HomeAI News
parameter list

parameter list

March 30th, 2023
View OriginalTranslated by Google
Parameters are options added to the prompt that change how the image is generated. Parameters can change the aspect ratio of the image, switch the Midjourney model version, change the Upscaler used, and more.

Arguments are always added to the end of the prompt. You can add multiple parameters to each prompt.

Are iOS devices supported? Many iOS devices automatically change double dashes (--) to em dashes (—). Midjourney supports both symbols.

Basic parameters

aspect ratio

--aspect or --ar Change the generated aspect ratio.


--chaos <number 0–100> Change the chaos of the result. Higher values produce more unusual and unexpected results.


--no negative hint, --no plants will try to remove plants from the image.


--quality <.25, .5, 1, or 2> or --q <.25, .5, 1, or 2> How much time you want to spend rendering quality. The default is 1, with higher values costing more and lower values costing less.


--seed <integer between 0–4294967295> The midjourney robot uses the seed number to create visual noise fields, such as TV stills, as a starting point for generating an initial image grid. The seed number is randomly generated for each image, but can be specified with the --seed or --sameseed arguments. Using the same seed number and hint will produce a similar final image.


--stop <integer between 10–100> Use the --stop parameter to advance jobs during processing. Stopping a job at an earlier percentage may result in blurry, less detailed results.


--style <4a, 4b or 4c> Switch between Midjourney model version 4.


--stylize <number> or --s <number> parameter will affect the degree to which Midjourney's default aesthetic style is applied to Jobs.


--uplight results closer to the original mesh image. The enlarged image has less detail and is smoother.


--upbeta results closer to the original grid image. Zoomed-in images add significantly less detail.

Default (Model Version 4)

Default (Model Version 5)

  • Aspect ratios greater than 2:1 are in beta and may produce unpredictable results.

Model version parameter

Midjourney regularly releases new model versions to improve efficiency, consistency and quality. Different models are good at processing different types of images.


--niji A mod focused on anime-style images.

High Definition

--hd Use an earlier model, which can generate larger, less consistent images. This algorithm may work for abstract and landscape images.


--test Use the Midjourney special test model.


--testp Use Midjourney's test model specifically for photography.


--version <1, 2, 3, 4, or 5> or --v <1, 2, 3, 4, or 5> Use a different version of the Midjourney algorithm. The current algorithm (V4) is the default.

upscaling parameter

Midjourney first generates a grid of low-resolution image options for each job. You can use the Midjourney upscaler on any grid image to increase size and add extra detail. There are various upscaling models available to improve image quality.


--uplight results closer to the original mesh image. The enlarged image has less detail and is smoother.


--upbeta results closer to the original grid image. Zoomed-in images add significantly less detail.


Use an alternative upscaler trained to be compatible with the --niji Midjourney Model. This upscaler was created specifically for this purpose.

Other parameters

These parameters apply only to specific early Midjourney models.


--creative Modify test and testp models to be more diverse and creative.

Image Weight

--iw sets the scale of image hint weight relative to text weight. The value range is 0.5~2, the larger the value is, the higher the proportion of the original image will be.

Currently only V5 version is supported. If you want to use V5 version of Midjourney, you can go to 【v5 Tutorial】How to open Midjourney v5 version to learn.


--sameseed Seed value creates a single large random noise field that is applied to all images in the initial grid. When --sameseed is specified, all images in the initial grid use the same starting noise and will produce very similar images.



Save a progress video of generating the initial image grid. When the emoji responds to the completed grid of images in the form of ✉️, it triggers sending the video to your direct message. --video doesn't work when zooming in on images.


Model version and parameter compatibility

deprecated parameters

  • --width and --w (replaced with --aspect)
  • --height and --h (replaced with --aspect)
  • --fast (replaced with --quality)
  • --vibe (now called V1)
  • --hq
  • --newclip
  • --nostretch
  • --old
  • --beta


no dataCoffee time! Feel free to comment