Pre-Generation

Mode Raw

• No styling applied by Midjourney.

• Relies solely on user entered style description.

• Use this when you want more control over the outcome by entering a detailed prompt.

Stylization

• Lower value = more accurate

• Higher value = more creative

Weirdness

• Controls how unusual an image is compared to typical Midjourney images.

• Higher value = quirky, unique, unexpected results

Variety

• Higher value = the resulting 4 images are more varied in style and composition

• Use this when you’re not sure about the style that you want, therefore you get drastically different results in order to find the style you’re looking for.

Speed

• Relax mode: takes longer to generate images

  • The more you use it the slower it gets

  • Uses less GPU time = you get more generations with your plan

• Fast mode: default mode

  • Speed is consistent and not affected by the mount of uses

• Turbo mode: experimental

  • extremely fast generations

  • uses twice the amount of GPU time

Image Reference

• Upload an image by clicking the ‘Plus’ (+) button.

• Character Reference: Hover over the uploaded image and click the person icon

• Style Reference: Hover over the uploaded image and click on the attach button

Post-Generation

Vary

• Subtle: keeps the image structure and changes some elements

• Strong: changes the structure and elements but keeps the styling

Upscale

• Subtle: doubles the size of your image and keeps details very similar to the original.

• Creative: doubles the size of your image but adds new details to the image.

Remix

• Allows you to modify the prompt on a generated image

• Subtle: tries to keep the image as close as possible to the original one

• Strong: Deviates further from the original image to apply the prompt changes (works better for changing text)

Pan

• Adds an additional area that is equal to the original area, in the direction of the arrow, by seamlessly generating the content

• You can use it multiple times to generate a panoramic image or a vertical strip

Zoom

• Zooms out and seamlessly generates the surrounding of the original prompt

• Hover over the ‘2x’ button and click edit to enter a custom zoom-out amount

Rerun

• Submits the original prompt again to come up with new results

Vary Region

• Opens an editor that allows you to select a specific area of the image to modify

• Optionally you can modify the prompt to specify the changes

Use

• Image: Use an image as a structure reference for a new prompt

• Style: Use an image as a style reference for a new prompt

• Prompt: Copies the prompt to the imagine field

Tips & Tricks

Generating Text

• Use quotation marks to determine the text to be generated in the image

• eg. a neon sign that says "Cyberpunk”

• Single quote marks/apostrophes won't work. Midjourney only recognizes "words and phrases in double quotations”

• works better with style raw or lower stylization

• Use ‘Vary Region’ to correct an area with inaccurate text

Generating Tiles

• Type --tile at the end of your prompt

• Generates images that can be used as repeating tiles to create seamless patterns for fabrics, wallpapers and textures.

Prompt Structure for best results

• Subject: person, animal, character, location, object

• Medium: photo, painting, illustration, sculpture, doodle, tapestry

• Environment: indoors, outdoors, on the moon, underwater, in the city

• Lighting: soft, ambient, overcast, neon, studio lights

• Color: vibrant, muted, bright, monochromatic, colorful, black and white, pastel

• Mood: sedate, calm, raucous, energetic

• Composition: portrait, headshot, closeup, birds-eye view