【Stable Diffusion】ポーズの呪文集プロンプト一覧を紹介！
TLDRThe video provides a comprehensive guide on using 'pose spells' to enhance image generation with Stable Diffusion. It emphasizes the importance of entering the correct pose spell to achieve desired poses, such as standing, sitting, or lying down. Tips include adding specific phrases to the prompt to increase the likelihood of generating poses like 'fold-arms' or 'one hand on hip'. The video also suggests using aspect ratios and adjusting the spell's influence with a numerical value (up to 1.3) for better results. It advises prioritizing the pose spell at the beginning of the prompt for higher chances of generation. Lastly, the video recommends using 'open pose editor' for poses not covered in the guide and encourages viewers to experiment with different prompts and ratios to achieve the ideal pose.
Takeaways
- 📚 Use specific pose spells to increase the variation of poses in stable diffusion, as it often requires the correct spell to generate a desired pose.
- 🧍 If you want a standing pose, use the 'standing' spell. Adding 'standing, raise one leg' can help generate a pose with one leg up.
- 🤚 Adding 'fold-arms' or 'crossed arms' can help hide hands, which is useful as AI often struggles with detailed finger depiction.
- 🤲 Adding 'arms behind back' or 'hand between legs' can also be used to hide hands, improving the quality of image generation.
- 🚗 For poses like holding a steering wheel, use descriptive spells such as 'hold the steering wheel of a car', and adjust the image aspect ratio for better results.
- 👆 'Put up index finger' can be used to depict a character wanting or thinking about something, but be aware of common AI limitations like thumb bending.
- 🧘♀️ For sitting poses, use 'sit' or 'sitting', and consider a vertically long image size for a well-balanced image.
- 🏃 Use 'walking', 'run', or 'running' for movement poses, and be patient as it may take several attempts to generate the desired pose.
- 🛌 For lying down poses, use 'lie down' and adjust the aspect ratio to vertical for a supine position, or horizontal for side or stomach poses.
- 😴 To generate sleeping poses, use 'sleep' and 'closed eyes' for a more natural depiction, and consider the aspect ratio for the best results.
- 📐 Changing the aspect ratio and adjusting the spell's influence with a numerical value (up to 1.3) can help overcome difficulties in generating specific poses.
Q & A
What is the purpose of using 'pose spells' in Stable Diffusion?
-Pose spells are used in Stable Diffusion to increase the variation of poses and to improve the chances of generating specific poses that the user has in mind. They are particularly useful when the desired pose is not easily achieved through the default settings.
How can adding 'standing' to the prompt affect the generation of a pose in Stable Diffusion?
-Adding 'standing' to the prompt increases the probability of generating a standing pose. It is especially beneficial when you want to ensure that the generated image depicts a person standing up.
What is the recommended method to hide hands in AI illustrations?
-To hide hands, you can use spells like 'fold-arms', 'crossed arms', 'arms behind back', or 'hand between legs'. Hiding the hands is recommended because AI illustrations often struggle to create detailed and beautiful fingers.
What is the effect of using 'arms raised in the air' in a pose spell?
-Using 'arms raised in the air' in a pose spell can generate poses where the character is making a gesture like banzai, waving with both hands, or saying goodbye. It is a common pose that artists often use.
How can you increase the influence of a pose spell in Stable Diffusion?
-You can increase the influence of a pose spell by adding a numerical value after the spell, with a colon separating them, like in '(run1.3)'. The recommended value is at most 1.3 to avoid deteriorating the overall quality of the generated image.
Why is it suggested to place the pose spell at the beginning of the prompt in Stable Diffusion?
-Placing the pose spell at the beginning of the prompt and at the top of the priority list makes it more likely for the desired pose to be generated, as the system tends to prioritize spells that are listed first.
What aspect ratio is best for generating standing or sitting poses?
-A vertical aspect ratio is recommended for generating standing or sitting poses as it tends to produce better results for these types of poses.
How can changing the aspect ratio help in generating poses that are difficult to achieve?
-Changing the aspect ratio to vertical, square, or landscape can help in generating poses that are difficult to achieve. It allows for trial and error to find the most effective ratio for the desired pose.
What is the recommended approach when you want to generate a pose that involves lying down?
