Understanding and using MidJourney parameters is like unlocking a treasure chest of creative possibilities. These parameters are special tools that help you fine-tune your artwork, giving you more control over how your images look. By getting comfortable with adding these parameters to your prompts, you can create stunning, customized visuals that stand out. Whether you’re aiming for a specific style, adjusting the size of your image, or experimenting with different effects, knowing how to use MidJourney parameters can take your creativity to the next level. Plus, mastering these tools can make your projects more unique and engaging, helping you get noticed in the crowded world of digital content or AI Art.
MidJourney commands and parameters might seem tricky at first, but most people pick them up quickly with a little practice. In this this guide we’ll explain what each parameter, command, and advanced prompt technique does. But first, let’s clarify the differences between these categories:
- Parameters start with double dashes “–“. Although often grouped with “commands,” they aren’t the same. Parameters are options added to your MidJourney prompts to fine-tune the images you create.
- Commands begin with a “/”. They adjust MidJourney’s settings and usually apply to all your future images, like making all your images private or changing how fast they generate.
- Advanced prompts are activated in various ways. This broad category includes everything from using images in your prompts to adjusting the importance of different parts of your prompt.
Image Parameters
Style Reference Numbers (--sref [number])
Purpose: Dictates the artistic style the AI should emulate. These styles can range from specific art movements (like Renaissance) to modern digital styles (like cyberpunk).
Usage: Can be combined for hybrid styles, e.g., --sref 104 --sref 108 for a watercolor cyberpunk look.
Aspect Ratio (--ar [aspect ratio])
Purpose: Defines the shape of the image. A 1:1 ratio creates a square image, while 16:9 is widescreen.
Considerations: Aspect ratio affects the composition and framing of elements within the image.
Quality (--q [quality]) range: 0-2
Purpose: Higher quality values increase detail and resolution but require more processing time.
Range: 0.25 for quick drafts, 1 for standard quality, up to 2 or more for high detail.
Seed (--seed [number])
Purpose: Sets a deterministic seed for the random elements of image generation, allowing for repeatable results.
Application: Useful for iterating on a specific composition or when consistency is required.
Stylization (--stylize [number]) range: 0-1000
Purpose: Adjusts the degree of artistic interpretation. Lower values maintain realism, while higher values add more stylization.
Range: From subtle (e.g., --stylize 250) to highly artistic (e.g., --stylize 1000).
Chaos (--chaos [number]) range: 0-100
Purpose: Introduces randomness in the image generation process, affecting layout and element placement.
Range: 0 for controlled compositions, up to 100 for highly unpredictable layouts.
Exclusion (--no [element])
Purpose: Ensures specific elements are omitted from the generated image, useful for avoiding unwanted themes.
Example: --no people to avoid human figures.
Weirdness Level (--weird [number], --w [number]) range: 0-3000
Purpose: Adjusts the level of creative distortion in your images. Use –weird to introduce more unusual and abstract elements, or dial it down for more realistic results.
Application: Ideal for projects where you want to push the boundaries of creativity. Experiment with different levels of weirdness to find the right balance between reality and surrealism, depending on your artistic vision.
Weirdness vs. Chaos vs. Stylize
Chaos: Adjusts the variety among the initial grid images. Use –chaos to create more diverse or more consistent options.
Stylize: Controls how much MidJourney’s default style influences the final image. Higher values create more stylized, artistic outputs.
Weird: Modifies how unusual or abstract the image becomes compared to typical MidJourney outputs. Use –weird to push the boundaries of creativity and generate more unconventional visuals.
Tile (--tile)
Purpose: Creates an image that can seamlessly tile, ideal for backgrounds or textures.
Application: Useful in design work where repeating patterns are needed.
Niji (--niji)
Purpose: Applies a specific art style, often characterized by vibrant colors and a particular aesthetic.
Application: Ideal for creating images with a distinct, colorful style.
History: Niji is a colaboration between Midjourney and and Indie Game/AI Developer Spellbrush.
