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

TLDRThe video introduces a new photorealistic model called 'Breathing Evomix,' developed by Studio Log and featuring a blend of models from East Asian backgrounds. It emphasizes the model's ease of use under its permissive license and recommends settings like Easy Negative or Badbras for optimal results. The video also guides viewers on setting up an environment using Google Colab and demonstrates the model's capabilities by generating images with various prompts, highlighting its potential for diverse image creation alongside other popular models like Childlimix and BRA.

Takeaways

  • 🎉 The introduction of a new model, Brashing Evomix, by the well-known studio artist, who also created the popular Child Remix.
  • 🌟 Brashing Evomix is centered around an open learning model called Open Blar, created by the famous BRA, both of which are highly regarded in the real-photo model community.
  • 📄 The official page of Brashing Evomix provides detailed information on licensing and recommended settings for users, encouraging them to read before utilizing the model.
  • 💡 Brashing Evomix is licensed for commercial use, making it highly accessible and user-friendly.
  • 🚫 While Easy Negative and Badbras are recommended settings, Easy Negative 2 is discouraged due to a higher human structure breakdown rate compared to Easy Negative.
  • 🔍 High-resolution settings are noted to be effective for background and texture quality, with many fine details documented for users to check.
  • 💻 The script provides a step-by-step guide to set up the environment using Google Colab, including commands for those interested in trying it out.
  • 🎨 The script showcases examples of images generated using Brashing Evomix, demonstrating its capability to create various images alongside Child Remix and BRA models.
  • 🖌️ ControlNet is also utilized for image generation, with reference-based preprocessors used to achieve a good result, maintaining the essence of the original image.
  • 📸 OpenPose is experimented with for image generation, proving to be effective in creating appealing images as well.
  • 🌐 The video emphasizes the versatility of using different models like Brashing Evomix, Child Remix, and BRA to generate a diverse range of images, even using existing prompts from Child Remix and BRA.

Q & A

  • Who is the creator of the new model, Bragging Evobmix?

    -The new model, Bragging Evobmix, has been released by the well-known studio member, Lou Gehrig, who is also the author of the popular Child Remix.

  • What type of model does Bragging Evobmix belong to?

    -Bragging Evobmix is a photo model that belongs to the East Asian category.

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

    -Both Child Remix and BRA are highly renowned in the realm of photo models, and their creators' collaboration can be felt in the new model, Bragging Evobmix, which is centered around the learning model called Open Blara.

  • What is the recommended setting for using Bragging Evobmix?

    -The recommended settings for Bragging Evobmix include Easy Negative or Badbras, with Easy Negative 2 being discouraged due to a higher rate of human structure collapse compared to Easy Negative.

  • How can the Bragging Evobmix license be utilized?

    -The Bragging Evobmix license allows for commercial use of the generated images, making it very user-friendly.

  • What is the significance of the High-Resolution setting in Bragging Evobmix?

    -The High-Resolution setting in Bragging Evobmix is effective for enhancing the background and texture of the generated images.

  • How can one set up the environment to try Bragging Evobmix?

    -The environment for trying Bragging Evobmix can be set up using Google Colab, with detailed commands provided in the video script and a URL link to the commands available in the summary section.

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

    -The demonstration uses version 1.21 of the Stable Diffusion Web UI, as the latest version, v1.30, generates images with filenames that differ from the usual defaults.

  • How does the use of Control Nets with reference preprocessors affect the image generation in Bragging Evobmix?

    -Using Control Nets with reference preprocessors in Bragging Evobmix can lead to the generation of images with a good feel, maintaining the atmosphere of the original image.

  • What can be inferred about the versatility of Bragging Evobmix compared to Child Remix and BRA?

    -Bragging Evobmix, along with Child Remix and BRA, can generate different facial features, allowing users to create a variety of images by utilizing these models separately.

  • How can existing prompts from Child Remix and BRA be used effectively with Bragging Evobmix?

    -Existing prompts from Child Remix and BRA can be directly applied to Bragging Evobmix to generate images with a similar feel, making it easy for users to start using the model effectively.

  • What is the value of subscribing to the channel that discusses image generation?

    -Subscribing to the channel provides viewers with daily content on image generation themes, which can serve as an inspiration and support for those interested in the subject.

Outlines

00:00

🎨 Introduction to the Bragging Evomix - A New Photorealistic Model

This paragraph introduces the release of a new photorealistic model called Bragging Evomix by the well-known Studio Log. The model is centered around an East Asian human model, similar to the famous BRA model by the author Housaku. Both models are popular in the real-image modeling community and have been featured on this channel before. The Bragging Evomix is described as a 'sauce bread' of photorealistic models, and the script will guide the audience through its features and usage. The official page of Bragging Evomix provides detailed information on licensing and recommended settings for users, emphasizing the ease of use and commercial viability of the model. The license allows for the generation of images that can be used commercially, making it highly accessible.

05:03

🖼️ Utilizing Bragging Evomix for Image Generation - Tips and Techniques

This paragraph delves into the practical application of the Bragging Evomix model for image generation. It discusses the recommended settings for optimal results, such as Easy Negative or Badbras, with a preference for Easy Negative over Easy Negative 2 due to lower human structure breakdown rates. The script also mentions the importance of high-resolution settings for background and texture quality. The audience is encouraged to check the detailed documentation for these settings. The video then transitions into setting up the environment for trying out Bragging Evomix using Google Colab, providing commands and URLs for easy replication. The script mentions the use of Stable Diffusion Web UI v1.21 for this purpose, despite the latest version being v1.30, due to filename inconsistencies in the latest release. Additional commands for Easy Negative and Control Net introduction are also included. The paragraph concludes with the anticipation of generating images using the recommended settings and promises to showcase some of the created images later in the video.

