使用Ai绘图工具创作连续性人物和角色 - Leonardo, Midjourney, Stable Diffusion的操作技巧
TLDRIn today's episode, Xiao Wei explores the creation of continuous characters across AI art platforms like Stable Diffusion, Midjourney, and Leonardo. She highlights Leonardo's user-friendly interface and the use of fine-tuned models for generating high-quality images with continuity by incorporating character names into prompts. Midjourney's version updates and their impact on character continuity are discussed, alongside the versatility of Stable Diffusion for modifying characters while maintaining consistency. Xiao Wei also shares practical tips for naming characters and altering their backgrounds, hairstyles, and settings to achieve continuity. This tutorial is a must-watch for anyone looking to create consistent characters across different AI art platforms without spending money.
Takeaways
- 🎨 The video discusses techniques for creating consistent characters across Stable Diffusion, Midjourney, and Leonardo AI art generation platforms.
- 🌐 Leonardo is an online drawing site based on the Stable Diffusion model, offering free users enough points and a variety of features for high-quality AI drawing.
- 📝 To create a consistent character in Leonardo, input simple commands and use prompt words effectively to generate high-quality images.
- 🔍 For character consistency, assign a name to the character using a random name generator and include it in the prompt to help the system remember facial features.
- 💇 Modify the character's appearance by changing attributes like hair color and background while keeping the name constant for high similarity in generated images.
- 🌏 In Midjourney V4, continuity was achieved with Seed and URL, but V5 de-emphasized Seeds, requiring a combination of URL, prompt, and parameters.
- 🖼️ Upload reference images and provide detailed descriptions to generate similar characters in Midjourney, though without a fine-tuned model, exact consistency may not be achieved.
- 🎭 Test different styles like 'niji 5 mode' in Midjourney for varied character appearances.
- 🔄 In Stable Diffusion, download images from Civitai community, view model and parameter details, and replicate them using the WebUI and additional models like NeverEnding Dream.
- 🌄 Modify prompts in Stable Diffusion by translating and altering descriptions to change the background and setting while keeping the character's appearance intact.
- 📢 The video encourages viewers to like, subscribe, and share the content, and to reach out with questions or for further assistance.
Q & A
What is the main topic of the video?
-The main topic of the video is the creation of continuous character images using AI platforms such as Stable Diffusion, Midjourney, and Leonardo.
How does Leonardo help in achieving continuous character creation?
-Leonardo, an online drawing website based on the Stable Diffusion model, allows users to create continuous character images by using simple commands and prompt words. It also enables users to input a character name, which helps the system remember the facial features for consistent character representation.
What is the role of the 'random name generator' in the Leonardo platform?
-The 'random name generator' is used to create a unique name for the character, which assists the Leonardo system in recognizing and maintaining the character's facial features across different images, thus ensuring continuity in character creation.
How can users modify character images while maintaining continuity on Leonardo?
-Users can modify character images on Leonardo by changing elements like hair color and background while keeping the character's name constant. This allows the system to retain the character's facial features and maintain continuity.
What is the significance of the Seed value in Midjourney for continuous character creation?
-In Midjourney V4, the Seed value was used to ensure the continuity of character images. However, in V5, the Seed's role has been diminished, and users now rely on a combination of URL, prompt, and parameters to achieve continuity.
How can users generate continuous character images in Midjourney V5?
-In Midjourney V5, users can generate continuous character images by uploading reference images, providing detailed descriptions including character traits, background, and lighting, and then adding image ratio parameters to their prompt commands.
What is the process for creating continuous character images in Stable Diffusion?
-In Stable Diffusion, users can create continuous character images by downloading a preferred work from the Civitai community, copying its data including model, prompts, and parameters, and then modifying the prompts while keeping the original model and reference image unchanged.
How can users change the background and costume of a character in Stable Diffusion?
-Users can change the background and costume of a character in Stable Diffusion by modifying the prompt words that describe these elements. For example, changing 'flower petals' to 'snowflakes' and 'uniform' to 'swimsuit' and then translating the modified prompts into English for the system to generate new images with the updated settings.
What is the role of DeepL in the process of creating continuous character images?
