Stable Diffusion 올바로 사용하기 #2 - 프롬프트에 강조 주기 (원하는 이미지로 쪼옥~ 뽑아서 만들기)

DigiClau (디지클로) Lab
29 Mar 202305:28

TLDRThe video script introduces viewers to the use of Stable Diffusion, an AI image generation model. It explains how users can input text prompts to create images, and how the model sometimes makes its own interpretations of the prompts. The video demonstrates the process of using prompts with examples, highlighting features like dresses, accessories, and settings. It also covers the use of prompt weights, such as parentheses and numerical values, to emphasize certain aspects of the image. The video concludes by encouraging viewers to use these techniques to create more targeted and precise images.


  • 📝 The video discusses the use of Stable Diffusion for image generation based on text prompts.
  • 🖌️ Stable Diffusion allows for the creation of images by interpreting text prompts and adding its own artistic touch.
  • 🎨 The AI sometimes ignores certain words in the prompt or follows its own interpretation to complete the image.
  • 👩 The example given describes a woman with specific attributes and clothing, captured in a cityscape.
  • 👗 The prompt mentions a black mini skirt, but some generated images may not include it, showing variations like other colored skirts.
  • 🐰 The example also notes the presence of a rabbit headband, but some images may lack it or have different headwear.
  • 👜 The video introduces the concept of emphasizing specific words in the prompt by using parentheses or 'prompt weights'.
  • 🔢 Prompt weights can be assigned numerical values to increase the emphasis on certain elements, with values above 1 intensifying the effect.
  • 📈 The term 'prompt weight' is also referred to as 'prompt waits', which can be found in prompts available online.
  • 🛠️ Errors in the prompt, such as mismatched parentheses, can be highlighted with a red border for easy identification and correction.
  • 🎯 By effectively using prompt weights, users can target and create more precise images according to their preferences.
  • 📌 The video encourages viewers to subscribe and set alarms for updates, highlighting the helpful nature of the content provided.

Q & A

  • What is the speaker's assumption about the audience at the beginning of the video?

    -The speaker assumes that the audience is already familiar with using Stable Diffusion or is new to it and learning the basic usage from a video on the right and above.

  • How does the Stable Diffusion AI create images based on the provided text?

    -Stable Diffusion creates images based on the text provided to it, but it also uses its own judgment to complete the drawing according to its preferences, even if not all the required words from the prompt are included.

  • What is an example of a prompt the speaker uses in the video?

    -The speaker uses a prompt describing a woman in a black mini skirt, wearing a dress, with a choker, beautiful eyes, and earrings, standing on the streets of a city like Bern, holding a hot beverage, and wearing a rabbit headband, giving off a hot actress vibe.

  • How does Stable Diffusion handle prompts with specific words that are not followed?

    -Stable Diffusion may ignore certain words in the prompt that are not followed, even if they are specifically added by the user.

  • What is a method to emphasize a particular word in the prompt?

    -One method to emphasize a word in the prompt is by using parentheses. For example, if the user wants to emphasize a handbag, they can write 'handbag' in parentheses.

  • How can the emphasis on a word be increased using parentheses?

    -By using multiple layers of parentheses around a word, the emphasis on that word is increased. This technique is also known as prompt weighting.

  • What is an alternative way to apply prompt weight?

    -An alternative way to apply prompt weight is by adding a colon followed by a numerical value after the word. The number represents the weight, with higher values applying more emphasis, up to a maximum value greater than 1.

  • What happens when the prompt weight is set to a value higher than 1?

    -When the prompt weight is set to a value higher than 1, it means that the emphasis is increased by a percentage greater than 100% of the base weight.

  • How can users correct mistakes in the prompt weight application?

    -Users can correct mistakes in prompt weight application by adjusting the parentheses and ensuring the correct weight is applied. A red border in the text indicating the mistake can be used as a guide to correct the issue.

  • What is the benefit of using prompt weights effectively?

    -Effective use of prompt weights allows users to target and create images that are more precise and aligned with their desired outcomes, making the process easier and more accurate.

  • What does the speaker suggest at the end of the video for viewers who found it helpful?

    -The speaker suggests that viewers who found the video helpful should subscribe and set an alarm for notifications to receive updates on future content.



🖌️ Introduction to Stable Diffusion and Prompt Usage

This paragraph introduces the video's focus on the use of Stable Diffusion, an AI image generation model. The speaker, Chloe, assumes that the viewers are already familiar with the tool but also offers guidance for beginners on how to learn the basics from a video on the right. The paragraph explains how Stable Diffusion takes a text prompt and generates images based on it, filling in any missing words with its own interpretations to complete the visual. It also touches on how the AI can ignore certain specified words if they don't align with its generated concept. The speaker then demonstrates the use of a prompt with a detailed description of a woman in a black dress, holding a bag, and wearing a bunny headband, set on the streets of Vancouver. The video showcases multiple images generated from the prompt, highlighting how some images follow the prompt closely while others incorporate the AI's unique interpretations, such as different skirt colors and the absence of a headband. The paragraph concludes by discussing the ability to emphasize specific words in the prompt by using parentheses, a technique known as 'prompt weighting,' to guide the AI more accurately in generating the desired image.


