No.03 AI繪圖Stable Diffusion提詞完全攻略!! "(())、{{}}" 到底差在哪?

杰克艾米立
14 Jan 202310:04

TLDRIn this informative script, the speaker, 艾粒, discusses the intricacies of AI-generated art and shares tips for effective communication with AI to create desired images. They delve into various methods of prompting, including natural language, emojis, tags, and reverse prompting to avoid undesired elements. The importance of weight adjustment using brackets and the combination of colors using AND syntax are highlighted. The script also touches on the impact of image size on the complexity of the generated art and suggests further enhancements through tools like SD upscale and Photoshop adjustments.

Takeaways

  • 🎨 AI绘图技术允许用户通过描述来创造图像，节省了传统绘画的长时间磨练。
  • 🖌️ 使用自然语言和表情符号可以直接与AI沟通创作细节，但需注意AI对表情符号的理解可能存在偏差。
  • 📝 Tag提词法通过使用关键词和调整权重来精确控制AI创作内容，但可能需要细致的调整以达到预期效果。
  • 🚫 反向提词可以避免不希望出现的元素，如通过排除特定元素的描述来减少其在画面中的出现。
  • 🎨 通过调整提词的权重，可以控制画面中元素的突出程度，如增加背景描述来减少人物数量。
  • 🌈 混合使用不同颜色和提词技巧，如AND语法，可以创造出具有特定色彩和特征的图像。
  • 📏 控制画面大小也是影响画面内容的重要因素，如降低宽度可以更高效地创作单一人物图像。
  • 📚 参考元素法典可以获取更多关于提词的细节和技巧，以提高创作质量。
  • 🛠️ 利用其他工具和功能，如SD放大、图生图修改提词、局部重绘和Photoshop调整，可以进一步提升作品质量。
  • 💡 创作过程中的试验和错误是不可避免的，通过不断尝试和调整，可以逐渐找到最适合自己的创作方法。

Q & A

  • What is the main advantage of AI in creating images?

    -The main advantage of AI in creating images is that it can generate images from various descriptive inputs, allowing for the creation of complex and imaginative visuals more efficiently than traditional manual drawing methods.

  • How can natural language be used in AI image generation?

    -Natural language can be used directly in AI image generation by inputting sentences similar to everyday communication, such as 'Your mother is behind you, She looks very angry,' to create the desired image.

  • What is the significance of emoticons in AI image generation?

    -Emoticons can be used to convey specific emotions or expressions in AI image generation. However, it's important to note that the AI's interpretation may vary, and additional context might be needed to achieve the desired result.

  • What is the Tag method in AI image generation?

    -The Tag method involves using single words or phrases to communicate ideas to the AI, and then adjusting the weight of these tags using specific syntax to control the prominence of certain elements in the generated image.

  • How can reverse tagging be used effectively in AI image generation?

    -Reverse tagging can be used to specify elements that the user does not want to appear in the image. By indicating unwanted features, such as 'fused fingers' or 'twisted hands,' the AI can reduce the likelihood of these elements appearing in the final image.

  • What is the impact of the position of a prompt word in AI image generation?

    -The position of a prompt word in AI image generation affects its weight. The earlier a prompt word appears in the input, the greater its influence on the final image, as it is given higher priority by the AI.

  • How can the 'highest quality' and 'masterpiece' prompts be utilized in AI image generation?

    -The 'highest quality' and 'masterpiece' prompts are commonly used at the beginning of the input to ensure that the generated image is of the highest possible quality, as these prompts are given significant weight by their placement.

  • What is the role of image size in controlling the elements in AI image generation?

    -The image size can influence the number of elements and the focus on the main subject. For instance, reducing the width can efficiently result in a single-character image, but over-emphasizing the background might cause the characters to disappear.

  • How can the 'element lexicon' be referenced for AI image generation?

    -The 'element lexicon' is a resource that provides a collection of prompts with high影响力的强度 for image generation. It can be referenced for insights and inspiration on how to craft effective prompts.

  • What is the purpose of the automatic prompt generator mentioned in the script?

    -The automatic prompt generator is intended to streamline the process of inputting prompts for AI image generation by compiling a set of commonly used prompt combinations, making it more convenient for users.

  • How can the AND syntax be used in AI image generation for color mixing?

    -The AND syntax allows for the specification of multiple colors for an element, such as hair, with each color followed by a weight that determines its prominence in the final image. This can create a multicolored effect, with the colors distributed according to the specified weights.

Outlines

00:00

🎨 AI in Art and Imagery

The speaker, 艾粒, discusses the transformative impact of AI in the field of art and image creation. They explain how AI allows for the creation of images through descriptive inputs, enhancing both work and personal life. The speaker shares their experience with using AI for image generation, particularly in handling images for Jack's videos. They introduce various methods of prompting AI, including natural language, emojis, and reverse prompting to avoid unwanted elements. The speaker emphasizes the importance of balancing positive and negative prompts and the use of specific tags to control the output, such as NSFW and family_friendly. They also mention the significance of high-quality prompts and the use of terms like 'masterpiece' to ensure superior image quality.

05:03

🖌️ Advanced Prompting Techniques

艾粒 delves deeper into the intricacies of AI prompting, focusing on the use of symbols to adjust the weight of elements in an image. They describe their personal system of using parentheses and colons to fine-tune the prominence of certain features. The speaker provides insights into creating multicolored hair using specific syntax and discusses the AND, addition, and removal syntaxes for more control over the AI's output. They also touch on the use of OR syntax for分段式 image generation. 艾粒 concludes by suggesting further enhancements through tools like SD upscaling, image editing, and Photoshop adjustments, promising to share more about these techniques in future videos.

