Mastering Midjourney: A Comprehensive Guide to Parameters

Midjourney is a powerful tool for generating images from text prompts. To get the most out of it, understanding its parameters is essential. These parameters let you fine-tune your creations, overriding default settings to achieve specific artistic visions. This guide dives into the world of Midjourney parameters, offering tips and tricks to elevate your image generation game.

Understanding Midjourney Parameters

When you submit a text prompt to Midjourney, the tool uses default settings to control various aspects of the image creation, such as aspect ratio and model version. Parameters allow you to override these defaults, giving you more control over the final result. Parameters are added at the end of your prompt, starting with two dashes followed by the parameter name (or abbreviation), a space, and the desired value.

For example, to specify a 3:4 aspect ratio instead of the default square, you would include `--aspect 3:4` in your prompt. Multiple parameters can be used by separating them with spaces.

Want to supercharge your Midjourney workflow? Check out the Midjourney Automation Suite from TitanXT to streamline your process and unlock new creative possibilities.

Key Midjourney Parameters Explained

Version (--version or -v)

The version parameter determines which Midjourney model is used to generate your images. Different models have different aesthetics and prompt understanding capabilities. Using an older model can create a different look. It's a great way to see how the tool has developed over time.

• To use version 5, add `--version 5` or `-v 5` to your prompt.

• To use the test model, add `--test` to your prompt.

• To use the anime-trained model, add `--ni` followed by the model number.

Aspect Ratio (--aspect or -ar)

Aspect ratio defines the relationship between the width and height of your image. The default is 1:1 (square). To change it, use `--aspect` followed by the desired ratio (e.g., `--aspect 16:9` for widescreen). Note that there are limits to how tall or wide images can be.

Chaos (--chaos or -c)

Chaos controls the diversity of the images in the grid. The default value is 0, and the range is 0 to 100. A higher chaos value results in more diverse and unexpected results. Using a bit of chaos can help you discover new ideas. Consider trying `--chaos 5` to mix up the results a bit.

Stylize (--stylize or -s)

Each Midjourney model has a default visual style. Stylize controls the strength of this style. The default value is 100, and the range is 0 to 1000. Higher values produce more stylized images but may reduce prompt adherence.

Style Reference (--style_reference or -sref)

Style reference is one of Midjourney's most powerful tools. It allows you to apply the aesthetic elements of reference images (color palette, texture, style) to your results. You can use image URLs, style reference codes, or the word "random" after the `sref` parameter.

Style Weight (--style_weight or -sw)

Style weight controls the influence of the style reference on your images. The default is 100, and the range is 0 to 1000. A higher value gives the style reference more weight. For a more subtle effect, try a value of 15.

Style (--style)

The style parameter lets you minimize Midjourney's default aesthetic. Many users use `style raw` to create more photorealistic images.

Ready to take your Midjourney skills to the next level? The Midjourney Automation Suite from TitanXT can help you experiment with these parameters more efficiently.

Quality (--quality or -q)

Quality affects how much time Midjourney spends generating your images. The default value is 1. Lower values generate images faster but with less detail. Higher values take longer but produce more detailed results. Different Midjourney models support different quality values.

Speed: Turbo, Fast, and Relax

These parameters control the speed of image generation. Turbo and fast use your fast GPU hours, while relax does not. Relax mode is not available with all subscriptions. You can set your default speed, but override for a specific job by using the parameter.

Repeat (--repeat or -r)

Repeat tells Midjourney to run the same prompt multiple times. It can only be used with fast or turbo mode. For example, `--repeat 5` will run the prompt five times.

Video (--video)

Video creates a short time-lapse of your initial image grid being generated. To use it, include `--video` in your prompt. This parameter works on both the web and Discord, but it's recommended to use it on Discord. To access the video, react to the completed image grid with the envelope emoji.

Stop (--stop)

Stop ends image generation early, creating softer, more abstract results. It takes values between 10 and 100. For example, `--stop 50` stops at 50% completion.

Image Weight (--iw)

Image weight can only be used with image prompts. It controls how much influence the image has on the results. The default value is 1, and it takes decimal values between 0 and 3.

Character Reference (--cref)

Character reference attempts to match a character from a reference image. Use `--cref` followed by the URL of the image. On the web, you can drag and drop the image and select the character icon.

Character Weight (--cw)

Character weight controls how closely Midjourney adheres to the clothing and hairstyle of the reference character. The default is 100, and the range is 0 to 100. When character weight is 100, Midjourney will lock onto the face and clothing. At 0, it only locks onto the face.

Personalize (--personalize or -p)

Personalize lets you create your own custom Midjourney model based on your preferred visual style. You first need to teach Midjourney about your style. Once enabled, your personalized model overrides the default Midjourney aesthetic.

No (--no)

No tells Midjourney what you don't want in your image. Add `--no` followed by the elements to exclude, separated by commas. This is more effective than exclusionary phrases within the prompt text.

Tile (--tile)

Tile creates repeating patterns. Just add `--tile` to your prompt. The resulting images can be tiled seamlessly. Be sure to use version 6.1 or higher when upscaling, as previous upscalers may break the seamless pattern.

Seed (--seed)

Each Midjourney job starts with a canvas of visual noise linked to a seed number. If you want to use a specific seed number, use the seed parameter (e.g., `--seed 1234`). Using the same seed number multiple times will produce similar results.

Weird (--weird or -w)

Weird creates strange and unexpected results. The default value is 0, and it takes values up to 3000. Higher values create increasingly unusual images. It can take a bit more experimentation to see what value you prefer for a given prompt. I've found that `--weird 5` is a good value to start with.

Conclusion

Mastering Midjourney's parameters empowers you to create truly unique and personalized images. By understanding and experimenting with these tools, you can unlock new levels of creativity and achieve your artistic vision. Don't be afraid to try different combinations and values to see what works best for you.

Want to automate the parameter testing process, and discover hidden combinations that create your vision? The Midjourney Automation Suite from TitanXT can help you discover them!