The Ultimate Guide for Midjourney Parameters Midejourney parameters: aspect ratio, chaos, prompt weight, image weight, no, quality, repeat, seed, stop, style, stylize, and tile

The Ultimate Guide for Midjourney Parameters


Midjourney is a text-to-image artificial intelligence (AI) generative tool. It uses machine learning to generate images from natural language prompts. The tool allows users to create unique and original images quickly and easily, without the need for advanced technical skills. If you have not used Midjourney before, please follow this tutorial to set up your account and get started.

In this tutorial, we will take an in-depth look at Midjourney’s parameters. We will cover:

  • How to change the width-to-height ratio of the Midjourney generated images?
  • How to control the level of variance in AI-generated images?
  • How to adjust the prompt weight and image weight to get the desired output?
  • How to exclude specific elements from the generated images?
  • How to control the quality and styles of the image?
  • How to reproduce an image generation or create repeated prompts?

Resources for this post:

  • Video tutorial for this post on YouTube
  • More video tutorials on AI Art
  • More blog posts on AI Art
The Ultimate Guide for Midjourney Parameters – GrabNGoInfo.com

Let’s get started!

Aspect Ratios

Aspect ratio in Midjourney determines the width-to-height ratio of the generated images. The default aspect ratio in Midjourney is 1:1, which creates square images. However, users can change the aspect ratio by using the --aspect or --ar parameter followed by the desired ratio.

The output below is created using the prompt cute cat , the width and the height for the images are the same because the default aspect ratio is 1 to 1.

Image by Amy @GrabNGoInfo

After adding the aspect ratio parameter and use the prompt cute cat --ar 2:3 , we get the output with the width-height-ratio being 2 to 3.

Image by Amy @GrabNGoInfo

There are four common Midjourney aspect ratios according to the Midjourney website.

--aspect 1:1 Default aspect ratio.
--aspect 5:4 Common frame and print ratio.
--aspect 3:2 Common in print photography.
--aspect 7:4 Close to HD TV screens and smartphone screens.

It is suggested to keep the aspect ratios less than 2:1 because a ratio greater than 2 may produce unpredictable results.

Chaos

Chaos in Midjourney is a parameter that controls the level of variance in the generated images. The chaos parameter adds more randomness to the initial image grids, resulting in more stylized and unexpected image outputs.

The value of the parameter ranges from 0 to 100. A higher value results in more chaotic and varied images, and a lower value results in more reliable and repeatable results. The default chaos value is 0.

Users can adjust the chaos parameter by using the --chaos or --c parameter followed by the desired value (e.g. --chaos 50).

For example, the prompt cat avocado hybrid creates the images below with the default chaos value of 0.

Image by Amy @GrabNGoInfo

The prompt cat avocado hybrid --chaos 100 created images that are more different from one another and have unexpected items such as an orange in the images. Each time the prompt was rerun, the results are varied.

Image by Amy @GrabNGoInfo

Prompt Weight

Prompt weight in Midjourney is used to assign importance to specific parts of a text prompt when generating images. It uses a double colon :: to separate different parts of a prompt, followed by a number immediately after the double colon to assign a weight.

Prompt weights help to emphasize or de-emphasize certain elements of a prompt, giving users more control over the generated images. The prompt weight command tells the Midjourney bot to consider two or more separate concepts within a prompt individually. With prompt weights, users can adjust the emphasis on different parts of a prompt, helping the Midjourney AI to generate more relevant and accurate images.

For example, the prompt ice cream gives us the images of four ice creams.

Image by Amy @GrabNGoInfo

Adding two colons after ice separates the two word, ice and cream, and give them equal importance (Prompt: ice:: cream).

Image by Amy @GrabNGoInfo

When we set the prompt weight to make ice twice as important as cream (Prompt: ice::2 cream), we get images mainly focused on ice.

Image by Amy @GrabNGoInfo

Image Weight

Midjourney can take images as prompts. When a prompt has both image and text, image weight controls the weight of an image in relation to text when generating AI images.

