Inpaint using ANY typical SDXL model in ComfyUI

Nerdy Rodent
12 May 202415:50

TLDRIn this video, the host dives into the world of image inpainting using the latest SDXL models in ComfyUI. The technique allows for creative editing by removing, changing, or adding elements to images. The host explores the capabilities of two new models, Brushnet SDXL and Power Paint V2, comparing them to the special SDXL inpainting model. Through demonstrations, viewers learn how to use masks, prompts, and various settings to achieve desired outcomes, such as transforming faces into different styles or objects. The video also covers advanced features like object removal, content-aware editing, and blending in painting to maintain the original image's context. The host provides a step-by-step guide on how to set up and use these models in ComfyUI, including downloading required files and custom nodes, ensuring a comprehensive understanding for viewers interested in image editing and creative digital art.


  • 🎨 **Inpainting Technique**: The video discusses inpainting, a technique that allows for the editing of images by removing, changing, or adding elements, which opens up creative possibilities.
  • 🔍 **Power Paint Paper**: Power Paint Paper is highlighted as a tool for inpainting, with examples shown where elements are added or removed from images based on a mask.
  • 🆕 **New Models**: Introduction to new inpainting models such as Brush Net SDXL and Power Paint V2, which are improvements over older models.
  • 🛠️ **Comfy UI Usage**: The video demonstrates how to use the new inpainting models within Comfy UI, a user interface for image editing.
  • 📘 **Mask and Prompt**: Emphasizes the necessity of an input image and a mask for inpainting, along with the importance of a prompt for the desired outcome.
  • 🦬 **Brush Net SDXL**: Brush Net SDXL is shown to work with typical SDXL models, allowing users to transform any Sable Diffusion 1.5 base model into an inpainting model.
  • 🧩 **Mask Options**: The video explores different mask options, such as segmentation and random masks, and their impact on the inpainting process.
  • 🔗 **Workflow Considerations**: Not all workflows and nodes are compatible with Brush Net due to its structure, and model changes should be made before Brush Net nodes.
  • 📈 **Scaling Effects**: The effect of scaling on the inpainting process is discussed, showing how it can increase the influence of the prompt.
  • 🖌️ **Blended Inpainting**: The Blended Inpaint node is introduced, which helps to blend the inpainted areas more naturally, especially useful when dealing with text in images.
  • 📚 **Downloading and Setup**: The video provides a guide on downloading the required model files from Google Drive and Hugging Face, and setting them up within Comfy UI.

Q & A

  • What is inpainting and how does it enhance creative options in image editing?

    -Inpainting is a technique that allows the editing of images by removing, changing, or adding elements. It opens up creative possibilities by enabling users to modify existing images in various ways, such as adding new objects or removing unwanted elements.

  • What are the new inpainting models mentioned in the script?

    -The new inpainting models mentioned are BrushNet SDXL and Power Paint V2. These models offer advanced capabilities for image editing through inpainting.

  • How does Power Paint V2 differ from the older models?

    -Power Paint V2 takes inspiration from BrushNet and introduces the ability to transform any Sable Diffusion 1.5 base model into an inpainting model, offering more versatility and creative control.

  • What is the role of a mask in the inpainting process?

    -A mask is used to define the area of the image that will be subject to inpainting. It determines where elements will be added, removed, or changed, allowing for precise control over the editing process.

  • How can one use an existing standard SDXL model for inpainting with BrushNet?

    -With BrushNet, users can utilize any standard SDXL models they already have by selecting the model within the BrushNet loader and proceeding with the inpainting process without needing a dedicated inpainting model.

  • What is the significance of the 'random mask' option in BrushNet?

    -The 'random mask' option allows for more creative freedom by not adhering as strictly to the mask shape. This can result in more varied and unique outputs, especially when attempting to create complex transformations like changing a human face to a rodent face.

  • How does the scale parameter in BrushNet affect the inpainting results?

    -The scale parameter in BrushNet adjusts the level of detail and prompt influence in the inpainting results. A lower scale value increases the prompt's power, leading to more significant changes in the image, while a higher value retains more of the original image's details.

  • What is the purpose of the 'Blended in Paint' option in the inpainting process?

    -The 'Blended in Paint' option is used to seamlessly blend the inpainted areas with the rest of the image, particularly useful when the original masked area's surroundings are also slightly altered and need to look natural.

  • How does Power Paint V2's object removal function work?

    -Power Paint V2's object removal function uses the 'segment anything' prompt to automatically detect objects and create a mask for them. It then replaces the masked objects with elements specified in the prompt, such as turning leaves into a luxury bed.

  • What are the recommended prompt additions when using Power Paint V2 for specific functions?

    -For object removal, add 'empty scene blur' to the prompt. For context aware and out painting, include 'empty scene' in the prompt to guide the model towards better results.

  • How can users get started with these inpainting models in their ComfyUI setup?

    -Users need to install the BrushNet custom nodes through the manager in ComfyUI, download the required model files from sources like Google Drive or Hugging Face, and follow the instructions for setting up directories and renaming files to match the structure provided by the custom node's documentation.



🖌️ Exploring Advanced Inpainting Techniques

This paragraph introduces various inpainting models and techniques that enable image editing by adding, removing, or altering elements within an image. It discusses the use of inpainting for creative image manipulation, highlighting models such as BrushNet SDXL and Power Paint V2. The segment showcases how these tools allow for extensive modifications in images, such as changing faces and adding objects, using masks and different inpainting methods. Additionally, the video demonstrates the setup and usage of these models through a practical example involving the transformation of an image to include a bearded hipster face instead of the original.


