How to Use Negative Prompts in Midjourney

Tokenized AI by Christian Heidorn
18 Dec 202222:07

TLDRChristian Heidorn from Tokenized AI presents a tutorial on negative prompting in Midjourney, a technique used to remove objects or concepts from generated images. He explains that while writing prompts is additive, negative prompting is subtractive, allowing for the removal of elements from a scene. However, he notes that Midjourney lacks specific editing tools like in-painting, which are available in other tools such as Dali 2 or Stable Diffusion. Heidorn demonstrates the use of the 'no' parameter and negative image text weights to adjust the generated images by removing certain elements like candles from a birthday cake or flowers from a meadow. He emphasizes that negative prompting is less effective with complex scenes and becomes more abstract with each added layer of complexity. The tutorial serves as a guide on how to use negative prompts to refine and control the content of AI-generated images, though it is not without its limitations and requires a process of trial and error.

Takeaways

  • 📌 Negative prompting is a technique used to remove objects or concepts from generated images in Midjourney.
  • ➕ Positive prompts add content to the image, while negative prompts subtract elements from the scene.
  • 🚫 Midjourney does not support in-painting or direct image editing like some other tools, such as Dali 2 or stable diffusion.
  • 🔍 The 'no parameter' is used to specify what should be removed from the image, but it's not always effective with complex scenes.
  • 📈 Text weights can also be used for negative prompting, where a negative value reduces the prominence of a concept in the generated image.
  • 🔄 Remix mode allows you to work off an existing image and make specific changes, like removing elements.
  • 🕵️‍♂️ Negative prompting works best when trying to remove non-core parts of the image concept, not the central theme.
  • 🌐 As complexity increases with multiple negative prompts, the generated images tend to become more abstract and less realistic.
  • 🔄 Using both positive and negative text weights can help construct intricate scenes with specific elements emphasized or removed.
  • 🎨 Negative prompting is not a perfect science and requires trial and error to achieve the desired outcome.
  • ⏱️ While Midjourney currently does not have an in-painting feature, it's possible that such a feature may be added in the future.

Q & A

  • What is the main topic of the tutorial?

    -The main topic of the tutorial is how to use negative prompts in Midjourney to remove objects and concepts from generated images.

  • What is negative prompting?

    -Negative prompting is the process of removing something from your prompt, allowing you to eliminate elements from a specific scene or image that you've created or described.

  • How does negative prompting differ from the regular prompt writing process?

    -While regular prompt writing is additive, adding content and transformation to your prompt, negative prompting is subtractive, allowing you to remove elements from the generated image.

  • Why can't Midjourney edit images like some other tools?

    -Midjourney cannot edit images due to its chosen interface, Discord, which is not a web application and would make implementing such features complex and cumbersome.

  • What is the 'no parameter' used for in negative prompting?

    -The 'no parameter' is used in negative prompting to specify the object, thing, or concept that the user wants to remove from the generated image.

  • How does the effectiveness of negative prompting change with the complexity of the scene?

    -The effectiveness of negative prompting decreases as the complexity of the scene increases. It is not meant to be used in very complex situations where super specific details of the image need to be edited.

  • What is the alternative method to the 'no parameter' for negative prompting?

    -The alternative method to the 'no parameter' is using negative image text weights, which allows for more control over the removal of concepts from the generated image.

  • Why might negative prompting not work as expected with certain concepts?

    -Negative prompting might not work as expected when trying to remove a core part of the concept of the image, as the image generation system may struggle to interpret the prompt without that fundamental element.

  • How can text weights be used to achieve a similar result as the 'no parameter'?

    -Text weights can be used to assign a negative value to a concept, effectively telling Midjourney to remove it from the generated image, similar to using the 'no parameter'.

  • What is the significance of using both positive and negative text weights?

    -Using both positive and negative text weights allows for the construction of entire scenes with more control over which elements are included or excluded, leading to more refined and intentional image generation.

  • Why does the image become more abstract with each added layer of complexity in negative prompting?

    -As more complexity is added through negative prompting, the image generation system may struggle to maintain a natural and realistic composition, resulting in more abstract and varied outcomes.

  • What is the advice given for using negative prompting effectively?

    -The advice is to use negative prompting on scenes or images where you're trying to remove something that is not a core part of the concept, and to be prepared for a process of trial and error to achieve the desired result.

Outlines

00:00

🎨 Introduction to Negative Prompting

Christian Heidorn from Tokenized AI introduces the concept of negative prompting in mid-journey, a method to remove objects or concepts from generated images. He clarifies that unlike in-painting features in other tools, mid-journey does not allow for specific image editing but uses negative prompts to alter the content of the generated scenes. The video promises to explore how to use both positive and negative text weights to construct scenes.

05:00

🚫 Using the 'No' Parameter for Negative Prompting

The video demonstrates the use of the 'no' parameter in negative prompting with an example of removing candles from a birthday cake image. It explains that the 'no' parameter is a subtractive process and shows how to use it in conjunction with remix mode. However, it also highlights the limitations, such as the inability to remove core concepts from a scene due to the way mid-journey interprets prompts.

