実写系モデルの新星『BracingEvoMix』のご紹介 (Stable Diffusion web UI)

サファはユーチューバー【AIイラスト】
2 Jun 202307:04

TLDRIn this video, the spotlight shines on 'Blazing EvoMix,' a new photorealistic model released by the well-known Studio Roukaku, famous for its Chilled Remix. Blazing EvoMix, described as a photomodel of Far East Asian descent, is built upon the OpenBlur model developed by Hui Kai, known for the widely acclaimed BRA model. The video delves into the collaborative essence of these creators, suggesting Blazing EvoMix as a pinnacle of photorealistic modeling. Viewers are guided through its commercial-friendly license, recommended settings for optimal use, and a demonstration of image generation using Google Colab and specific settings like Easy Negative. The video concludes with showcasing generated images, proving the model's versatility in creating diverse visuals, and inviting users to explore this tool for enriching their image generation projects.

Takeaways

  • 📺 Studio Roukaku has released a new model named Bracing Evomix, recognized for its lifelike Far East Asian photo model quality.
  • 📸 Bracing Evomix is built upon the famous real-life models Chilled Remix and BRA, further developed by the creator Wan Kai, focusing on the OpenBlurr learning model.
  • 🎥 Both Chilled Remix and BRA are renowned for their realism in photographic modeling, making Bracing Evomix a culmination of these efforts.
  • 📈 The official page for Bracing Evomix provides detailed licensing information, recommending the model for commercial use, highlighting its user-friendly license.
  • 🛠 Recommended settings for using Bracing Evomix include Easy Negative or Badbras, with Easy Negative 2 being advised against due to higher rates of anatomical distortion.
  • 🔍 The document provides intricate details on high-resolution settings for backgrounds and textures, suggesting users to thoroughly review it before use.
  • 💻 The tutorial uses Google Colab for environment setup, with specific commands provided for ease of replication by viewers.
  • 👨‍💻 Version 121 of the Stable Diffusion Web UI is chosen over the latest version to maintain consistency in generated image filenames.
  • 📱 The video demonstrates image generation using Bracing Evomix, applying recommended settings and comparing results with previous models like BRA.
  • 📲 ControlNet and OpenPose were also tested for image generation, showing versatility in creating lifelike images with different pre-processors.

Q & A

  • Who is the creator of the new model, Bracing Evomix?

    -The creator of Bracing Evomix is Studio楼閣, who is also well-known for creating the Childlike Remix model.

  • What type of model is Bracing Evomix?

    -Bracing Evomix is a photorealistic model featuring an East Asian person.

  • What is the significance of the collaboration between the creators of Childlike Remix and BRA models?

    -Both Childlike Remix and BRA models are popular in the photorealistic model community, and their creators' collaboration in the creation of Bracing Evomix suggests a synergy that can produce high-quality models.

  • What is the current licensing situation for Bracing Evomix?

    -As of the current date, Bracing Evomix can be used for commercial purposes, making its license very user-friendly.

  • What are the recommended settings for using Bracing Evomix?

    -The recommended settings include Easy Negative or Badbras, with Easy Negative 2 being discouraged due to a higher rate of anatomical breakdowns compared to Easy Negative.

  • How detailed is the documentation for Bracing Evomix?

    -The documentation for Bracing Evomix is quite detailed, covering even the finest aspects such as high-resolution effects on background and texture.

  • What is the process for setting up the environment to use Bracing Evomix?

    -The environment setup involves using Google Colab and following a series of commands provided in the script. The commands are designed to be easily copied and pasted for those utilizing a similar setup.

  • Which version of Stable Diffusion Web UI is used for the demonstration?

    -The demonstration uses version 1.21 of the Stable Diffusion Web UI, as the latest version at the time of the script (version 1.30) had different default filenames for generated images.

  • How does the use of ControlNet affect the image generation with Bracing Evomix?

    -ControlNet, when used as a reference preprocessing method, can help generate images with a good feel, maintaining the atmosphere of the original image.

  • What was the result of using Bracing Evomix with the recommended settings and existing prompts?

    -The images generated using Bracing Evomix with the recommended settings and existing prompts from other models like Childlike Remix and BRA resulted in high-quality, photorealistic images that smoothly continue the feel of the original prompts.

  • How can the channel mentioned in the script help users interested in image generation?

    -The channel creates daily videos on image generation themes, which can serve as both inspiration and guidance for users interested in the topic.

Outlines

00:00

🌟 Introduction to Blazing EvoMix: A New Realistic Photo Model

In this segment, the video introduces a new realistic photo model called Blazing EvoMix, developed by the renowned Studio Loukaku, known for Chilld Remix. Blazing EvoMix is described as a Far East Asian type photomodel, leveraging learning models from the creator of BRA, Hui Kai, particularly focusing on OpenBlur. This model represents a synergy between two highly acclaimed realistic models, Chilld Remix and BRA, suggesting Blazing EvoMix as a pinnacle of realistic photo model development. The video aims to showcase the capabilities and features of Blazing EvoMix, emphasizing its significance in the realm of realistic photo modeling.

05:03

📸 Generating Diverse Images with Blazing EvoMix

The second paragraph delves into the practical application and advantages of using Blazing EvoMix for image generation. It suggests that Blazing EvoMix, along with Chilld Remix and BRA, can produce a wide range of images, making it a versatile tool for creators. The segment highlights how existing prompts used with Chilld Remix and BRA can be seamlessly adapted for Blazing EvoMix, facilitating an easy transition for those familiar with realistic image generation. The encouragement for channel engagement through subscriptions and likes suggests the video's aim to foster a community around image generation themes, ending with a promise to showcase images generated during the session.