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 used in the video to refer to a specific type of photo model that is well-known in the real-photo model community. It is mentioned as one of the models that the channel has previously introduced and is being compared with the new model, ブレーシングエボミックス (Breathing Evomix). The comparison illustrates how different models can generate distinct types of images, emphasizing the diversity and flexibility of the new model being introduced.

💡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 newly released model discussed in the video. It is described as a type of photo model that is part of the 極東アジア人系 (East Asian) category. The model is noted for its detailed and realistic features, which allow for the generation of high-quality images. The video emphasizes its ease of use under a user-friendly license and provides recommendations for settings to optimize image generation.

💡実写系モデル (Real-photo Model)

実写系モデル (Real-photo Model) refers to a category of models used in image generation that are based on real photographs. These models are known for their realistic textures and details, which can produce high-quality, lifelike images. In the context of the video, the term is used to describe the type of models being discussed, such as チルドリミックス and ブレーシングエボミックス, and to highlight the capabilities and potential applications of these models in creating various types of images.

💡BRA

BRA is another real-photo model mentioned in the video, known for its detailed and realistic features. It is used as a point of comparison to チルドリミックス and the new ブレーシングエボミックス model. The video suggests that by understanding the different characteristics of these models, users can choose the most suitable one for their image generation needs.

💡ライセンス (License)

ライセンス (License) in the context of the video refers to the legal permissions and restrictions associated with using the ブレーシングエボミックス model. It is important for users to understand the terms of the license to ensure they are using the model in compliance with its intended use. The video highlights that the ブレーシングエボミックス model has a user-friendly license that allows for commercial use, making it accessible for a variety of creative projects.

💡イージーネガティブ (Easy Negative)

イージーネガティブ (Easy Negative) is a recommended setting mentioned in the video for using the ブレーシングエボミックス model. It is a parameter that affects the image generation process, likely related to the balance between detail and stylization in the final image. The video suggests that this setting, along with others like Badbras, should be considered when generating images to achieve the desired outcome.

💡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 using the ブレーシングエボミックス model. High resolution indicates that the images have a greater level of detail and clarity, which can be particularly beneficial for background textures and overall image quality. The video suggests that users should consider this setting when generating images to enhance the visual appeal and professional quality of their work.

💡Googleコラボ (Google Colab)

Googleコラボ (Google Colab) is a cloud-based platform for machine learning and programming that is mentioned in the video as a tool for users to try out the ブレーシングエボミックス model. It allows users to run Python code in their browser and use pre-configured environments, making it accessible for those who may not have the necessary setup on their local machines.

💡ステーブルディフュージョンWEB UI (Stable Diffusion WEB UI)

ステーブルディフュージョンWEB UI (Stable Diffusion WEB UI) refers to a specific version of a user interface for the Stable Diffusion model, which is used for image generation. The video mentions using version v121 instead of the latest v130 due to differences in the default file naming conventions, indicating a preference for stability and familiarity over the latest features.

💡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 or feature mentioned in the video that is used to refine the image generation process. It likely involves controlling certain aspects of the generated images to ensure they meet specific criteria or styles. The video suggests using Control Net in conjunction with reference-based preprocessors to achieve a desired look in the final images.

💡オープンポーズ (Open Pose)

オープンポーズ (Open Pose) refers to a method or technology used in the video for generating images. It is likely a feature that allows for the creation of images with a focus on specific poses or body language. The video suggests that using Open Pose can result in the generation of high-quality images that capture a particular atmosphere or style.

💡プロンプト (Prompt)

プロンプト (Prompt) in the context of the video refers to the text or input used to guide the image generation process. It is a crucial element that helps determine the content and style of the generated images. The video suggests that existing prompts used with other models like チルドリミックス can be smoothly transitioned to the ブレーシングエボミックス model, indicating compatibility and ease of use.

Highlights

The release of a new model, 布蕾西ングエボミックス, by the well-known studio楼閣さん, who is also the creator of the popular チルドリミックス.

The new model, 布蕾西ングエボミックス, is a photorealistic model focused on the East Asian demographic.

The model is centered around the learning model known as Open, ブラ, created by 挽回さん, who is famous for BRA, another photorealistic model.

Both チルドリミックス and BRA are highly renowned photorealistic models, and their creators' collaboration can be felt in the new 布蕾西ングエボミックス model.

The official page of 布蕾西ングエボミックス provides detailed information on licensing and recommended settings for the model.

The model allows for commercial use of the generated images, making its license very user-friendly.

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

High-resolution settings are effective for background and texture quality, as detailed in the documentation.

The video provides a step-by-step guide on how to construct the environment using Google Colab for those interested in trying it out.

The video utilizes Stable Diffusion Web UI v1.2.1 for the demonstration, opting for this version over the latest due to filename differences in the latest release.

Control nets are also utilized for a more refined image generation process.

The video showcases some generated images using the recommended settings, demonstrating the model's capabilities.

The model is capable of producing diverse faces, different from チルドリミックス and BRA, offering more variety in image generation.

Existing prompts from チルドリミックス and BRA can be smoothly utilized with the new model, making it easy to start using it right away.

The video emphasizes the channel's commitment to creating daily content on image generation, serving as an inspiration for viewers.

The video concludes with a showcase of the images generated using the 布蕾西ングエボミックス model, highlighting its practical applications.