Stable Diffusionで高品質な美少女ができるプロンプト呪文集をまとめて紹介

AIジェネ【AIイラスト生成の情報発信】
2 Sept 202313:15

TLDRこのビデオでは、高品質の美少女画像を生成するための「呪文」を紹介しています。初心者は「stable diffusion」から始めると良いが、効果的なプロンプトを知らないと高品質の画像は生成できません。ビデオでは、全体の品質を向上させるための「masterpiece」や「best quality」、「ultra quality」などの呪文を紹介し、詳細に「detailed」や「ultra detailed」などの呪文を組み合わせて使用することで、さらに品質を向上させることができます。また、手首や指などの細部を改善するため、否定的なプロンプトとして「bad hands」や「missing fingers」を使用することも提案されています。さらに、VAEを使用することで全体の明るさと目などのハイライトを修正し、品質を向上させることができます。最後に、高品質の画像を生成するためのおすすめの3つの呪文は「masterpiece」、「best quality」、「detailed」です。全ての呪文を使えば品質が下がる可能性があるため、3つ以内に絞り込んでおくことが重要です。

Takeaways

  • 🎨 Start with the 'stable diffusion' technique for beginners to create high-quality images of beautiful girls.
  • 📝 Use the 'any lora' model from the Civitai site and enter prompts like 'girl, upper body' to begin generating images.
  • 🔍 Experiment with different prompts and spells, such as 'masterpiece', 'best quality', and 'detailed', to improve image quality.
  • 🌟 The 'masterpiece' spell is essential for enhancing the overall style and quality of the generated images.
  • 📏 Comparing results with and without half-width characters can yield different image qualities.
  • 🧩 Combining spells like 'masterpiece' and 'best quality' can further improve the quality of the generated images.
  • 👗 The 'detailed' and 'ultra detailed' spells are recommended for adding intricate details to the clothing and other elements.
  • 👀 Using 'high resolution' or 'hires' can make the eyes appear more lively with highlights.
  • 📱 The '4k' and '8k' spells can enhance the overall quality, with '8k' generally producing higher quality images.
  • 🎮 The 'playstation5' and 'unreal engine 5' spells can create highly realistic and game-like characters.
  • 🖼️ For a more realistic, non-anime style, use the 'photo realistic' spell along with 'illustration' for 2.5D effects.
  • ✅ Stick to a maximum of 3 quality-affecting spells in the prompt to avoid quality deterioration.
  • 🧙‍♂️ Use corrective spells in the negative prompt to fix specific issues like 'bad hands' or 'missing fingers'.
  • 🚫 Avoid using all the spells at once; instead, focus on a few that work well together.
  • 🔄 The 'EasyNegative' spell can significantly improve the quality of anime-style images when used in the negative prompt.
  • 🌟 Using VAE (Variational Autoencoder) can enhance the overall image quality by brightening and correcting highlights.
  • 🔧 Changing the 'sampling method' to one ending with 'karras' can help generate higher quality images.
  • 🖼️ Increasing the image size to 1024 or more and using 'hires.fix' can dramatically improve the quality, especially for full-body images.
  • 📚 Refer to the prompt content provided by Civitai for the model to understand which spells generate high-quality images.

Q & A

  • What is the recommended starting point for beginners looking to create high-quality images using stable diffusion?

    -Beginners can start with stable diffusion by entering simple prompts like 'girl, upper body' and then adding various spells to enhance the image quality.

  • What is the significance of the 'masterpiece' spell in generating high-quality images?

    -The 'masterpiece' spell is a well-known spell that greatly improves the overall style, hair, eyes, and other aspects of the generated image, making it a must-have for high-quality outputs.

  • How does the 'best quality' spell compare to the 'masterpiece' spell in terms of image quality?

    -The 'best quality' spell is considered to be slightly lower in quality compared to 'masterpiece', but when combined with it, the quality increases more effectively than using 'masterpiece' alone.

  • What is the role of the 'detailed' spell when generating images?

    -The 'detailed' spell, when combined with the 'masterpiece' spell, significantly enhances the quality of the image, particularly in terms of the details on the clothes and other elements.

  • How does the 'high resolution' spell affect the quality of the generated images?

    -The 'high resolution' spell improves the image quality by adding highlights and details, such as the white light part in the eyes, making the image look more lively and realistic.