🔍 Improving Image Integrity with Blended Inpainting

The second paragraph addresses challenges and solutions related to preserving the integrity of unmasked areas during inpainting. It explores how blending techniques can prevent unwanted changes to critical parts of an image, such as text, ensuring they remain readable post-editing. The discussion includes comparing different inpainting models, focusing on their effectiveness in integrating edits seamlessly into the original image. Through examples, it explains the use of various masking and inpainting strategies to enhance image authenticity without altering essential content.


🐶 From Dogs to Cats: Advanced Contextual Editing

Here, the focus shifts to more advanced inpainting applications, using a playful example of replacing a dog’s head with a cat’s head within an image. This paragraph explains how different inpainting models handle context and masking, and the importance of proper mask application for effective results. It also touches on technical aspects such as setting up and installing necessary components and model files for inpainting tasks in a home setup, providing practical guidance for users to follow along and execute similar projects.


🚀 Leveraging Custom Nodes and Online Resources

The final paragraph rounds up the discussion by exploring the broader implications and possibilities offered by custom nodes and model integrations. It guides viewers through the process of installing custom nodes and downloading model files from various online sources. The narrative also emphasizes the importance of organizing and renaming files to manage resources effectively. Additionally, it encourages viewers to explore further by visiting GitHub and other platforms for more examples and detailed tutorials, enhancing their understanding of the subject.




Inpainting is a technique used in image editing that allows for the removal, change, or addition of elements within an image. It is a creative process that can be used to fix imperfections, alter scenes, or add new details. In the video, inpainting is the central theme, with the host demonstrating how to use various models to edit images creatively.


ComfyUI is a user interface mentioned in the video that is used for image editing tasks, specifically for the inpainting technique. It is the platform where the host shows how to apply different inpainting models and custom nodes to achieve desired effects on images.

💡SDXL Model

SDXL stands for Sable Diffusion XL, which is a type of model used in AI image generation. In the context of the video, the host discusses using an SDXL model for inpainting tasks, transforming it into an inpainting model to edit images in various ways.


A mask in image editing is a selection that defines which parts of an image will be affected by a certain operation. In the video, masks are used to specify areas of the image that the inpainting models will focus on, such as adding or removing elements.


In the context of AI image generation, a prompt is a description or instruction given to the model to guide the output it generates. The host uses prompts to instruct the inpainting models on what changes to make in the images, such as transforming a face into a 'bearded hipster dude'.

💡Brush Net

Brush Net is a specific model or tool discussed in the video that is used for inpainting. It works with typical SDXL models and allows for various creative edits to images, including the addition or removal of elements.

💡Power Paint V2

Power Paint V2 is a newer version of a model or tool for image editing that has been inspired by Brush Net. It offers various functions for inpainting, such as object removal, content-aware editing, and image outpainting. The host demonstrates how to use Power Paint V2 for advanced image editing.

💡Segmentation Mask

A segmentation mask is a type of mask used in image editing that separates different regions of an image for individual manipulation. In the video, the host uses a segmentation mask to select and edit specific parts of an image, such as changing the background.

💡Random Mask

A random mask, as discussed in the video, is a type of mask that does not adhere strictly to a specific shape, allowing for more varied and less predictable edits. The host illustrates how a random mask can create more unique and creative results in image editing.

💡Image Blending

Image blending is the process of smoothly integrating edited parts of an image with the original, unaltered areas. The host mentions the use of a 'blended in paint' option to ensure that the edited text in an image remains readable and naturally fits with the rest of the image.

💡Custom Nodes

Custom nodes are user-created or third-party components that can be added to a software interface like ComfyUI to extend its functionality. In the video, the host discusses installing custom nodes for Brush Net and Power Paint V2 to enable the use of these specific inpainting models within ComfyUI.


Inpainting is an image editing technique that allows for the removal, change, or addition of elements within an image.

Power Paint Paper showcases examples of inpainting, including adding and removing items from images.

The video demonstrates the use of new inpainting models, Brush Net SDXL and Power Paint V2.

Power Paint V2 is inspired by Brush Net and can transform any Sable Diffusion 1.5 base model into an inpainting model.

Comfy UI is used to demonstrate the inpainting process with the new models.

Brush Net allows the use of any standard SDXL models for inpainting without needing a specific inpainting model.

The video provides a step-by-step guide on using inpainting with masks and prompts to achieve desired outputs.

Different mask shapes can lead to unique and creative results in inpainting.

The scale parameter in Brush Net can significantly change the output, demonstrating the power of prompt control.

Blended in-painting is a feature that helps maintain the original tone and texture of the image.

Text within an image can be preserved using the Blended in-painting option to avoid changes to the text.

The video explains the limitations and correct workflow when using Brush Net with other nodes in Comfy UI.

Power Paint V2 offers various functions such as text-guided, shape-guided, object removal, and content-aware image out-painting.

The author recommends specific prompt phrases when using certain functions of Power Paint V2 for optimal results.

Object removal in Power Paint V2 is better suited for background objects rather than foreground ones.

The video provides instructions on how to integrate the new models into an existing Comfy UI setup, including downloading and installing custom nodes.

The GitHub page for Brush Net and Power Paint offers more options, settings, and examples for users to explore.

The video includes a Patreon mention for additional examples and notes related to Brush Net and Power Paint.