 unless changed in your settings (Stylize – Midjourney). Style Reference (--sref) Matches the look and feel of one or more reference images. Upload an image and append --sref to have Midjourney adopt its style (colour palette, textures and mood) without replicating objects (Style Reference – Midjourney). You can include multiple style references separated by spaces; each contributes to the final aesthetic. Use with descriptive prompts. Style Weight (--sw) controls the influence of the style reference from 0 – 1000 (default 100). Higher values make the style reference more dominant (Style Reference – Midjourney). Style Version (--sv) selects which style‑version algorithm to use. Version 6 is default; version 4 replicates the v7 style explorer styles. You can also use --sref random to pick a random style code (Style Reference – Midjourney, Style Reference – Midjourney). Tile (--tile) Creates a seamless repeating pattern where the edges of the image line up, perfect for wallpapers and textures. Not recommended for upscaling because the pattern can break (Tile – Midjourney, Tile – Midjourney). Version (--version or --v) Selects the Midjourney model version. The default is Version 7. You can switch to previous versions (e.g., --v 6) when available (Version – Midjourney, Version – Midjourney). Draft Mode (--draft) Generates a low‑fidelity draft up to 10× faster using half the GPU time. Draft images can later be enhanced to full quality via the “Enhance” button. Use --draft to enable this mode or select it in the /settings menu (Draft & Conversational Modes – Midjourney, Draft & Conversational Modes – Midjourney, Draft & Conversational Modes – Midjourney). Weird (--weird or --w) Adds quirky and unconventional elements to your output. Default weirdness is 0; values up to 3000 produce increasingly odd results. Not fully compatible with seeds (Weird – Midjourney, Weird – Midjourney). GPU Speed Modes (--fast, --relax, --turbo) Fast Mode uses your monthly allotment of fast GPU minutes to render images quickly. Relax Mode offers unlimited rendering at slower speeds; available to Standard, Pro and Mega plans. Note that relax mode does not support permutations or repeat parameters (GPU Speed (Fast, Relax, Turbo) – Midjourney). Turbo Mode is the fastest but consumes double the fast GPU minutes; only available on Version 6 and 7 and not compatible with Omni Reference or some features (GPU Speed (Fast, Relax, Turbo) – Midjourney). You can switch modes via the /settings UI or by adding --fast, --relax or --turbo to your prompt (GPU Speed (Fast, Relax, Turbo) – Midjourney). Image Weight (--iw) Controls how strongly an image prompt influences the result when mixing images and text. In Version 7 the weight range is 0 – 3 (default 1). Higher values give the image more influence relative to the text (Image Prompts – Midjourney). Niji (--niji) Switches to Midjourney’s anime‑focused model (Niji 6) for stylised, Japanese animation‑inspired art. Permutations / Multi‑Prompt & Weights Use curly braces {} to create permutations within a prompt. For example, {red|blue|green} dragon produces three prompts (“red dragon”, “blue dragon”, “green dragon”). Each set is limited based on your subscription: Basic 4, Standard 10, Pro/Mega 40 variants (Permutations – Midjourney). You can also vary parameters by putting them in braces (e.g., {--ar 16:9|--ar 9:16}). Permutations require Fast or Turbo mode and use GPU minutes per permutation (Permutations – Midjourney). Stop (--stop) Stops image generation early, at a percentage of completion. Accepts values from 1–100; smaller values result in blurrier, less developed images, while 100 represents a fully rendered image (✋ What Is The –Stop Parameter & How To Use It). Creative uses include stopping around 50 for dreamy, painterly looks or low values for sketch‑like outputs (✋ What Is The –Stop Parameter & How To Use It). --stop cannot be used during upscaling (✋ What Is The –Stop Parameter & How To Use It). You can combine multiple parameters together to fine‑tune your prompts for different workflows—for example: a neon cityscape at night --ar 16:9 --s 200 --c 30 --seed 1234 --repeat 3 --v 7 This reference provides a comprehensive overview of all image‑specific parameters available in Midjourney as of October 2025. Select a repo