How to Write the Most Accurate Midjourney Prompts - /describe Beginner Tutorial

This tutorial video teaches viewers how to write accurate prompts for the Midjourney AI image generation tool. It explains the use of the 'describe' feature on Discord to get AI-generated descriptions of an image and how to refine the prompts by adding a reference picture, adjusting weights, and modifying stylization and chaos values for more creative or literal results. The video also mentions the limitations of the 'describe' feature, such as its inability to recreate text from images. The creator offers a prompt pack for purchase to save time and provide examples of effective prompts, and the video concludes with a reminder of the tool's imperfections and a call to like the video for wider sharing.

Takeaways

• 🖼️ To write an accurate prompt for an image, use the 'describe' feature in Discord by typing '/describe' and uploading the image.
• 🔍 Mid-Journey will provide four different options based on the image, which can be selected and submitted for results.
• 📎 If unsatisfied with the initial results, add the original reference image as an image prompt to improve the accuracy of the generated prompts.
• 🔗 Copy the image address and paste it into the describe prompts to use the reference picture as a foundational picture for the prompts.
• ⚖️ Adjust the importance of the reference picture by adding a weight (IW) with a number between 0.5 and 2 to the prompt.
• 🔄 Experiment with different weights to see how it affects the alignment of the generated images with the reference picture.
• 🎨 Adjust the 'stylize' value to make the generated images more literal or creative, and the 'chaos' value for more variety in the grid.
• 🔄 Use the 're-roll' feature on the describe prompt to get additional variations of the generated images.
• 🌐 Be aware that the describe feature uses a model that may not be native to Mid-Journey, which means it may not perfectly recreate the image even if it describes it well.
• 📐 Understand that aspect ratios may not be exact due to Mid-Journey's rounding to the nearest 32-pixel value.
• 💡 The describe feature can interpret made-up names, and sometimes these are included in the prompt without a hyperlink.
• 📦 The creator offers a prompt pack with favorite prompts and examples for a fee, as a way to support the channel without sponsors.

Q & A

What is the main purpose of using the 'describe' feature in Discord with Midjourney?

The 'describe' feature in Discord is used to help users generate prompts for images by asking Midjourney what it sees in a saved image, which assists in creating more accurate prompts.

How does the 'describe' feature provide options for an image prompt?

After using the 'describe' feature and uploading an image, Midjourney offers four different options or interpretations of the image, which the user can select and submit.

If the initial results from the 'describe' feature are not satisfactory, what can be done to improve them?

Users can add the original reference picture as an image prompt, adjust the weight of the reference picture, modify the stylize or chaos value, and use the re-roll feature for additional prompts.

How does adding the original reference picture as an image prompt affect the generated prompts?

Adding the original reference picture provides a foundational picture for the prompts, making the generated images closer to the original image.

What is the significance of adjusting the weight of the reference picture in the prompt?

Adjusting the weight of the reference picture (using --IW followed by a number between 0.5 and 2) determines the importance of the reference picture relative to the other words in the prompt, influencing how closely the generated images resemble the reference.

What is the role of the stylize and chaos values in the prompt generation process?

The stylize value, when lowered, makes Midjourney follow the prompt more closely, while raising it allows for more creative freedom. The chaos value, when increased, introduces more variety in the generated grid of images.

What does it mean if the 'describe' feature mentions an artist without a hyperlink?

If an artist is mentioned without a hyperlink, it suggests that the name does not exist and might be a made-up name that Midjourney has interpreted.

How does Midjourney handle aspect ratios in the 'describe' feature?

Midjourney outputs aspect ratios rounded to the nearest 32-pixel value, so exact ratios like 16:9 may not be precisely achieved.

Why might Midjourney's 'describe' feature be able to describe an image accurately but fail to recreate it?

The 'describe' feature uses a model that is not inherently native to Midjourney's generation capabilities. Thus, even if it can describe an image perfectly, it may not accurately recreate complex elements like text within an image.

What is the purpose of the prompt pack created by the video creator?

The prompt pack is a collection of the creator's favorite prompts with example images, designed to save users time and provide a clear guide on how to make amazing images with Midjourney.

How can users get more prompts that are similar but not the same to the ones generated by the 'describe' feature?

Users can utilize the re-roll feature within the 'describe' function to obtain four more prompts, which will be similar but offer slight variations.

What is the recommended approach if the generated images are not as literal as intended due to a complex and specific prompt?

If the images are not as literal as intended, the user can lower the stylize value to make the output more literal, or they can provide clearer and more specific instructions in their prompt to guide Midjourney more effectively.

Outlines

00:00

🖼️ Image Prompt Creation with Discord's Describe Feature

The first paragraph explains how to create prompts for images using the 'describe' feature in Discord. It details the process of uploading an image and receiving four different prompt options. The user can select one of these prompts and submit it for a more tailored result. If the initial results don't match the original picture closely enough, the user can add the original image as a reference, which can then be weighted to increase its influence on the final output. The paragraph also discusses adjusting the 'stylize' and 'chaos' values to refine the generated images further.