10:01

🌼 Removing Non-Core Concepts with Abstract Prompts

Christian shows how to use negative prompting with more abstract prompts, like 'Nature's Meadow,' to remove elements like flowers and birds while keeping the composition similar. He explains that as complexity increases, the results become more abstract and less realistic. The video illustrates how negative prompting can be effective for non-core elements and how it changes the overall scene.

15:02

🔍 Exploring Text Weights for Negative Prompting

The video explains the alternative method of negative prompting using text weights, which is equivalent to using the 'no' parameter but allows for more control. It provides an example of creating a winter landscape with a buffalo herd, without trees or snow, by assigning different weights to various elements of the scene. This method is shown to be effective for crafting detailed and specific images.

20:02

🔧 Negative Prompting as a Trial and Error Process

Christian concludes by emphasizing that negative prompting is not a perfect science and requires experimentation. He advises that for simple removals, using the 'no' parameter might be more efficient, but for more complex scenes, text weights offer greater control. He acknowledges that negative prompting in mid-journey is different from in-painting in other tools and encourages viewers to play around with it to achieve desired results.

Mindmap

Keywords

💡Negative Prompting

Negative prompting refers to the technique of specifying what should not be included in the generated image by an AI image-generating tool. In the context of the video, it is used to remove objects or concepts from the AI-generated images. For example, if an image of a children's room already contains a teddy bear that the user does not want, they can apply negative prompting to remove it.

💡Midjourney

Midjourney is the name of the AI tool discussed in the video that generates images based on textual prompts. It is used to illustrate the process of negative prompting. The video explains that while Midjourney is powerful, it has limitations such as the inability to perform in-painting or detailed image editing.

💡No Parameter

The 'no' parameter is a specific syntax used within prompts to indicate elements that should be excluded from the generated image. It is a form of negative prompting. For instance, if a user enters 'no candles' in their prompt, Midjourney will attempt to generate an image without candles.

💡Remix Mode

Remix mode is a feature within Midjourney that allows users to modify existing images rather than starting from scratch. It is activated to work off an existing image and make minor changes, such as removing certain elements, using negative prompting.

💡Text Weights

Text weights are numerical values assigned to words or concepts in a prompt to control their importance or emphasis in the generated image. A higher weight increases the prominence of that element, while a negative weight, such as -0.5, indicates that the element should be downplayed or removed, which is a form of negative prompting.

💡In-Painting

In-painting is a feature of some AI image editing tools that allows for the modification of existing images, such as removing or adding elements with a more precise control than negative prompting. The video mentions that Midjourney does not currently support in-painting, which is a limitation when compared to other tools like Dali 2 or Stable Diffusion.

💡Composition

In the context of the video, composition refers to the arrangement of elements within an image. When using remix mode in Midjourney, the composition of the original image is maintained, but certain elements can be removed or altered through negative prompting.

💡Concepts

Concepts in the video refer to the ideas or themes that are central to the image generated by Midjourney. Negative prompting can be used to remove certain concepts from these images. However, the effectiveness of negative prompting depends on whether the concept is core to the overall theme or a removable element.

💡Abstract Prompts

Abstract prompts are less specific and more open to interpretation than concrete prompts. They can result in a wider variety of images and are used in the video to demonstrate how negative prompting can be applied to generate creative and artistic results.

💡Scene Construction

Scene construction is the process of creating a detailed image by combining various elements and concepts. The video shows how both positive and negative text weights can be used to construct scenes in Midjourney, emphasizing certain elements while de-emphasizing or removing others.

💡AI Software

AI software, as discussed in the video, refers to artificial intelligence applications like Midjourney that can perform complex tasks such as image generation based on textual descriptions. The video tutorial focuses on how to use negative prompting within the context of AI software to refine and direct the creative output.

Highlights

This tutorial demonstrates how to use negative prompts to remove objects and concepts from images in mid-journey.

Negative prompting is a subtractive process that allows you to remove elements from a specific scene or image.

Midjourney, unlike some other tools, does not support image editing features like in-painting.

Negative prompting in mid-journey involves removing concepts from the message sent to the AI, rather than specific image editing.

The effectiveness of negative prompting decreases with more complex scenes.

There are two methods for negative prompting in mid-journey: using the 'no' parameter and negative image text weights.

The 'no' parameter is used to specify objects or concepts to be removed from the generated image.

Using the 'no' parameter can lead to replacement objects if the composition is too specific.

Negative prompting does not work well with very clearly defined concepts that are core to the image.

An abstract prompt can be more effective for negative prompting, as it allows for broader interpretation.

Negative prompting can be used to create more abstract and artistic images by removing certain elements.

Adding complexity to a prompt with negative prompting can result in more abstract and less realistic images.

The 'no' parameter is equivalent to setting a text weight of -0.5 for the concept to be removed.

Text weights can be used to control the presence and emphasis of different elements in an image.

Crafting scenes with text weights and negative prompting can lead to the creation of unique and compelling images.

Negative prompting is not a perfect science and requires trial and error to achieve the desired result.

The possibility of mid-journey implementing an in-painting feature in the future is not ruled out.