The default image weight is 1 in version 5, but users can adjust this parameter based on their needs between the range of 0 to 2.

Higher image weight values will result in more visually prominent image features in the generated output, while lower values will give more emphasis to the text portion of the prompt. The image weight parameter can be adjusted using the --iw parameter followed by the desired value (e.g. --iw 0.5).

Let’s get the image URL from my previous blog post here. We will use the robot artist image as the image prompt. Right-click the image and click Copy Image Address.

Image by Amy @GrabNGoInfo

In the Midjourney prompt, paste the image URL as the image prompt, cute cat as the text prompt, and set the image weight at a low value of 0.5 (https://s.mj.run/PzfW5aE16bI cute cat — iw 0.5 ).

We can see that the generated images are all cats standing because the robot in the image prompt is standing.

Image by Amy @GrabNGoInfo

After increasing the image weight to the highest value of 2, the generated images are more like robot cats, and the theme color is similar to the input image.

Image by Amy @GrabNGoInfo

No

In Midjourney, the --no parameter is used to prevent the AI from adding a specific element or object to the generated image. For example, using --no cats would instruct the AI to avoid including cats in the generated image. Multiple elements can be excluded by separating them with commas (e.g. --no cats, dogs, trees). The --no parameter can be useful for refining the generated images and focusing on specific aspects of the prompt.

From the previous output, we can see that most of the cat images Midjourney generated are white cats or brown cats. Let’s try to use the no parameter to ask Midjourney to create cats of other colors.

Using the prompt cute cat --no white, brown , Midjourney created four cats with other colors.

Image by Amy @GrabNGoInfo

Quality

The quality parameter in Midjourney is a parameter that determines how much time is spent generating an image and affects the level of detail and complexity in the generated output.

Higher-quality settings take longer to process but produce more detailed images, while lower-quality settings generate images more quickly but with less detail.

The quality parameter can be adjusted using the --quality or --q parameter followed by the desired value (e.g. --quality 0.5). According to Midjourney documentation,

The default --quality value is 1.
--quality only accepts the values: .25, .5, and 1 for the current model. Larger values are rounded down to 1.
--quality only influences the initial image generation.
--quality works with Model Versions 4, 5 and niji 5.

For a prompt skeleton --quality 0.25 , we get the image below.

Image by Amy @GrabNGoInfo

After increasing the quality value from 0.5 to 1, we get the images with more details.

Image by Amy @GrabNGoInfo

Repeat

The Repeat parameter in Midjourney is a feature that enables generating multiple sets of images. To use this feature, you can add the --repeat parameter followed by the number of times you want to repeat the prompt. The range for the repeat parameter is 1 to 40.

This feature is often used as a productivity booster, saving clicks and speeding up iterations for Midjourney users. According to Midjourney documentation,

--repeat is available for Standard and Pro subscribers
--repeat accepts values 2–10 for Standard subscribers.
--repeat accepts values 2–40 for Pro subscribers.
The --repeat parameter can only be used in Fast GPU mode.
Using the redo (re-roll) 🔄 button on the results of a --repeat Job will only re-run the prompt once.

After the prompt input of cute cartoon cat --repeat 2 , Midjourney Bot will ask to confirm if you want to imagine the prompt twice. Click the green Yes button to run the prompts.

Below are the 8 images generated from the repeated prompt.

Image by Amy @GrabNGoInfo
Image by Amy @GrabNGoInfo

Seed

The seed parameter in Midjourney is a value to create a field of visual noise, which acts as a starting point to generate the initial image grids. By default, the Midjourney AI selects a random seed value for each generation process, but users can also specify the seed value using the --seed or --sameseed parameter to create consistent results.

The seed value is an integer between 0 and 4294967295. Seed numbers are not static and should not be relied upon between sessions.

To find the seed number and job ID for a Midjourney job, you can right-click the Midjourney output, then select Add Reaction -> envelope.