  • What is the 'EasyNegative' extension, and how does it help in generating higher quality images?

    -The 'EasyNegative' extension is a tool that, when used in the negative prompt, can significantly improve the overall quality of the generated image, especially when creating anime-style characters.

  • What is the recommended approach when using multiple quality-enhancing spells in the prompt?

    -It is advised to use up to three spells that affect overall quality, such as 'masterpiece, best quality', and then add specific part-targeting spells or negative prompts to fix any issues with particular body parts.

  • How can the 'sampling method' be adjusted to improve the quality of generated images?

    -Changing the default 'sampling method' to a sampler with 'karras' at the end of its name can help generate higher quality images. Beginners can start with 'karras' if they are unsure.

  • What is the impact of image size settings on the quality of the generated images?

    -Increasing the image size to 1024 or more on one side and using 'hires.fix' can dramatically improve the quality, making it harder for details like eyes to collapse, even with the same prompt content.

  • How can the '4k' and '8k' spells be utilized effectively in generating high-quality images?

    -The '4k' spell can be improved when combined with the 'master piece' spell, while the '8k' spell tends to produce higher quality images on its own. It is recommended to experiment with these spells to determine which works best for a particular model.

  • What is the effect of the 'unreal engine 5' spell when generating images?

    -The 'unreal engine 5' spell, when used in the prompt, can generate images with surprisingly high quality, tending to produce results that resemble 3D game characters. It is recommended to use it in combination with the 'illustration' spell for better results.

  • What is the recommended strategy for correcting specific parts of the generated image that may not be of high quality?

    -For correcting specific parts like hands or fingers, it is suggested to use a corrective spell in the negative prompt, such as 'bad hands, bad fingers, missing fingers', to improve the quality of those parts.

Outlines

00:00

🎨 Generating High-Quality Images with Spell Prompts

This paragraph introduces the process of creating high-quality images using stable diffusion and various 'spell' prompts. It emphasizes the importance of using specific prompts like 'masterpiece', 'best quality', and 'detailed' to enhance the quality of the generated images. The video also discusses the use of the 'any lora' model from the Civitai site and the impact of half-width characters on the generation process. It provides a comparison between different quality spells and recommends combining certain spells for better results. The paragraph concludes with the suggestion to experiment with multiple generations and adjustments rather than changing spells frequently.

05:01

🖼️ Enhancing Image Quality with Specific Spells and Techniques

The second paragraph delves into additional spells that can be used to improve the quality of generated images, such as 'unreal engine 5' for a 3D game character look and 'photo realistic' for a more live-action appearance. It advises on the selection of spells based on the desired outcome and the model used. The paragraph also addresses the limitations of spell prompts and the need for corrective spells in negative prompts to fix specific issues like 'bad hands'. It highlights the 'EasyNegative' spell for its ability to significantly improve image quality and recommends a cautious approach to combining multiple spells to avoid quality deterioration. The use of VAE (Variational Autoencoder) for further enhancement and the importance of the sampling method and image size are also discussed.

10:05

📚 Best Practices for High-Quality Image Generation

The final paragraph summarizes the best practices for generating high-quality images. It suggests using a combination of 'master piece', 'best quality', and 'detailed' spells in the prompt while cautioning against overloading the prompt with too many spells. The paragraph recommends adding negative prompts to exclude low-quality elements and emphasizes the importance of testing different combinations to find the optimal balance. It also encourages users to refer to the Civitai site for prompt content guidance and to utilize tools like VAE and specific samplers for improved results. The video concludes with a call to action for viewers to subscribe to the channel for more informative content on generation AI.

Mindmap

Keywords

💡Stable Diffusion

Stable Diffusion is a term referring to a type of artificial intelligence model used for generating images from textual descriptions. In the context of the video, it is the AI model that the user is working with to create high-quality images of beautiful girls. The script mentions starting with Stable Diffusion for beginners, indicating its user-friendly nature for image generation.

💡Prompt

A prompt in the context of AI image generation is the input text that guides the AI in creating the desired image. The video discusses various prompts that users can enter to generate high-quality images, emphasizing the importance of the right prompts for achieving the best results.

💡Spell

In the video, 'spell' is used metaphorically to describe specific terms or phrases that are added to the prompt to influence the style and quality of the generated image. For instance, 'masterpiece' is a spell that improves the overall quality of the image, as explained in the transcript.

