How To Write Effective Midjourney Prompts

Midjourney is a popular generative AI art tool that allows users to generate images from simple textual descriptions. By using Midjourney, you can draw inspiration from generated images to boost your creativity, quickly try diverse artistic styles, and explore imaginative concepts.

However, you’ll need to input clear and descriptive prompts to allow Midjourney to produce images that align with your goals. Think of prompts as commands that guide AI art tools to produce coherent outputs.

Mastering prompt writing can help you harness the power of artificial intelligence in your workflow. Vague prompts can allow Midjourney to generate images that may not suit your needs. At the same time, Midjourney can misinterpret or miss crucial points in overly complex prompts with too many intricate details. Good Midjourney prompts should have a balance between simplicity and complexity.

Keep reading to discover how to create effective prompts to produce more relevant and accurate outputs in Midjourney.‍

What is Midjourney?

Midjourney describes itself as an independent research lab focused on expanding the imaginative powers of the human species.

Like other AI tools, Midjourney is powered by deep learning models, machine learning algorithms, and natural language processing (NLP) technology. NLP enables Midjourney to process conversational inputs and determine the user’s intent, while deep learning and machine learning models facilitate the actual image-generation process.

Midjourney has been trained on large datasets to enable it to generate images. As a result, it can imitate works of real artists to produce original images in different styles, including minimalist, photorealistic, surrealist, abstract, and more.

You must write and submit a prompt to get started with image generation in Midjourney. Prompts are instructions you enter in generative AI tools to guide them in producing desirable outputs. In this case, Midjourney prompts allow it to create AI images.

While there’s no specific format for structuring different prompts, an effective text prompt should be simple, clear, and descriptive. As we will demonstrate, you can include image URLs and different parameters in your Midjourney prompts to facilitate more tailored AI-generated art.

Midjourney prompt best practices

From knowing your objectives to modifying your prompt styles, here are the best practices for writing prompts.

Understand your objective

Midjourney was trained on vast datasets and can help you explore different creative fields and styles, from watercolor paintings to modern digital art. But to leverage Midjourney’s capabilities effectively, you must have a clear objective that outlines what you hope to achieve.

So, before you create a prompt, the first step is to define the purpose of your art. Determine if the art should evoke emotion, tell a story, or explore a visual concept. For example, if you’re creating characters for a children’s book, the purpose of your art might be to convey a story.

A clear objective will guide your prompt creation process, leading to clear commands. With a good objective, you can ensure that AI tools like Midjourney generate more relevant and accurate outputs. You can also quickly tell when generated images are too broad and don’t align with your purpose and ideal final image.

Configure your settings

While you can produce stunning images in Midjourney out-of-the-box, adjusting the settings can allow you to specify your desired default outputs.

Using the /setting command in Midjourney, you can adjust different values, including public and stealth modes, fast, relax, turbo modes, and the sticky style. You can also enable remix mode to modify more values, including parameters, aspect ratios, and prompt structure. For example, remix mode must be enabled for you to edit specific areas on your images.

Incorporate specific art styles

Midjourney is a robust AI tool capable of generating images in different styles. When you give Midjourney creative freedom via vague or extremely simple prompts, you may end up with images that don’t align with your purpose.

To increase the likelihood of getting the stylized output you want, you can specify your preferred art styles within text-to-image prompts. Some examples of artistic styles to add to prompts include hyper-realistic, cyberpunk, anime, mosaic, oil painting, and impressionism. You should also consider providing the color palette for your design, mentioning the specific colors you would like in the image.

To illustrate, a prompt like “short girl, wide eyes, red hair, green blouse, blue pants, anime style” will help generate an anime character. Here’s what Midjourney produced.

We get the following image when we use the same prompt but change the image style to hyper-realistic.

Prompt: Short girl, wide eyes, red hair, green blouse, blue pants, hyper-realistic style

Midjourney also provides the style parameter you can use to generate images. Supported image styles include raw, original, expressive, and cute. To use this parameter, simply write a prompt and then add --style <INSERT STYLE> at the end. Like this: “short girl, wide eyes, red hair, green blouse, blue pants --style raw.” Here’s Midjourney’s output.