🎨 Utilizing Prompt Weights for Image Generation

This paragraph delves deeper into the concept of prompt weighting, explaining how it can be used to influence the AI's image generation process. The speaker illustrates the use of parentheses to emphasize certain words in the prompt, which the AI then prioritizes more highly. The paragraph further explains the concept of 'prompt waits,' where additional weight is given to specific words by enclosing them in double parentheses, tripling their influence on the image generation. The speaker also introduces another method of prompt weighting by assigning numerical values to words using colons, with higher values leading to greater emphasis on those features. The speaker then demonstrates how adjusting the weights can lead to more precise image generation, using the example of a woman with an emphasized black mini skirt and a handbag, requesting the AI to generate an image with these features in focus. The video also shows how to correct mistakes in the prompt by adjusting brackets, which are indicated by a red border. The paragraph concludes by encouraging viewers to use prompt weighting effectively to create images that closely match their desired outcomes and to select the best images from multiple generations. The speaker also invites viewers to subscribe and set alarms for notifications if they found the video helpful.



💡Stable Diffusion

Stable Diffusion is an AI-based image generation model that creates images based on textual prompts provided by users. In the context of the video, it is the primary tool being discussed, and the video aims to educate viewers on how to effectively use it to generate desired images. The model is noted for its ability to interpret and sometimes creatively deviate from the exact instructions given in the prompts.


Text-to-Image refers to the process of converting textual descriptions into visual images using AI technology. In the video, this concept is central as it explains how Stable Diffusion takes textual prompts and translates them into corresponding images, allowing users to generate custom visual content by describing it in text form.


A prompt, in the context of this video, is a textual description or a set of instructions that guides the AI in creating an image. Prompts are crucial as they determine the elements and style of the generated images. The video discusses the importance of crafting effective prompts to achieve the desired output from Stable Diffusion.


Customization in the video refers to the ability of users to tailor the output of Stable Diffusion according to their preferences by adjusting the prompts and using various techniques to emphasize or de-emphasize certain aspects of the image. This allows for a more personalized and targeted image generation process.


Weighting in the context of the video is the process of assigning importance to specific aspects of a prompt to influence the AI's interpretation and the final image generation. This can be done using parentheses, repeated weighting, or numerical values to indicate the level of emphasis on certain elements, ensuring that the AI prioritizes them accordingly.

💡Visual Deviation

Visual Deviation refers to the instances where the AI-generated image does not strictly adhere to the textual prompt provided and introduces elements not mentioned or deviates from the expected output. The video discusses how Stable Diffusion sometimes applies its own interpretation, leading to variations in the final images.

💡Image Generation

Image Generation is the process of creating visual content using AI models like Stable Diffusion. It involves providing textual prompts that describe the desired image, and the AI model then generates an image that corresponds to those descriptions. The video focuses on teaching viewers how to effectively use this technology to create images that match their vision.

💡Textual Description

A textual description is a written account that details the visual elements and characteristics that a user wants to be included in the generated image. In the video, textual descriptions are essential for guiding the Stable Diffusion model in creating the desired images, and the video provides insights on how to craft effective descriptions.

💡AI Interpretation

AI Interpretation refers to how the AI model, such as Stable Diffusion, understands and processes the textual prompts to generate images. The video highlights that while the AI generally follows the instructions in the prompts, it may also apply its own logic or style, leading to images that may not exactly match the prompt but still convey the intended concept.

💡Creative Control

Creative Control in the video refers to the level of influence a user has over the final output of the image generated by Stable Diffusion. By manipulating the prompts and using weighting techniques, users can exercise a higher degree of control over the visual aspects and style of the images produced, allowing for a more personalized and targeted creative output.

💡Prompt Weights

Prompt Weights are numerical values assigned to specific elements within a textual prompt to indicate their relative importance in the image generation process. The video explains that by using prompt weights, users can guide the AI to prioritize certain aspects of the image, such as emphasizing the size or color of an object, to achieve a more accurate representation of the user's vision.


Introduction to Stable Diffusion as a tool for image generation based on text prompts.

Explanation that even experienced users of Stable Diffusion may be watching the video for further insights.

Mention of the basic usage tutorial available for beginners in the video description.

Description of how Stable Diffusion takes text prompts to create images, sometimes adding its own interpretation to the prompts.

Example of a detailed prompt provided to illustrate the process of image generation.

Discussion on how Stable Diffusion may not always follow the exact words of the prompt, showing its autonomy in image creation.

Demonstration of creating multiple images based on the provided prompt, showcasing the range of outputs.

Explanation of instances where the generated images did not strictly adhere to the prompt's specifications.

Introduction of the method to emphasize certain words in the prompt by using parentheses.

Description of the 'prompt weight' concept and how it can be applied to increase the influence of specific words.

Illustration of how to use double parentheses for triple the weight of a word in the prompt.

Introduction of another way to assign weight to prompt words by using a colon and a numerical value.

Clarification that the numerical value for weight can exceed 1, thus intensifying the influence of the word more than the base weight.

Example of how to apply prompt weight to the detailed prompt, aiming for a more precise image generation.

Explanation of how to identify and correct errors in the prompt using the red border feature.

Conclusion that effective use of prompt weights can help achieve the desired image more easily and accurately.

Encouragement for viewers to subscribe and set alarms for more content, highlighting the educational value of the video.