Model Version (--version, --v)
Purpose: Allows you to choose different model versions in MidJourney to optimize for coherency, efficiency, quality, and style. Each model version is designed to excel at producing specific types of images.
Application: Use the –version parameter (or –v for short) to switch between model versions. Available options include 1, 2, 3, 4, 5, 5.0, 5.1, 5.2, 6, and 6.1. The latest default model is –v 6.1, offering the most advanced features. You can also change model versions through the /settings command by selecting your preferred option.
Development: Midjourney’s Hisory has seen significant advancements from version 1 to version 6.1, with each iteration bringing major improvements in image quality, coherency, and style. Starting from basic, abstract outputs in V1, the models have evolved to produce highly detailed and realistic images in V6.1. The progress is so remarkable that future versions may create images indistinguishable from actual photographs, showcasing the potential of AI-generated art.
Upscale/Adjustment Parameters
Upscale – Pan/Direction
Purpose: Select the image from the grid to upscale (U1, U2, U3, U4)
Effect: Upscales original image, shifts the image up, down, left, or right
Vary Region + Remix
Purpose: Allows you to selectively regenerate specific parts of an upscaled image by combining Remix Mode with the Vary Region editor. This tool lets you refine and adjust areas of an image using a new or modified prompt.
Application: After upscaling an image, the Vary (Region) button becomes available. You can select a specific area, and the new result will be guided by the content of the original image, the selected region, and your modified prompt. This feature is compatible with MidJourney Model Versions V5, V6, niji 5, and niji 6, enabling precise and creative edits.
After varying the region and achieving the desired changes, you can upscale the image again to enhance its detail and resolution. This process allows you to continue refining the image, making further adjustments or building on the existing design. By repeating these steps, you can create increasingly sophisticated and polished visuals, perfecting your image until it matches your vision.
Once you’ve chosen your final images, you can refine them further to create dramatically different visuals. By making additional adjustments, altering specific elements, or experimenting with new prompts, you can transform the refined images into something entirely new. This process allows you to explore different styles and concepts, resulting in images that may be vastly different from the original, showcasing the flexibility and creative potential of MidJourney.
Repeat (--repeat, --r) range: 0-40
Purpose: Automatically runs a prompt multiple times, allowing for quicker exploration of variations. Combine with other parameters like –chaos to rapidly experiment with different visual outcomes.
Application: Ideal for when you need to generate multiple versions of an image. Basic subscribers can use –repeat with values from 2 to 4, Standard subscribers from 2 to 10, and Pro and Mega subscribers from 2 to 40. Note that –repeat works only in Fast and Turbo GPU modes.
Video (--video)
Purpose: Generates a short movie showing the creation process of your initial image grid. This feature allows you to see how your image evolves from start to finish.
Application: Use the –video parameter to create a video of the image grid generation. Once the process is complete, react to the finished job with the envelope ✉️ emoji to receive a link to the video via Direct Messages from the MidJourney Bot. Note that –video is only applicable to image grids, not upscales, and works with Model Versions 5.2, 6, niji 5, and niji 6, as well as Legacy Model Versions 1, 2, 3, 5.0, 5.1, test, and testp. Generated videos are deleted 30 days after creation.
Above video generated using prompt “jellyfish –video”, after generation respond with :envelope: to receive download link for video of image generation
Fast and Slow Modes (--fast, --slow)
Purpose: Balances speed versus quality. Use --fast for quicker drafts and --slow for detailed final renders.
Application: Choose based on project requirements and deadlines.
