Midjourney has updated how style references work. Most srefs on this website now require to specify version with
If you are using the Midjourney Web UI, some of the parameters are accessible directly on the interface. However, you can also use parameters directly in your prompt. Here's a cheat sheet to help you understand the parameters you can use with Midjourney.
--ar width:height
Sets image width-to-height ratio. Default is 1:1
--no item1, item2
Tells Midjourney what not to include in your image. Add multiple items separated by commas. Using --no is more effective than saying "without" or "don't" in the prompt. For example, "still life --no fruit" works better than "still life without fruit".
Note: --no item1, item2 is equivalent to using a negative weight in multi-prompting (item1::-.5, item2::-.5)
--quality .25, .5, or 1 or --q .25, .5, or 1
Determines rendering time and quality. Higher values consume more resources. Default is 1.
--seed 0-4294967295
Sets starting point for generation. Same seed and prompt produces similar images.
--version 1-6 or --v 1-6
Selects the Midjourney algorithm version.
--c 0-100 or --chaos 0-100
Controls result variability. Higher values yield more unexpected outcomes.
--style raw
Applies alternative aesthetics. Raw offers less automatic beautification.
--stylize 0-1000 or --s 0-1000
Controls strength of Midjourney's default aesthetic style.
--fast or --relax or --turbo
Controls processing speed and GPU usage for image generation.
--sref
Uses a number or an image URL as a style reference to maintain consistent visual style across different prompts. Add "random" to use a random style from Midjourney's internal style library.
--sw 0-100
Sets the strength of style influence when using --sref. Higher values create images more similar to the reference style.
--cref
Uses an image URL as a character reference to maintain consistent character appearance across different prompts and scenes.
--cw 0-100
Sets the strength of characterization when using --cref. Lower values focus on face only, higher values include hair and clothing. Default is 100.
--iw 0-3
Adjusts influence of image prompts vs text prompts. Default is 1.
--niji 4 or 5
Generates images in an anime style.
--p
Makes the job private, not visible in public feeds.
--repeat 1-40 or --r 1-40
Runs multiple jobs from a single prompt.
--stop 10-100
Ends generation process early, resulting in less detailed images.
--tile
Generates images suitable for seamless patterns.
--video
Saves a progress video of the initial image grid generation.
--weird 0-3000 or --w 0-3000
Introduces quirky and offbeat qualities to images.
For a comprehensive list and detailed explanations, refer to the official Midjourney documentation.