Mindmap

Keywords

💡AI drawing

AI drawing refers to the creation of images through artificial intelligence based on textual descriptions. This concept is central to the video's theme, highlighting the transformative impact of AI on visual creativity. The script mentions how AI drawing allows for the visualization of ideas directly from one's mind, offering a new level of imagination and detail that exceeds even the descriptions provided. Examples include creating images from natural language inputs, emojis, and kaomojis, demonstrating the AI's understanding and creative capabilities.

💡Natural syntax

Natural syntax describes the use of everyday language to communicate with AI systems for image creation. It is emphasized in the video as a straightforward method for prompting AI, allowing users to describe scenes just as they would in normal conversation. For instance, the video mentions using sentences like 'Your mother is behind you, She looks very angry' to guide the AI's drawing process, showcasing the AI's ability to interpret and visualize human language.

💡Emoji and Kaomoji

Emojis and Kaomojis serve as non-verbal cues that can be understood by AI to create images. The script illustrates their use in AI drawing prompts to test the AI's interpretive skills, leading to diverse outcomes including human emotions and textures of buildings. This highlights the AI's flexibility in understanding and transforming a wide range of symbolic inputs into visual content.

💡Tag prompting

Tag prompting involves using specific words or phrases as cues to direct the AI's image generation, often with adjustments to influence the prominence of certain elements. The video discusses its use in fine-tuning AI drawings, such as emphasizing or de-emphasizing the appearance of hands by adjusting the tags. This method illustrates the strategic manipulation of AI's focus and output quality through selective wording.

💡Inverse prompting

Inverse prompting is a technique mentioned in the script where prompts specify what should not appear in an AI-generated image, reducing the likelihood of unwanted elements. For example, by suggesting 'merged fingers' or 'distorted hands,' the speaker encourages the AI to avoid these features, thereby exerting indirect control over the image outcome. This strategy is presented as a way to balance and ease the content modification process.

💡NSFW and family-friendly

The terms 'NSFW' (Not Safe For Work) and 'family-friendly' are used in the video to illustrate how tagging can filter the content of AI-generated images. Incorporating these tags helps to manage the inclusion or exclusion of adult elements, ensuring the appropriateness of the imagery. This highlights the capacity for AI drawing tools to cater to varied audience sensitivities through careful prompt construction.

💡Quality enhancement tags

Quality enhancement tags like 'highest quality' or 'masterpiece' are recommended in the script to improve the artistic value of AI-generated images. Placing these tags at the beginning of prompts is suggested to assign them higher priority, demonstrating a technique to influence the AI's focus on producing visually superior results.

💡Weight syntax

Weight syntax involves using special symbols (like parentheses and colons) to adjust the influence of certain prompt elements on the AI's output. The video explains this as a method to fine-tune the depiction of various aspects, such as background details or character prominence, by assigning numerical weights that increase or decrease an element's impact. This advanced prompting technique allows for nuanced control over the composition and focus of images.

💡Color blending

Color blending is discussed in the context of achieving specific hair colors or artistic effects through prompt syntax that alternates between colors until a desired outcome is reached. The video's mention of creating a character with 'silver and purple hair' exemplifies how AI can be guided to mix colors in a controlled fashion, showcasing the flexibility and precision of AI in rendering complex visual details.

💡Symbolic syntax

Symbolic syntax refers to the use of special characters and formatting within prompts to guide AI behavior, such as the inclusion or exclusion of elements at specific stages of image development. The video covers the strategic use of brackets and colons for adding ('[') or removing (':') content, illustrating how temporal adjustments in the drawing process can achieve dynamic compositions or sequential storytelling in AI art.

Highlights

AI绘图能够通过各种描述方式创造出脑中想要的图像，甚至比描述的更有想象力。

自然语法可以直接用于AI绘图，如日常沟通的语句和表情符号。

表情符号在AI绘图中的使用，可以证明AI是否理解表情符号的含义。

顏文字在AI绘图中的解读可能会有多种解释，如人类表情或建筑纹理。

Tag提词法通过使用单词或短语进行交流，并调整权重来控制AI绘图的输出。

反向提词可以用来减少不希望出现的元素，如手部的绘制。

通过正向提词强调需要的元素，如手部细节，可以提高其在画面中出现的机率。

使用特定词汇如NSFW和family_friendly可以有效地过滤成人内容。

提高提词的权重可以增加特定元素在画面中的比重，如提高画质。

景深和广角的提词可以丰富画面的层次感。

画面大小的控制可以通过提词中的景物描述来实现，影响人物数量的呈现。

元素法典提供了丰富的提词强度，有助于提升画面的创作质量。

使用括号和冒号可以调整提词的权重，影响元素在画面中的存在感。

彩色头发的绘制可以通过混合不同颜色并控制权重比例来实现。

AND语法可以通过大写AND分隔不同颜色，并控制颜色的分配比例。

添加和移除语法可以用来控制元素在画面中出现和消失的步骤。

OR语法可以用来分配不同元素在画面中的绘制比例和步骤。

除了提词，还可以通过其他功能如SD放大、图生图修改提词、局部重绘等来提升作品质量。