-For lying down poses, it is recommended to use a vertical aspect ratio. If you want a specific type of lying down pose, such as on the back or face down, you should include 'lie down', 'lie face down', or other relevant spells in your prompt.
How can the 'open pose editor' be used to generate an ideal pose in Stable Diffusion?
-The 'open pose editor' allows for more precise control over the generation of poses. If a specific pose is not in the introduced spells or cannot be generated using the standard methods, the 'open pose editor' can be used to manually specify the desired pose.
What is the significance of the aspect ratio in generating poses like sitting on the ground or lying face down?
-For poses where the character is sitting on the ground or lying face down, a square or landscape aspect ratio is recommended. These ratios are more conducive to generating balanced images for these types of poses.
Outlines
🎭 Introduction to Pose Spells for Image Generation
The first paragraph introduces the concept of 'pose spells' which are specific prompts used to guide AI in generating images with desired poses. It emphasizes the importance of using these spells to increase the variety of poses and improve the success rate of generating stable images. The paragraph covers various pose spells, such as standing, sitting, and lying down, and provides tips on how to modify these spells to achieve specific outcomes, like hiding hands or adjusting the position of arms. It also discusses the challenges of generating images with certain poses and offers solutions like changing the aspect ratio or using a specific numerical value to enhance the spell's influence.
🖼️ Tips for Generating the Ideal Pose
The second paragraph delves into strategies for successfully generating the ideal pose using AI. It suggests changing the aspect ratio of the image to vertical, square, or landscape to influence the type of pose that is generated. The paragraph also recommends adjusting the numerical value after the spell to fine-tune the outcome, with a caution that values higher than 1.3 may degrade image quality. It highlights the importance of prioritizing certain spells by placing them at the beginning of the prompt to increase the likelihood of generating the desired pose. Additionally, it mentions the use of 'open pose editor' for more complex or specific poses and encourages viewers to refer to the video for a demonstration on using this tool.
📈 Summary of Pose Spell Techniques
The third paragraph summarizes the key points discussed in the video script. It reiterates the effectiveness of changing the aspect ratio and adjusting the spell's influence value to generate desired poses. The paragraph also emphasizes the importance of prioritizing the pose spell in the prompt for better results. It concludes by encouraging viewers to use the provided information to generate images with their ideal poses and to subscribe to the channel for more content on generated AI.
Mindmap
Keywords
💡Stable Diffusion
💡Pose Spell
💡Image Generation
💡Aspect Ratio
💡Hands
💡Numerical Value
💡Open Pose Editor
💡Walking
💡Lying Down
💡Kneeling
💡Lotus Position
Highlights
Stable diffusion requires the correct pose spell to generate poses.
Adding a standing spell increases the probability of generating a standing pose.
Combining spells like 'standing, raise one leg' can generate specific poses.
Using 'fold-arms' or 'crossed arms' can help hide hands and improve image generation quality.
Adding 'arms behind back' is a recommended method for those wanting to hide hands.
Including 'hand between legs' and 'sit' can facilitate better pose generation.
Poses like 'put hands hip' and 'one hand on hip' can be used for model-like poses.
Combining 'own hands together' can be challenging due to finger rendering issues.
Entering 'hold the steering wheel of a car' can help generate specific action poses.
Using 'reaching out' can be difficult due to finger count and shape complexities.
Artistic poses like 'arms raised in the air' can be achieved with the correct spell.
Adding 'piece sign' can help generate poses with a specific hand gesture.
Using 'put up index finger' can create a pose that indicates wanting or thinking.
For balanced images, sitting poses are better generated with a vertically long image size.
Inserting 'walking', 'run', or 'running' can generate dynamic movement poses.
Adding 'jumping' can create poses with moving hair, but it's challenging to generate well.
To generate a supine pose, using a vertical aspect ratio is recommended.
For sleeping poses, 'sleep' and 'closed eyes' can be combined for more realism.
Experimenting with aspect ratios can help generate desired standing or sitting poses.
Adjusting the spell's influence with a numerical value up to 1.3 can improve pose generation.
Placing the pose spell at the beginning of the prompt increases the likelihood of generation.
Using 'open pose editor' can help generate poses not covered in the video.