05:01

🎨 Advanced Techniques for Image Generation with Mid-Journey

The second paragraph delves into more advanced techniques for refining image generation with Mid-Journey. It covers the use of the 're-roll' feature on the describe prompt to get additional variations, the interpretation of made-up artist names by the AI, and the nuances of aspect ratios in image outputs. The paragraph also touches on the limitations of the describe feature, particularly its inability to recreate text from images accurately. Finally, it mentions a prompt pack created by the speaker for a fee, offering 51 favorite prompts with 69 examples to save time and provide a guide for creating impressive images.

Keywords

Midjourney

Midjourney is an AI-based tool that assists in the creation of images from textual prompts. In the context of the video, it is the primary subject being discussed, with the focus on how to utilize its 'describe' feature to accurately interpret and generate images based on a given prompt or reference picture.

Describe Feature

The 'describe' feature is a function within Midjourney that analyzes images and provides a textual description, which can then be used to generate prompts for creating new images. It is crucial for understanding how to write accurate prompts and is the main tool demonstrated in the tutorial.

Prompts

In the context of the video, prompts are the textual descriptions or commands that guide AI tools like Midjourney in generating specific images. They are the essential input that the AI uses to create the desired visual output.

Image Upload

Image upload refers to the process of transferring a picture from the user's device to the Midjourney platform to be analyzed and used as a reference for generating prompts. This is a necessary step to ensure that the AI can accurately understand and respond to the user's request.

Reference Picture

A reference picture is the original image that a user wants the AI to base its generated images on. By including the reference picture in the process, the user can guide the AI to produce outputs that closely resemble the desired theme or subject.

Weight

In the context of Midjourney, weight is a numerical value assigned to a reference picture to indicate its level of importance in the image generation process. A higher weight means the AI will prioritize the reference picture more, resulting in outputs that closely follow the original image's theme or style.

Stylize and Chaos Values

Stylize and chaos values are parameters within Midjourney that control the level of adherence to the prompt and the degree of creative freedom the AI has. A lower stylize value means a more literal interpretation, while a higher value allows for more creative latitude. Chaos value introduces variability in the generated images, with higher values resulting in more diverse outputs.

Re-roll

Re-roll is an option in Midjourney that allows users to generate additional prompts based on the same image or prompt. This feature provides more options and variations for the user to choose from, helping to refine the desired outcome.

Aspect Ratios

Aspect ratios refer to the proportional relationship between the width and height of an image. In the context of the video, aspect ratios may not always produce the exact dimensions expected due to Midjourney's output rounding to the nearest 32-pixel value.

Describe Model

The describe model is the AI component within Midjourney that interprets and describes images. It is not inherently part of Midjourney's main functionality but is used to analyze images and produce descriptions that can guide the image generation process.

Prompt Pack

A prompt pack is a collection of pre-made prompts and example images created by experienced users of Midjourney. It is designed to help newcomers or other users save time and learn from the examples provided by those who have extensive experience with the AI tool.

Highlights

Using the 'describe' feature in Discord, you can ask Midjourney what it sees by uploading an image and receiving different prompt options.

By clicking on the generated prompts and hitting submit, you can see how well they match the original image.

If unsatisfied with the results, you can add the original reference picture as an image prompt to improve accuracy.

You can copy the image address and paste it into the describe prompts to use it as a foundational picture for your prompts.

Adding a weight to the reference picture with '--IW' followed by a number can increase its importance in the prompt.

Adjusting the stylize or chaos value can influence how closely Midjourney follows your prompt or allows creative freedom.

Lowering the stylized value to zero makes Midjourney follow the prompt more closely, while raising it enhances the visual appeal.

Adding chaos value increases the variety in the generated grid, with 100 resulting in completely different images.

The 'describe' feature uses a model to interpret images, which may not be native to Midjourney, and thus may not perfectly recreate the described image.

The 'describe' feature can sometimes mention artists and attach hyperlinks, or mention names without hyperlinks that do not exist.

Midjourney outputs images rounded to the nearest 32-pixel value, which may result in slight variations from the exact aspect ratio you input.

The 're-roll' feature on the describe function can provide additional prompts, offering more options to refine a particular style.

The video offers a prompt pack available for purchase, containing 51 favorite prompts with 69 examples to save time and demonstrate image creation.

The author emphasizes the importance of understanding the limitations of the 'describe' feature and its impact on the generation process.

The video provides a comprehensive guide on how to get the most out of Midjourney's 'describe' feature for accurate image generation.

The author acknowledges the community contribution by thanking Squire Zed on Discord for their insights on the describe feature.

The video concludes with an invitation to like the video to share it with more people and a farewell message from the author.