Mastering Midjourney Parameters: The Complete Guide
Midjourney’s true power lies in its parameters, which offer you precise control over how your images come to life. Parameters are like a toolkit that allows you to fine-tune every aspect of your creations. By adjusting parameters, you can guide the AI’s behavior—from the composition of the image to the colors, lighting, and even the artistic style.
Master these tools, and you can unlock Midjourney’s full creative potential, transforming your prompts into visually stunning pieces of art with greater accuracy and style.
You’ll find a detailed breakdown below of each parameter—what it does, how to use it, and the best scenarios for applying it. With these insights, you’ll be able to navigate Midjourney more confidently, bringing your creative visions to life in ways that feel truly personalized.
Parameters (Click for More Details)
1. Aspect Ratio (--ar)
What it does:
Changes the dimensions (aspect ratio) of your image.
How to use it:
Add --ar X:Y at the end of your prompt, e.g., --ar 16:9 for a widescreen format.
When to use it:
Use this when your image needs to fit a specific layout, like for social media posts (1:1 for Instagram) or landscape scenes (16:9). It’s useful for controlling the final composition based on where the image will be displayed.
2. Chaos (--chaos)
What it does:
Adjusts how random or unpredictable your image will be.
How to use it:
Add --chaos <number 0–100>. Higher numbers lead to more unexpected and varied results.
When to use it:
Ideal for when you’re experimenting with creative concepts or want to push the AI into more unpredictable outcomes. If you're seeking unique or abstract imagery, bump chaos to 70+; for consistency, keep it lower.
3. Character Reference (--cref)
What it does:
The --cref parameter allows you to use an image reference for a character, ensuring that the AI generates consistent visual elements of that character across different images.
How to use it:
Add --cref [image URL] to your prompt. For example, if you have a specific character design in mind, you can include an image link like this: --cref [https://example.com/character.jpg]. This instructs MidJourney to take the visual elements from that image—such as facial features, clothing, or even the character's overall style—and apply them to your new image.
When to use it:
This parameter is perfect for projects where consistency is key, such as graphic novels, animation storyboards, or any creative work featuring the same character in multiple scenes or poses.
4. Fast (--fast)
What it does:
Forces the MidJourney bot to run a job in Fast Mode, even if your default setting is Relax Mode.
How to use it:
Add --fast at the end of your prompt to override the default mode and speed up generation time.
When to use it:
Use Fast Mode when you’re in a hurry or want to test multiple variations quickly without waiting.
5. Image Weight (--iw)
What it does:
Balances the influence of your image prompt relative to the text prompt.
How to use it:
Add --iw <0–3>, where the default is 1. A higher number gives more weight to the image than the text.
When to use it:
If you’re using an image and want the generated result to stay visually similar, increase the image weight. This is handy when you want to stick closely to a visual reference while experimenting with new ideas through text.
6. Niji (--niji)
What it does:
Uses the specially trained Niji model for generating anime-style images.
How to use it:
Add --niji <v1 5 6> to switch to the anime model.
When to use it:
Use different versions to take advantage of updates, improved features, or explore the strengths of older models.
7. No (--no)
What it does:
Removes specific elements from the image.
How to use it:
Add --no <element>, e.g., --no trees to exclude trees from the generated image.
When to use it:
When you want to avoid certain features, like cluttered backgrounds or unwanted objects, use this to narrow the focus of your image. Great for simplifying scenes.
8. Quality (--quality or --q)
What it does:
Sets how much time and rendering power is spent on generating the image.
How to use it:
Use --q <.25 .5 or 1>; 1 is the default for maximum quality, while lower values speed up rendering with less detail.
When to use it:
If you’re working on drafts or exploring ideas, use .5 for quicker results. For final images or high-quality outputs, stick to --q 1.
9. Random Style (--style random)
What it does:
Randomizes the style of the image using built-in style codes.
How to use it:
Add --style random or --style random-16, --style random-64, --style random-128 for different levels of randomization.
When to use it:
This is great when you’re open to creative surprises and want to see the AI’s interpretation of a prompt in various unexpected styles.
10. Relax (--relax)
What it does:
Overrides your default setting and runs the job in Relax Mode, which takes longer but doesn’t use up GPU minutes.