Mindmap

Keywords

💡チルドリミックス (Childlike Remix)

チルドリミックス (Childlike Remix) is a term mentioned in the script that refers to a previously known model used for image generation. It is a type of AI model that is renowned in the community for creating realistic human-like images. In the context of the video, it is one of the models compared to the newly released ブレーシングエボミックス (Breathing Evomix), highlighting the evolution and variety in AI image generation models.

💡ブレーシングエボミックス (Breathing Evomix)

ブレーシングエボミックス (Breathing Evomix) is a new model introduced by the script's author, which is also a part of the real-photographic model category. It is centered around an open learning model called OpenBler, developed by 挽回 (Wanhui), a notable author in the same field. This model is described as being very user-friendly with a permissive license for commercial use, making it accessible for various applications.

💡実写系モデル (Real-photographic Model)

実写系モデル (Real-photographic Model) refers to a category of AI models that generate images resembling real photographs of humans. These models are known for their ability to create highly realistic and detailed images, often used in various creative and commercial applications. In the video, the script compares two such models, チルドリミックス (Childlike Remix) and ブレーシングエボミックス (Breathing Evomix), to demonstrate their capabilities and potential uses.

💡BRA

BRA is another renowned real-photographic model used for generating realistic human images. It is mentioned alongside チルドリミックス (Childlike Remix) as a comparison to the newly introduced ブレーシングエボミックス (Breathing Evomix). The script suggests that these models have their unique characteristics and can be used for different purposes, offering a variety of options for image generation.

💡ライセンス (License)

ライセンス (License) in the context of the script refers to the legal permission and terms under which the ブレーシングエボミックス (Breathing Evomix) model can be used. It is mentioned that the model has a permissive license that allows for commercial use, making it highly accessible for creators and businesses. This is an important aspect for users to consider when choosing an AI model for their projects.

💡推奨設定 (Recommended Settings)

推奨設定 (Recommended Settings) refers to the configuration parameters suggested by the creators of the ブレーシングエボミックス (Breathing Evomix) model for optimal results. These settings guide users on how to best use the model to generate images, ensuring that the output meets certain quality standards. The script mentions specific recommended settings like Easy Negative and Badbras, which are crucial for users to achieve the desired image quality.

💡Googleコラボ (Google Colab)

Googleコラボ (Google Colab) is a cloud-based platform for machine learning and data analysis that is mentioned in the script as the environment used to test the ブレーシングエボミックス (Breathing Evomix) model. It allows users to run Python code in a collaborative environment, making it an ideal choice for experimenting with AI models without the need for extensive local setup.

💡ハイレゾリューション (High Resolution)

ハイレゾリューション (High Resolution) refers to the quality of the images generated by the AI model, indicating that they have a high level of detail and clarity. In the context of the script, high-resolution settings are recommended for enhancing the background and overall texture of the generated images, resulting in more realistic and visually appealing outputs.

💡コントロールネット (Control Net)

コントロールネット (Control Net) is a technique used in AI image generation models to control the output based on certain parameters or references. It allows users to guide the model's output to match specific styles, themes, or characteristics, ensuring that the generated images align with the desired outcome. The script mentions using Control Net with reference-based preprocessing to achieve a particular look in the images.

💡オープンポーズ (Open Pose)

オープンポーズ (Open Pose) is a term related to image generation that involves creating images with a focus on pose and body language. It is mentioned in the script as another method tried for generating images with the ブレーシングエボミックス (Breathing Evomix) model. The use of Open Pose can help in creating images that have a natural and dynamic feel to them.

💡スムーズに使い始め (Smooth Start)

スムーズに使い始め (Smooth Start) refers to the ease with which users can begin utilizing the AI models for image generation. The script emphasizes that by using existing prompts from models like チルドリミックス (Childlike Remix) and BRA, users can quickly adapt to using the new ブレーシングエボミックス (Breathing Evomix) model without any significant learning curve.

💡画像生成 (Image Generation)

画像生成 (Image Generation) is the process of creating visual content using AI models, which is the central theme of the video. The script discusses various models and techniques for generating realistic images, highlighting the advancements in AI technology and its applications in the field of image creation.

Highlights

Studio楼阁发布了新的模型布蕾辛混合模型。

布蕾辛混合模型是极东亚洲人系照片模特类型。

布蕾辛混合模型围绕由挽回创作的开放布拉学习模型构建。

布蕾辛混合模型与BRA和チルドリミックス模型一样，都是实拍系模型。

布蕾辛混合模型的官方页面详细记录了许可证信息和推荐设置。

该模型生成的图像可用于商业用途，许可证非常易于使用。

推荐设置包括Easy Negative或Badbras，但Easy Negative 2由于人体结构破裂率较高，不推荐使用。

文档中详细记录了高分辨率对背景和质感的影响等细节。

使用Google Colab构建布蕾辛混合模型的试用环境。

推荐使用Easy Negative生成图像，并尝试使用控制网。

使用BRA模型常用的提示词，结合Easy Negative生成图像。

使用控制网和参考系预处理器进行图像生成尝试。

通过使用额外的处理步骤，成功生成了具有原始图像氛围的女性形象。

尝试使用Open Poses生成图像，也取得了良好的效果。

布蕾辛混合模型与BRA和チルドリミックス模型相比，可以生成不同风格的面孔。

使用这些模型可以根据不同的用途灵活地生成各种图像。

对于正在进行实拍图像生成的人来说，这是一个特别推荐的模型。

频道中每天都在创建有关图像生成的视频，可以作为激励观看。

频道注册和点赞将非常感激，因为这是频道持续创作的动力。

最后，介绍了使用新模型生成的图像，感谢观看。