Parameters
Parameters are what you add directly into your prompts. After typing two dashes “–“, you enter the parameter name and any additional information if needed. Parameter names often have both long and short versions. The following chart shows a range of common parameters that can be easily added to your prompts, for an even more in depth list check out Midjourney’s Full Parameter List:
| Parameter | Description | Usage Example | Notes |
|---|---|---|---|
--sref [number] | Specifies a style reference number to guide the AI in generating a specific artistic style. | --sref 104 | Use specific numbers associated with different styles. Can combine multiple styles (if supported). |
--ar [aspect ratio] | Sets the aspect ratio of the image, defining the width-to-height ratio. | --ar 16:9 | Common ratios: 1:1 (square), 4:3 (standard), 16:9 (widescreen). |
--q [quality] | Determines the quality level of the output. Higher quality uses more resources and time. | --q 2 | Common values: --q 0.5 (low), --q 1 (default), --q 2 (high). |
--seed [number] | Sets a specific seed for randomization, allowing for reproducible results. | --seed 12345 | Useful for recreating a particular output or experimenting with slight variations. |
--v [version] | Specifies the version of the AI model to use, which can affect the style and quality of the output. | --v 5 | Versions may differ in terms of style, quality, and available features. |
--stylize [number] | Adjusts the level of artistic stylization applied to the image. | --stylize 1000 | Range typically from 0 (minimal stylization) to 1000 or more (maximum stylization). |
--chaos [number] | Controls the randomness of the image composition and elements. | --chaos 50 | Range from 0 (more control, less random) to 100 (more randomness, chaotic compositions). |
--no [element] | Excludes certain elements from the image, ensuring they are not included. | --no trees | Can be used to avoid unwanted objects or themes. |
--fast | Prioritizes faster generation time at the expense of quality and detail. | --fast | Useful when quick drafts are needed. |
--slow | Prioritizes quality and detail over speed, using more time and resources. | --slow | Best for final renders where detail is crucial. |
--tile | Creates a tileable pattern, useful for creating textures and backgrounds. | --tile | Ideal for seamless backgrounds or repeating patterns. |
--niji | Uses the Niji mode, which applies a specific art style often associated with vibrant colors and anime. | --niji | Named after a particular style; check for updates or specific style descriptions. |
--r [number] | Re-rolls the prompt to generate a different image variation without changing the prompt itself. | --r 2 | Useful for exploring different compositions or interpretations of the same prompt. |
| –video | Generates a short movie showing the creation process of your initial image grid | –video | This feature allows you to see how your image evolves from start to finish. |
Commands
Commands in MidJourney are powerful tools that allow you to customize and control your image generation experience. Unlike parameters, which modify individual prompts, commands are used to change settings, manage your workspace, and interact with the MidJourney platform in various ways. By learning and using these commands, you can streamline your creative process, access advanced features, and better manage your projects. The following table lists essential MidJourney commands that can help you get the most out of your AI-generated art.
| Command | Description | Usage |
|---|---|---|
/imagine | Generates images based on a text prompt. | /imagine [your prompt] |
/settings | Opens a menu to adjust various MidJourney settings, such as model version, style strength, and more. | /settings |
/prefer option set | Sets a custom preference option, allowing you to create and save personal settings for future prompts. | /prefer option set [option_name] [value] |
/prefer suffix | Sets a suffix that will be automatically appended to the end of every prompt. | /prefer suffix [text] |
/private | Makes your image generations private so that only you can see them. | /private |
/public | Returns your image generations to public visibility. | /public |
/info | Displays your current usage stats, including jobs done, remaining fast hours, and subscription tier. | /info |
/blend | Blends two or more images together to create a new image. | /blend [image1] [image2] [image3...] |
/help | Provides a list of available commands and general help information. | /help |
/relax | Switches to relaxed mode, where image generation is slower but doesn’t consume fast GPU hours. | /relax |
/fast | Switches to fast mode, where image generation is quicker but consumes fast GPU hours. | /fast |
/subscribe | Provides a link to manage your subscription. | /subscribe |
/show | Displays a list of all the images you’ve generated recently. | /show |
/redo | Re-runs the last prompt you generated with the same settings. | /redo |
/upscale | Upscales a selected image to a higher resolution. | /upscale [image_id] |
/zoom | Allows you to zoom into a specific region of an upscaled image. | /zoom [image_id] [zoom level] |
/remix | Activates remix mode, which allows you to edit and refine images during the generation process. | /remix |
/beta | Switches to the beta version of MidJourney, which may have new features and settings. | /beta |
/public | Makes your next image generation public. | /public |