Image by Amy @GrabNGoInfo

A direct message will be sent to you from Midjouney Bot with the job and seed information. Note that if you are creating the images in a Discord server, the seed does not show up on the server. You need to go to the direct message with Midjourney Bot to get the information.

Image by Amy @GrabNGoInfo

An alternative way of getting the job ID and seed number is to right-click the Midjourney output, then select Apps -> DM Results. The information will be sent via direct message by Midjourney Bot.

Image by Amy @GrabNGoInfo

To reproduce this Midjourney cartoon cat output, use the prompt cute cartoon cat --seed 1339578580 .

Stop

The stop parameter in Midjourney is a value that sets the percentage of completion for the image generation process. The--stop parameter can take an integer value between 10 and 100. The default value for --stop is 100.

Depending on the situation, using the --stop parameter can help save time and resources. It can create blurrier, less detailed results. However, it is important to note that --stop does not work while upscaling.

Setting the parameter stop at 30% for the same cute cartoon cat image (prompt: cute cartoon cat --seed 1339578580) gives us the output below.

Image by Amy @GrabNGoInfo

Style

The style parameter in Midjourney is a parameter for switching between different versions of Midjourney models to achieve different styles and looks.

For example, the Niji (a Midjourney model for cartoon images generation) Model Version 5 has four style versions, cute, expressive, original, and scenic. According to Midjourney documentation,

--style cute creates charming and adorable characters, props, and settings.
--style expressive has a more sophisticated illustrated feeling.
--style original uses the original Niji Model Version 5, which was the default before May 26th, 2023.
--style scenic makes beautiful backgrounds and cinematic character moments in the context of their fantastical surroundings.

The prompt cat wearing sunglasses --niji 5 --style cute gives us the kawaii cute style cat images.

Image by Amy @GrabNGoInfo

Changing the style to expressive (Prompt: cat wearing sunglasses --niji 5 --style expressive) gave us two cats and two girls wearing cat ears.

Image by Amy @GrabNGoInfo

The original (Prompt: cat wearing sunglasses --niji 5 --style original) gives us images of different styles.

Image by Amy @GrabNGoInfo

The scenic value (Prompt: cat wearing sunglasses --niji 5 --style scenic) gives us four very cool cat images.

Image by Amy @GrabNGoInfo

Stylize

The stylize parameter in Midjourney is a parameter that affects the strength of the stylization of the generated images. With this parameter, users can adjust the level of artistic filter applied to the generated image, resulting in images with varying levels of detail, noise, and texture.

The stylize parameter is --stylize or -s . Low stylization values create images that are closely aligned with the prompt. High stylization values generate images that are very artistic but may not match the prompt very well.

The default value for --stylize is 100, and it can take the values ranges from 0 to 1000.

When we set the stylize parameter to a low value of 50 (Prompt: cat wearing sunglasses --stylize 50), we get relatively realistic images.

Image by Amy @GrabNGoInfo

After increasing the stylize value to 1000 (Prompt: cat wearing sunglasses --stylize 1000), some generated images are more artistic.

Image by Amy @GrabNGoInfo

Tile

The tile parameter in Midjourney is a parameter that generates images that can be used as repeating tiles to create seamless patterns for fabrics, wallpapers, and textures.

For example, we can generate seamless wild flower patterns using the prompt wild flowers --tile .

Image by Amy @GrabNGoInfo

Summary

Midjourney offers different parameters to create customized and visually appealing images, and gives users control in the image generation process. The possibilities are endless when it comes to image creation with Midjourney. Hopefully, this guide provided you with the necessary knowledge to start tweaking and experimenting with these parameters for your artwork. Be sure to follow me and subscribe if you would like to be notified when I publish new tutorials like this. Happy creating!


For more information about data science and machine learning, please check out my YouTube channel and Medium Page or follow me on LinkedIn.

Recommended Tutorials

Leave a Comment

Your email address will not be published. Required fields are marked *