💡Quality

Quality, in this script, refers to the visual and stylistic outcome of the generated images. The video focuses on how to enhance the quality of AI-generated images, with discussions on spells and techniques that lead to higher quality results, such as 'best quality' and 'ultra high resolution'.

💡Masterpiece

As a spell in the context of the video, 'masterpiece' is a prompt addition that significantly improves the style and quality of the generated image. It is described as a must-have spell for generating high-quality images with minimal discomfort.

💡Detailed

The term 'detailed' is used as a spell to enhance the intricacy of certain aspects of the generated image, particularly the decorations on clothes. It is suggested to combine this spell with 'masterpiece' to greatly increase the quality of the image.

💡High Resolution

High Resolution is a spell that improves the clarity and detail of the generated image, making it appear more vivid and lifelike. The video mentions that using 'high resolution' can make the highlights in the eyes of the generated character more pronounced, contributing to a higher quality image.

💡Photo Realistic

Photo Realistic is a spell aimed at creating images that closely resemble real-life photographs rather than anime or illustrative styles. It is recommended for users who want their AI-generated girls to have a more lifelike appearance.

💡Negative Prompt

A negative prompt is a spell used to omit or correct unwanted elements in the generated image. The video suggests using negative prompts like 'EasyNegative' to improve the overall quality by specifying what not to include in the image.

💡vae

VAE, or Variational Autoencoder, is a technique used in AI to improve the quality of generated images by correcting details and enhancing overall visual appeal. The video recommends using VAE for an effect that brightens the image and corrects highlights, thus improving quality.

💡Sampling Method

The sampling method refers to the algorithmic technique used by the AI to generate the image based on the prompt. The video suggests changing the default 'euler a' sampling method to a 'karras' sampler for better image quality.

Highlights

Beginners can start with stable diffusion to create high-quality images of beautiful girls.

Using the right prompts and negative prompt spells can enhance the quality of generated images.

The 'masterpiece' spell is a must-have for significantly improving overall image quality.

Combining 'master piece' with 'best quality' can further increase image quality.

The 'detailed' spell, when combined with 'master piece', greatly enhances the quality of details.

Using 'all detailed' or 'ultra detailed' can improve the details of clothing in the generated images.

The 'extremely detailed' spell can produce more flashy clothing details.

Combining 'extremely detailed cg unity 8k wallpaper' with 'master piece' can yield good overall results.

The 'intricate details' spell is ideal for creating illustrations with realistic hair.

Adding 'high resolution' or 'hires' can improve the highlights in the eyes, making them look more lively.

Using 'ultra high resolution' can produce high-quality images, especially when combined with 'master piece'.

The '4k' spell works better when combined with 'master piece', enhancing the overall image quality.

The 'playstation5' spell can generate high-quality images with a gaming machine theme.

Entering 'unreal engine 5' in the prompt can produce surprisingly high-quality 3D game character-like images.

The 'photo realistic' spell is recommended for creating images with a more realistic live-action appearance.

Adding 'detailed' or 'ultra detailed' to the 'master piece' spell can improve image quality in anylora models.

Using corrective spells in the negative prompt can fix issues with specific body parts, like hands.

Including 'EasyNegative' in the negative prompt can significantly improve the overall image quality.

Using 'worst quality, low quality, normal quality' in the negative prompt can further enhance image quality.

Utilizing VAE can brighten the image and correct highlights, improving overall quality.

Changing the sampling method to one ending with 'karras' can help generate high-quality images.

Increasing the image size to 1024 or more and using 'hires.fix' can dramatically improve image quality.

Referencing prompt content from the Civitai site can provide guidance on generating high-quality images.

To generate high-quality images, combine 'master piece', 'best quality', and 'detailed' in the prompt.

Casual Browsing

【Stable Diffusion】構図・アングル・視線の呪文集プロンプトをまとめて紹介

2024-03-28 23:10:00

【Stable Diffusion】顔・目・表情・ピアスの呪文集プロンプトを一覧で紹介

2024-04-23 02:40:00

【Stable Diffusion】背景を上手く生成させるためのコツと呪文集をまとめて紹介

2024-04-23 03:15:00

【Stable Diffusion】ポーズの呪文集プロンプト一覧を紹介！

2024-04-23 04:35:00

Stable Diffusion Web UIでモデルを追加してアニメ風の美少女を生成する方法

2024-04-12 17:25:00