How to use it:
Add --relax at the end of your prompt.
When to use it:
Ideal for times when you’re not in a rush and want to preserve GPU minutes for faster projects later.
11. Repeat (--repeat or --r)
What it does:
Repeats the same job multiple times with slight variations.
How to use it:
Add --repeat <number> to repeat a prompt up to 40 times.
When to use it:
Great for exploring different versions of an idea. If you’re unsure which variation you prefer, this helps you quickly generate multiple options.
12. Seed (--seed)
What it does:
Sets the initial seed to generate consistent images across different runs.
How to use it:
Add --seed <number> to lock in the starting point.
When to use it:
Use this when you want consistent results across multiple images. Perfect for making variations of the same image without losing the original vibe.
13. Stop (--stop)
What it does:
Stops a job partway through, creating a more blurred or incomplete result.
How to use it:
Add --stop <number> to halt the process at any stage, from 10% to 100%.
When to use it:
Use this when you want a softer or less detailed image. Stopping early can give a dreamy or ethereal quality to the output.
14. Style (--style)
What it does:
Switches between different versions of Midjourney’s models and aesthetic presets.
How to use it:
Add --style <raw 4a 4b 4c cute expressive scenic> to control style variants.
When to use it:
Use this to match a specific visual aesthetic. For example, --style cute is great for more playful, animated styles, while --style raw sticks to a more neutral look.
15. Style References Image (--sref)
What it does:
Allows you to reference an external image to guide the style of the image you're generating.
How to use it:
Add --sref [image URL] to your prompt. For example, you could say: --sref [URL of a vibrant painting] to apply the color and style of that image.
When to use it:
Use this parameter when you want your AI-generated image to follow the aesthetic of a specific visual style.
16. Style References Random (--sref random)
What it does:
Allows the AI to randomly choose a style based on a variety of internal aesthetic presets.
How to use it:
Simply add --sref random to your prompt, and Midjourney will apply a random style reference.
When to use it:
This is ideal when you’re experimenting and want a surprise element in your results. You can use this when you’re open to discovering new aesthetics and don’t have a specific style in mind.
17. Style Reference Seeds (--sref seeds)
What it does:
These seeds act like “cheat codes” for consistency, ensuring your images share the same structure and style. A seed is a unique number representing the visual elements of a specific style.
How to use it:
Add --sref [seed number] to your prompt, and the AI will generate images that share the core aesthetics of that seed. For example: --sref 12345 could lock in a certain visual style, allowing you to use it in various contexts.
When to use it:
Use style reference seeds when you want to maintain a consistent look across multiple images or refine an existing image by keeping its structure while making slight variations.
18. Stylize (--stylize or --s)
What it does:
Controls how strongly Midjourney’s default aesthetic is applied to the image.
How to use it:
Add --s <number 0–1000>, where higher values lead to more artistic embellishment.
When to use it:
If you want a more artistic or abstract result, increase the stylize value. For realistic images, keep the value lower.
19. Tile (--tile)
What it does:
Creates seamless, repeating patterns that can be tiled.
How to use it:
Simply add --tile to your prompt.
When to use it:
Perfect for creating wallpapers, textures, and backgrounds where the pattern needs to repeat seamlessly.
20. Turbo (--turbo)
What it does:
Runs a job in Turbo Mode for faster generation.
How to use it:
Add --turbo to your prompt.
When to use it:
Use Turbo Mode when you need results quickly, like in live demonstrations or when testing ideas fast.
21. Video (--video)
What it does:
Saves a time-lapse video of the image grid being generated.
How to use it:
Add --video to save the video to your DMs.
When to use it:
Great for educational content or social media posts, where you want to show the progression of an image.
22. Weird (--weird)
What it does:
Applies a strange and experimental aesthetic to your images.
How to use it:
Add --weird <number 0–3000> to apply varying levels of eccentricity.
When to use it:
Perfect for surreal, experimental, or abstract art where the goal is to push boundaries.
23. Version (--version or --v)
What it does:
Switches between different versions of the Midjourney model.
How to use it:
Add --v <1 2 3 4 5 5.1 5.2 6 6.1> to specify which version you want to use.
When to use it:
Use different versions to take advantage of updates, improved features, or explore the strengths of older models.