Balance detail with open-endedness

While you should be clear and descriptive in your prompts, you should also leave room for AI interpretation. Strike the right balance using synonyms, advanced prompts, and multi-prompts to add depth without over-constraining the creative process.

For advanced prompts, use image URLs alongside your prompts to give Midjourney more detail and enable it to produce more relevant outputs. You can upload any image to a Midjourney Discord channel and then simply right-click on it and copy the image address, as illustrated below.

You can also include parameters like style, models, and upscalers to allow Midjourney to generate more enhanced images.

As Midjourney puts it, “Word choice matters when drafting prompts.” Use synonyms and words that clearly describe what you’re trying to generate in Midjourney. For instance, rather than saying “small,” you can go with “tiny.”

You should also be specific in your prompts. For example, if you’re looking to generate an image with a few objects, including the actual number—like three people—can lead to better results.

If you don’t want a specific object to appear in your artwork, use the --no parameter. For instance, if you’re generating an image of soccer players on a field, you can remove the ball by adding --no soccer ball to your prompt.

Midjourney multi-prompts can also allow you to control how different objects are generated in your images. You use double colon (::) to specify multi-prompts. For example, if you want Midjourney to interpret starfish as separate words, You will write it as star:: fish.

Here’s the Midjourney output for starfish.

And here is an interpretation of star:: fish—which is a person surrounded by fish.

Tip: The Midjourney bot can process and interpret your prompt more accurately when you use simple and short sentences. Drafting long sentences can overwhelm the AI system, leading to less ideal results.

Specify output resolution

Aspect ratios can influence an image’s balance, emphasis, and visual appeal. In digital media, aspect ratio affects how an image is displayed on different screen configurations. For instance, mobile devices may better display images with smaller aspect ratios than larger ones. An image with an aspect ratio of 16:9 may fit well as a social media cover photo.

Midjourney allows you to specify the aspect ratio and resolution the generated image should have. You can add the --ar <ASPECT RATIO> at the end of your prompt. Supported aspect ratios in Midjourney include 1:1 (default in Midjourney v5), 5:4, 3:2, 7:4, and 16:9.

Here are examples of Midjourney-generated images with unique aspect ratios.

Prompt: A short girl, white hair, purple dress, laughing, Pixar style --ar 16:9

Prompt: A muscular hero, standing on a rooftop, looking down the city, cape blowing in the wind, anime style --ar 3:2

Use photography terms

Since Midjourney is tailored for image generation tasks, adding different photography terms in your prompts can help achieve specific image styles. Examples of photography terms include cinematic lighting, golden hour, and full-body shot.

You can also add camera angles like wide-angle, ultra-wide, eye-level, and medium shot to allow you to determine how generated images look from different points of view.

Additionally, incorporating camera lenses in your prompt allows Midjourney to try to match the quality of generated images to those on real-world professional cameras. Some camera lenses you can include in your prompt include Canon EF 50mm f/1.8 STM, Sigma 18-35mm F1.8 Art DC HSM, and Nikon AF-S DX NIKKOR 35mm f/1.8G.

To illustrate, we used photography terms and camera lenses in the “a family of five at a dinner in a hotel, food and drinks on the table, cinematic lighting, Canon EF 50mm f/1.8 STM --ar 16:9” prompt. Here’s the generated output.

Leverage inspiration and references

If you have any existing images or artwork you want to recreate, you can upload them to Midjourney for reference. These image URLs will then serve as nice starting points for your prompts.

To upload images, navigate to a Midjourney channel on Discord. Press the plus (+) icon adjacent to the input field.

You should see an option to upload a file, as shown below.

Simply click the Upload a File button to upload the images to Discord. Once an image is uploaded, you can right-click on it to copy its URL address. In our case, we uploaded the following image.

You can proceed to include the image URL in your prompt as follows.

https://media.discordapp.net/attachments/1144149411790852106/1197111190732668988/skinycharacter.webp?ex=65ba13ad&is=65a79ead&hm=3e54358bc7591179c4882de5d7816cb8c422a9c0f8112e3333b7383f529d3af9&=&format=webp&width=369&height=369 short girl, red hair, blue pants, light green shirt, Pixar style --ar 16:9

The above prompt will allow Midjourney to generate a character that looks similar to the reference image.

