Midjourney User Guide - Quick Start (2)

Prompts are short text phrases used by Midjourney to generate images. Midjourney breaks down the words and phrases in the prompts into smaller parts, called tags, which it can compare with its training data and then use to generate images. Well-crafted prompts can help create creative and high-quality images.

Structure of Prompts

Basic Prompts

Basic prompts are very simple and can be a single word, a phrase, or even an emoji.

Advanced Prompts

More advanced prompts can include one or more image URLs, multiple text phrases, and one or more parameters.

Image Prompts can add image URLs to the prompt to influence the final generated style and content. Image URLs always come at the beginning of the prompt.

Prompt Text is the text description of the image to be generated. Refer to the prompt information and tips below. Well-crafted prompts help generate eye-catching images.

Parameters change the way the image is generated. Parameters can change the aspect ratio, the model version used, the upscaler… Parameters always come at the end of the prompt.

Key Points for Writing Prompts

Length of Prompts

Prompts can be very simple. A single word (even an emoji!) can generate an image. Very short prompts will largely rely on Midjourney’s default style, so descriptive and specific prompts better reflect the effect you want. Of course, very long prompts do not necessarily result in good images. Focus on the core concept of the image you want to create.

Grammar

Midjourney Bot does not understand grammar, sentence structure, or words like humans do.

• Word choice is very important. In many cases, more specific synonyms work better. For instance, instead of “big,” use “gigantic.” Remove redundant words whenever possible. Fewer words mean each word has a stronger impact. Use commas, brackets, and hyphens to organize your thoughts, but know that Midjourney Bot does not interpret short words one-to-one. Note that Midjourney Bot does not consider capitalization.

Midjourney Model Version 4 is better at interpreting traditional sentence structure than other models.

Focus on the Core Concept

Describe what you want rather than what you don’t want. For example, if you say “a party without a cake,” the result may still include a cake. If you want to ensure that an object is not in the final image, try using the --no parameter for the prompt.

Consider Which Details Are Important

Anything not clearly expressed may surprise you. Anything omitted will be randomized. Of course, this is a good way to get diversity, but in this case, you can’t control the details of the image.

Any important background or detail, try to express it clearly. Think about:

• Subject: People, animals, characters, places, objects, etc.
• Medium: Photos, paintings, illustrations, sculptures, graffiti, etc.
• Environment: Indoors, outdoors, on the moon, in Narnia, underwater, in the Emerald City, etc.
• Lighting: Soft, ambient, overcast, neon, studio light, etc.
• Color: Vibrant, pastel, bright, monochromatic, colorful, black and white, soft, etc.
• Mood: Serious, calm, noisy, energetic, etc.
• Composition: Portrait, headshot, close-up, bird’s-eye view, etc.

These are also key points an artist needs to think about before designing.

Use Collective Nouns

Plural words are not always well expressed. Try specific numbers. “Three cats” is more specific than “cats.” Collective nouns can also be used, for example, “flock of birds” instead of “birds.”