-DeepL, an AI translation platform, is used to translate the positive prompts from English to Chinese and vice versa, allowing users to understand and modify the character's features and settings in their preferred language.
How can users apply the original image's data in Stable Diffusion?
-Users can apply the original image's data in Stable Diffusion by copying the image's data, including the model, prompts, and parameters, and using the 'send to text, to image' feature in the WebUI to generate a new image that closely resembles the original.
What platforms are discussed in the video for continuous character creation?
-The video discusses the use of Stable Diffusion, Midjourney, and Leonardo for continuous character creation, providing insights into how each platform can be utilized to maintain character consistency across different images.
Outlines
🎨 Creating Continuous Characters with AI Art Platforms
This paragraph introduces the concept of creating continuous characters using AI art platforms such as Stable Diffusion, Midjourney, and Leonardo. It highlights the ease of use and affordability of Leonardo, which offers a variety of fine-tuned models and sufficient points for free users. The process involves logging in, setting image parameters, using positive and negative prompts, and creating a character name for consistency. The example given is generating a female character named Pat Cameron and maintaining her likeness across different hairstyles and backgrounds. It also discusses remixing existing works and creating characters of different ages by adjusting the prompt.
🌐 Midjourney's Evolution and Character Continuity
This section discusses the evolution of Midjourney from version 4 to 5 and the changes in the method for creating continuous characters. It explains that while V4 used Seed and URL for continuity, V5 shifted to a URL+prompt+parameters approach. The process involves uploading reference images, adding detailed descriptions, and adjusting image ratios. The paragraph also explores the limitations of V5 in generating images identical to the original due to the lack of fine-tuned models. Additionally, it introduces the niji 5 mode, which produces a cuter style of images.
🖌️ Enhancing Art with Stable Diffusion and Civitai
The final paragraph focuses on using Stable Diffusion and Civitai for continuous character creation. It describes how to download images from Civitai, view and copy the image data, and use the WebUI to apply the data for new creations. The paragraph explains the need to install specific models like NeverEnding Dream and how to modify prompts for new images while retaining the original character's appearance and settings. It provides an example of changing the background and costume while keeping the character's features intact.
Mindmap
Keywords
💡Stable Diffusion
💡Midjourney
💡Leonardo
💡Prompt
💡Continuous Character Creation
💡Negative Prompt
💡Random Name Generator
💡Remix
💡Community
💡WebUI
💡Translation
Highlights
Introduction to creating consistent characters in AI art platforms like Stable Diffusion, Midjourney, and Leonardo.
Leonardo is an online drawing website based on the Stable Diffusion fine-tuned model, offering sufficient points and rich features for free users.
Users can generate high-quality images with simple commands and achieve character consistency by using prompt words effectively.
Creating a character name with a random name generator helps maintain facial features for consistent character portrayal.
Changing the character's hairstyle and background while keeping the name constant results in highly similar images.
In Midjourney V4, continuity is achieved using Seed and URL, but V5 de-emphasizes the role of Seed.
V5 version of Midjourney focuses on URL+prompt+parameters for consistent character creation.
Uploading photos of a subject and adding detailed descriptions can generate images with high similarity to the original.
Testing the niji 5 mode in Midjourney can produce more adorable images with the same prompt.
Stable Diffusion allows for continuity in character creation by copying image data and using it as a reference for new images.
Using the NeverEnding Dream and baked vae models in Stable Diffusion can generate images highly similar to the original.
Modifying prompts in Stable Diffusion and translating them can change the background and other elements while retaining the character's appearance.
Regardless of the platform, modifying prompts based on the original model and reference image enables consistent character creation.
The video concludes with a call to action for viewers to like, subscribe, and share the content, and to reach out with any questions.
The video provides practical applications for AI art enthusiasts looking to create consistent characters across different platforms.
The use of random name generators and detailed descriptions enhances the customization and continuity of AI-generated characters.
The video showcases the versatility of AI art platforms in generating characters with varying hairstyles, backgrounds, and outfits.
The transcript serves as a guide for users to navigate the features and capabilities of AI art platforms for character creation.