Before uploading images to Midjourney, you should consider copyrights. If you’re uploading original images that you own then, according to Midjourney’s terms of service:

You grant to Midjourney, its successors, and assigns a perpetual, worldwide, non-exclusive, sublicensable no-charge, royalty-free, irrevocable copyright license to reproduce, prepare derivative works of, publicly display, publicly perform, sublicense, and distribute text and image prompts You put into the Services….

In essence, you are giving Midjourney the right to use and re-use your image and derivative works of your image.

If you are uploading works to Midjourney hoping to recreate images that you do not own the rights to, then you should consider your use case and the AI’s output. If the output will only be used personally and not made public, that might be covered by fair use. But keep in mind that the owner of the original image has to permit a new version of the work to be created—no matter how much of the original is changed by AI.

If your intention with images created by Midjourney is to profit—for example, by using it to create a children’s book that you hope to sell—you may want to consult with an attorney about copyrights.

Continuously learn and adapt

As you work with Midjourney, remember to adapt your prompt style based on what works and what doesn’t. Tweaking your prompts can help produce images that closely align with your objectives. Additionally, staying updated with Midjourney version updates, AI advancements, and tutorial resources can also enhance your prompt engineering skills.

How to format your prompt

Though there’s no specific way to structure a prompt, a good Midjourney prompt should adhere to the following guidelines.

• Have keywords/tokens
• Feature specific synonyms that describe the character, scenes, or objects
• Use short phrases and sentences

While drafting Midjourney, you don’t have to pay much attention to grammar and capitalization. However, you should watch out for spelling mistakes since they could cause Midjourney to misinterpret or ignore certain portions of your prompt. Also, avoid including too much detail in your prompt since it may overwhelm the AI system.

Here is an example of a well-formatted Midjourney prompt:

A young boy, wide eyes, red hair, laughing, green trousers, black shoes, white t-shirt, anime style --ar 16:9

Midjourney prompt examples

During prompt engineering, you should be detailed enough to allow Midjourney to produce images that match your needs. For instance, a prompt like “a cat sitting on a chair” is too basic and lacks enough information. In this case, a good prompt could be “a fluffy, long-haired white Persian cat sitting elegantly on a vintage Victorian armchair.”

Here are other Midjourney prompt examples you can draw inspiration from.

Prompt: A view of a futuristic utopia with happy citizens, referencing “The Jetsons” style architecture, during golden hour, wide angle --ar 16:9

Prompt: Gothic cathedral interior, detailed shadows, and light rays penetrating stained glass windows in a hyper-realistic style

Prompt: Park scene depicting all four seasons in quadrants, from a fixed viewpoint, showing the location’s transformation over time

Prompt: Portrait of a woman in the style of a 1920s photograph, sepia tones, with soft lighting and a thoughtful expression

Prompt: A high fantasy book cover, featuring a dragon and a knight, with room for the title at the top

Prompt: Four birds on a power line, a flower garden below, vibrant colors --ar 16:9

Prompt:  An illustration for a new bag for college students, close-up, soft material, blue straps

Prompt: A cyclist going down a mountain, green grass, white helmet, white t-shirt, wide angle, hyper-realistic --no road

Find Midjourney work on Upwork

Midjourney is a robust AI image generator that can bring numerous benefits to your workflow. However, you’ll need to perfect the art of prompt engineering to get the most out of Midjourney.

When creating prompts, ensure a balance between simplicity and complexity. Use clear words, phrases, and synonyms separated by commas to describe scenes and characters—which AI systems can then interpret accurately. Using photography terms like cinematic lighting, bonker effect, and others can also add more quality to generated images.

Drafting Midjourney prompts that return relevant results may take time, especially for beginners. Consider working with Midjourney AI artists on Upwork to help you create good prompts to generate high-quality outputs.

And if you’re an expert looking for work, Upwork can connect you to different Midjourney jobs to help build your prompt engineering portfolio while earning extra income.

‍

Upwork is not affiliated with and does not sponsor or endorse any of the tools or services discussed in this article. These tools and services are provided only as potential options, and each reader and company should take the time needed to adequately analyze and determine the tools or services that would best fit their specific needs and situation.

This article is intended for educational purposes and should not be viewed as legal or tax advice. Please consult a professional to find the solution that best fits your situation.