Parameters are short settings you add to the end of a Midjourney prompt. They change the shape, style or randomness of the result without changing the subject. This guide reflects Midjourney’s documentation as of October 7, 2026, when V8.2 is the default model. Check the official parameter list for the latest, since options and ranges can change between versions.
Aspect ratio: --ar
Sets the shape of the image, written as width:height. The default is square (1:1).
a lighthouse on a basalt coast at blue hour, long exposure --ar 3:2Use --ar 16:9 for wide scenes, --ar 4:5 for portraits and social posts, and --ar 1:1 for squares.
Stylize: --stylize or --s
Controls how strongly Midjourney applies its own artistic look. The default is 100, and you can set it from 0 to 1000 in recent versions. Lower values stay closer to your words; higher values give Midjourney more freedom.
a quiet street market at dawn, soft light --s 250Chaos: --chaos or --c
Increases variety between the images in a set. The default is 0 and the range is 0 to 100.
an abstract poster of a city at night --chaos 30Weird: --weird or --w
Adds experimental, unconventional results. The default is 0, and values go up to 3000.
a lighthouse in a field of glass flowers --weird 500Exclude things: --no
Tells Midjourney to leave something out.
a minimalist living room, natural light --no clutter, peopleRepeat a starting point: --seed
Reuses a starting point so results are more comparable when you change only one word. It is a testing aid, not a guarantee of identical images.
a red fox in a snowy forest, soft light --seed 1234Version: --v
Chooses the model version. The current default is V8.2. Keep the version constant while you compare prompts.
Personalization: --p
Applies your personalization profile to the prompt. Personalization learns your taste from the images you rate, and your V7 global profile works with V8.2.
a cosy reading nook with rain on the window --pDraft mode: --draft
Creates images faster at lower GPU cost, for quick experiments. In V8.1 and V8.2, one draft prompt generates a batch of 24 images at 512 by 512 pixels and costs 0.4 minutes of GPU time.
a robot gardener watering neon plants --draftReferences and editing in V8.2
Omni Reference and Character Reference were replaced by the Edit Model, added on August 27, 2026. It edits images with instructions, accepts up to four image references and supports inpainting and outpainting. You can drag images into the prompt bar under “attach to prompt”, click “edit” in an image’s lightbox, or use the Edit tab. Style references (--sref), moodboards and personalization work with it.
Put it together
Start without parameters, then add one at a time and note what changes. Our beginner’s guide shows the full process, and the Midjourney review covers plans and models.