Creating prompts for the Midjourney AI image generator is an art in itself, and understanding the various parameters available can significantly enhance your ability to generate precise and visually stunning images.
Here’s a detailed explanation for 8 these parameters and how to effectively use them:
1. Aspect Ratio (–ar)
- Purpose: Controls the shape of the generated image, specifically its width-to-height ratio.
- Options:
--ar 1:1: Produces square images. Ideal for profile pictures or icons.--ar 3:2: Creates a landscape format, popular in print photography. Use this for wider scenes.--ar 7:4: Suitable for HD-TV and smartphone screens. Good for wallpapers.--ar 5:4: Common ratio for frames and prints.
2. Chaos (–chaos)
- Purpose: Increases variation and randomness in the results.
- Options:
--chaos 0: Minimal randomness, produces more predictable and consistent images.--chaos 50: Moderate level of variation, introducing some unexpected elements while retaining some predictability.--chaos 100: Maximum randomness, ideal for abstract or highly creative images.
3. Image Weight (–iw)
- Purpose: Determines the importance of an uploaded reference image relative to the text prompt.
- Options:
--iw 0.5: The reference image has a lesser impact, suitable when you want just a hint of its style.--iw 1(default): Balanced influence between the text prompt and the image.--iw 2: The reference image heavily influences the result, useful when replicating a specific style.
4. Quality (–quality or –q)
- Purpose: Affects the rendering time and final image quality.
- Options:
--q .25: Lower quality but faster generation. Useful for quick previews.--q 1(default): Standard quality and rendering time.--q 2: High quality, takes longer to generate. Ideal for final versions of an image.
5. Seed (–seed)
- Purpose: Uses a specific seed number to generate images similar to a previous result.
- Options: Any integer. The specific seed number relates to a previous image you’ve generated and want to mimic.
6. Stop (–stop)
- Purpose: Halts the image rendering process at a specified point.
- Options:
--stop 10: Stops very early in the process, resulting in a very abstract and undefined image.--stop 50: Stops midway, offering a balance between definition and abstraction.--stop 100: Completes the rendering process, providing a fully detailed image.
7. Stylize (–stylize or –s)
- Purpose: Adjusts the influence of Midjourney’s default style on the final image.
- Options:
--s 0: No influence of the default style, relying entirely on the prompt’s specifics.--s 500: Moderate influence of the default style.--s 1000: Maximum influence, resulting in a strong signature style of Midjourney.
8. No (–no)
- Purpose: Attempts to remove a specified element from the image.
- Options: Any descriptive term, like
--no treesto avoid trees in a landscape.
9. Switch Models (–v)
- Purpose: Indicate which model version to use. Each incremental version comes with unique features and capabilities. The current model as of this writing is –v 6. Understanding these model versions can significantly enhance your ability to create specific types of images.
- Using Parameters: Add
--v6,--v 5,--v 5.1,--v 5.2, or--niji 5to your prompt. - Using Settings Command: Type
/settingsin Midjourney and select your preferred version.
Midjourney Tips for Beginners:
- Start Simple: Begin with basic prompts and gradually introduce one parameter at a time to see its effect.
- Experiment: Don’t hesitate to try different combinations of parameters to see how they affect the outcome.
- Balance Is Key: Find a balance between the specificity of your prompt and the creative freedom you allow the AI.
- Create A Private Server: Instead of loosing your request in the continuously scrolling Discord feeds, create a private “room” where it’s just you and Midjourney Bot talking.
By understanding and experimenting with using these Midjourney AI parameters, you can greatly improve the quality and specificity of the images generated by Midjourney.
Remember, the best way to learn is by experimenting and observing the results of different combinations of prompts and parameters.


