Creating Effective Image Prompts

Introduction

Welcome to the Midjourney v5 Bot unofficial user guide. This comprehensive guide is designed to help beginners create effective image prompts, producing visually engaging and accurate images that meet their requirements. Explore the intricacies of clarity, format selection, artistic terminology, multi-prompt basics, advanced techniques, and natural language descriptions. To activate v5 on Discord, type “/settings” and select [MJ Version 5].

Image Prompt Generator

Clarity and Specificity

• For optimal results, provide a detailed and precise description of the scene, object, or character.

• Include crucial aspects like setting, key elements, and their relationships, which enables Midjourney v5 to generate accurate visual representations.

• Keep prompts concise, ideally above 20 words, but always under 60 words.

• Use double colons “::” to separate distinct parts of a prompt, such as objects, settings, and styles.

Example:
a tranquil beach at sunset:: real photo:: silhouettes of people playing volleyball

Selecting the Ideal Format and Style

• Specify your preference for a real photo, illustration, painting or drawing.

• Emphasize this choice in the prompt to ensure the generated image aligns with the desired level of realism and format.

• To create a specific aesthetic, reference at least two elements in the prompt, such as art movements, media types, art techniques, artist names, media genres, media titles, time periods, or geographic locations.

• If no specific style is mentioned, the system defaults to photography.

Example:
a lush forest with a winding path:: digital artwork, style inspired by anime films and japanese prints

Stylization with the --stylize or --s Parameter

• The Midjourney Bot has been trained to produce images that favor artistic color, composition, and forms.

• The “--stylize” or “--s” parameter influences how strongly this training is applied.

• Low stylization values produce images that closely match the prompt but are less artistic (e.g., a more true-to-prompt garden scene). High stylization values create images that are very artistic but may deviate more from the prompt (e.g., a more creative garden scene that resembles a field but is missing the garden).

• The “--stylize” default value is 100 and accepts integer values from 0 to 1000.

• To adjust the stylization level, include “--stylize” or “--s” followed by the desired value in the prompt, like “--stylize 500” or “--s 250”.

Example:
a colorful flower garden:: bright watercolor painting --s 700

Artistic Terms and Techniques

• Infuse your prompt with relevant artistic terms and techniques specific to photography, digital artwork, or painting.

• Experiment with styles and techniques from various eras to guide Midjourney v5 in creating unique and captivating images. Including specific artistic terms can result in images with a more distinct style and time period (e.g., impressionist style inspired by Monet produces an early 1900s setting).

• More detail often translates to improved results.

• Focus on only using visual descriptions.

Example:
a bustling city street in the rain:: painting, impressionist style, inspired by monet

Multi-Prompt Fundamentals

• Separate distinct parts of a prompt using a double colon “::” followed by a space.

• The double colon "::" followed by a space can be used to separate items, concepts, or composition.

• Do not use a single colon “:” in any circumstance.

• Make sure to include all essential elements in the prompt.

Example:
animated race car with two big eyes:: expressive, colorful and big shapes:: driving fast on a detailed rural dirt road, with a huge dust cloud building up behind the car:: in the style of beloved classic animation:: excited whimsical cartoon race car --s 500

Focus on Essential Elements with Weights

• Employ prompt weights to assign importance to specific elements relative to one another.

• Weights are generally useful when you want the bot to focus more on a specific aspect, or when you want to ensure a particular element is more or less prominent in the final image.

• To assign a weight, use the format “element::weight” and ensure there is a space " " after the weight value, before the next word.

• Weights can be positive or negative, provided their sum remains positive.

• Use weights to emphasize or de-emphasize elements in your prompt. The larger the weight, the more focus will be placed on that element relative to the other elements with weights.

• A prompt like “hot::2 dog” will double the weight of “hot” relative to the other elements.

• Use negative weights to remove unwanted elements or aspects and avoid using natural language to describe negative elements. This means that words like “not”, “without” or “missing” cannot be used in natural language to exclude items.

Example:
nature, central stockholm, sweden, humanity destroyed:: ivy and moss growing through cracks in buildings::4 broken signs and shattered brown old glass::2 ruined cracked streets::3 mud, dirt covering traditional decaying swedish architecture::4 infrastructure::-3 functional cars::-3 iso 800, f/5.6, 1/200s shutter speed --no order, clean, people, streetlights --ar 16:9 --s 500

• When you want to create a more specific and focused image prompt, you can combine styles, details, and weights together to guide the image generation.

Example:
a tree house nestled in the branches of a massive oak tree:: painting in style of graphic novel, detailed and dynamic::4 multiple levels connected by rope bridges and ladders:: wooden exterior with small windows:: the windows are covered by colorful drapes hinting at a cozy interior:: surrounded by a lush forest canopy --s 400 --ar 3:2

• In the example above, the style and details are combined to create a unique and specific prompt. The weight of "4" has been assigned to the "painting in the style of a graphic novel, detailed and dynamic" to emphasize this aspect. The other elements are described in natural language and are smoothly integrated into the prompt.

The --no Parameter

• Use the “--no” parameter to exclude undesired elements from the image.

• The “--no” parameter can be used once and must be placed at the prompt’s end but before “--ar”.

• The "--no" parameter automatically assigns a weight of -0.5 to each excluded item. To assign a greater negative value to an item, you can instead give it a negative value in the prompt before the “--no” parameter.

• To exclude multiple elements, separate them with commas, like this: “--no magic, wires, visual illusion, cropped”

Example:
a peaceful countryside scene:: fine art painting --no cars, buildings, power lines

Natural Language Descriptions

• Midjourney v5 understands natural language prompts. Describe the scene as if explaining it to a friend.

• Experiment with different prompting techniques for various scenes, and combine methods for optimal results.

• Starting and ending the prompt with the same or similar words can increase the focus on them.

• While using natural language descriptions, it is often best to separate the image based on concepts rather than items if using the “::” parameter.

• Natural language should never be used after the “--no” parameter.

Example:
a vibrant marketplace full of people shopping for fruits and vegetables during a summer rainstorm:: charcoal and pen artwork

Controlling Aspect Ratios

• Aspect ratios govern image proportions and can be adjusted by appending the “--ar” parameter at the prompt’s end, following the “--no” parameter when used together.

• Define the desired aspect ratio using standard formats like “1:1”, “4:3”, “5:4”, “3:2”, “16:10”, “16:9”, “2:1”, “21:9”, “3:1”, “4:5”, or “5:7”.

• Choose an aspect ratio that best suits the subject or scene being depicted.

Example:
a stunning mountain landscape with a crystal-clear lake in the foreground:: real photo using 1/30s shutter speed --no buildings, vehicles --ar 16:9

Incorporating Metadata and EXIF Data

• Including metadata and EXIF data from photos in the prompts can help generate more accurate and visually engaging images.

• Metadata like camera model, ISO, aperture, and shutter speed may enhance the image generation process and improve the results.

Example:
night_out.jpg was taken on august 12, 2005, at 10:30 pm with a canon powershot sd450 digital camera. the image shows two adult happy women in cocktail dresses, standing outside a nightclub in new york city. the camera settings were iso 400, f/2.8, and 1/30s shutter speed. the image was later edited to adjust the brightness and contrast --ar 4:3

Avoiding Banned Words

• Steer clear of banned words, including sexual and gore-related terms. Innocuous words may also be prohibited, as they could be employed to create inappropriate images. In most cases, common sense will suffice.

Example to avoid:
bare bum exposed

Correctly Combining Parts of the Prompt with Parameters

• When incorporating different elements and parameters in the prompt, ensure that double colons “::” are only used to separate distinct parts of the prompt.

• Do not use double colons “::” before or after parameters like “--s”, “--ar” or “--no”.

• Parameters should be placed at the end of the prompt, without double colons “::” and separated by spaces “ “ only.

Example:
a vibrant underwater scene with colorful coral and fish:: digital artwork --s 500 --no sharks, divers

Summary

By experimenting with various techniques, formats, artistic terms, and stylization levels, users can optimize their prompts and generate engaging visual content. While Midjourney V5 allows for the use of natural language, it cannot interpret things like smell and sounds other than through the data it has been trained on. All methods explained in this guide can be combined if needed, unless stated otherwise. Revisit the guide often to keep up to date. Always remember to adhere to the guidelines and avoid banned words to ensure a smooth and productive experience with the Midjourney v5 Bot.

Disclaimer

This guide is an unofficial resource created to assist users in generating effective image prompts for the Midjourney Bot V5. The author of this guide has no affiliation with Midjourney, Discord, or their respective parent companies and subsidiaries. The content of this guide is not endorsed by Midjourney or Discord, and the views and opinions expressed herein are solely those of the author. Users should refer to the official documentation provided by Midjourney and Discord for the most accurate and up-to-date information.

The use of any trademarks, trade names, or logos belonging to Midjourney, Discord, or other third parties on this website is for identification purposes only and does not imply endorsement or affiliation.

Contact: The Creator