How To install models in invoke ai And using ip adapter plus Face

28 Sept 202311:54

TLDRThe video script offers a step-by-step guide on installing and using IP adapters from, focusing on the 'IP adapter plus face' model. It explains the process of importing models, syncing them, and using them to generate images with specific facial features while maintaining the original image's context. The tutorial also covers adjusting settings for optimal results and fine-tuning the output through various actions and adjustments in the asset browser.


  • 📂 Visit to access a variety of models categorized under Concepts, Styles, and Tools.
  • 🔍 Check the model version compatibility by hovering over the yellow icon to ensure the correct encoder is downloaded and installed.
  • 🖼️ Select and add desired models, such as the IP adapter plus face, by copying the provided text and pasting it into the model manager's import section.
  • 🔄 Ensure models are synced in the settings to maintain consistency across different devices, if applicable.
  • 🖼️ Load images with maintained proportions (e.g., 384x512 and 512x768) for better processing.
  • 🎨 Utilize image-to-image mode for generating variations based on the input image and the chosen model.
  • 🔧 Adjust denoise strength to find a balance between maintaining the original image's features and generating new variations.
  • 🎨 Use blend mode to fine-tune the result, keeping the desired elements of the original image while applying the new face.
  • 🔄 Experiment with different settings and denoise strength levels to achieve the most accurate and desired outcome.
  • 🖼️ Send the final image to the unified canvas for further adjustments and refinements.
  • 📌 Remember to hide unnecessary elements (e.g., using Shift+C and Shift+H) to focus on the final desired image.

Q & A

  • What is the first step in installing models from

    -The first step is to visit the website and navigate to the 'models' section at the top of the page.

  • How can you determine the required version for a model?

    -By hovering your cursor over the yellow circle icon next to the model, it will display the necessary version information.

  • What is the purpose of the IP adapter plus face model?

    -The IP adapter plus face model allows users to input a face which is then used as a condition for generating a similar face in the image.

  • How do you add a model to the model manager?

    -You copy the model information from the website, navigate to the model manager (represented by a cube icon), go to 'import models', and paste the copied text.

  • What is the significance of the denoise strength setting?

    -The denoise strength setting affects how closely the generated image resembles the original. A higher setting will result in more variations and less similarity, while a lower setting will try to maintain the original image's features more closely.

  • How can you adjust the settings to keep the background and other elements of the image intact?

    -By adjusting the blend mode and denoise strength settings, you can control how much the background and other elements of the image are altered in the generated output.

  • What are the recommended proportions for the images used in the IP adapter face model?

    -The recommended proportions are not strict, but maintaining a consistent aspect ratio, such as 512 by 512, can help in achieving better results.

  • How can you view the progress of a model installation?

    -While the UI might not show the progress, you can open the terminal to monitor the installation process and see how long it will take to finish.

  • What is the benefit of syncing models in the settings?

    -Syncing models in the settings ensures that all models are updated and consistent across different projects, although it's unclear if this applies to the IP adapter models.

  • How do you fine-tune the face replacement process in the IP adapter?

    -You can fine-tune the face replacement by adjusting the blend mode, denoise strength, and manually painting over the face area in the unified canvas to ensure accuracy.

  • What is the process for using the generated image in another application?

    -After finalizing the adjustments and being satisfied with the generated image, you can export or save the image and then import it into another application for further editing or use.



📦 Installing Models from

The paragraph outlines the process of installing models from the website It begins with navigating the site to find a list of available models categorized under Concepts, Styles, and Tools. The speaker emphasizes the importance of having the correct version of the software, as older versions may not be compatible. The user is guided to download the necessary encoder and provides an example of selecting an IP adapter model called 'face plus IP adapter'. The instructions continue with copying the model's URL and importing it into the model manager, which is accessed through a cube icon. The paragraph also touches on syncing models in settings for convenience and concludes with loading images for use with the IP adapter.


🖼️ Using the IP Adapter with Images

This paragraph details the application of the IP adapter with images. The user is shown how to adjust settings such as skin tone and blend mode to achieve a desired outcome. The process involves experimenting with different denoise strength levels to find the right balance between maintaining the original image's details and generating variations. The speaker also discusses the importance of keeping the background and other elements consistent while focusing on the face. The paragraph concludes with the user making fine-tuned adjustments to achieve a more accurate facial representation.


🎨 Fine-Tuning the IP Adapter for Accuracy

The final paragraph focuses on fine-tuning the IP adapter for increased accuracy. The user is guided through the process of adjusting the image settings, such as the number of generated images and the blend mode, to refine the output. The speaker demonstrates how to use the unified canvas for making specific changes to the face while keeping the rest of the image intact. The paragraph emphasizes the iterative nature of the process, encouraging users to experiment with different settings to achieve the best results. The user is shown how to finalize the image by hiding unnecessary elements and selecting the most satisfactory outcome.




The term '' refers to a website that hosts a variety of models, which are essentially software representations or algorithms designed for specific tasks. In the context of the video, these models are used for image processing and generation. The website is mentioned as the source for downloading and installing models such as IP adapters, which are used to modify or generate images based on certain conditions or inputs.

💡IP adapter

An 'IP adapter' in this context is a type of model used for image processing. It takes an input, such as a face, and processes it to generate an image with specific characteristics or styles. The IP adapter is designed to adapt the input to the desired output, often used in tasks like face swapping or image stylization. The video demonstrates how to use an IP adapter to change the face in an image while keeping the background and other elements intact.


An 'encoder' is a software tool or model that translates data from one format into another, often for the purpose of compression or efficient storage and transmission. In the video, the encoder is a necessary component for the proper functioning of the models downloaded from It ensures that the models are compatible with the user's system and can be utilized effectively for image processing tasks.

💡model manager

The 'model manager' is a user interface or application feature that allows users to manage, organize, and import models for use in various tasks. In the video, the model manager is accessed by clicking on a cube-like icon and is used to import the downloaded models from It serves as a central hub for handling the different models and their settings.

💡denoise strength

Denoise strength is a parameter used in image processing models to control the level of noise or variation introduced into the generated image. A higher denoise strength value results in more significant alterations to the image, potentially moving away from the original input. Conversely, a lower value will attempt to preserve more of the original input's characteristics. In the context of the video, adjusting denoise strength allows the user to control the degree of similarity between the generated face and the input face.

💡blend mode

Blend mode refers to the way in which layers or elements are combined in image editing software. It determines how the colors and tones of one layer interact with those of another layer beneath it. In the video, the blend mode is used to adjust how the generated face integrates with the rest of the image, ensuring a natural and seamless transition between the face and the original image's background.

💡image to image

The term 'image to image' describes a process in which one image is transformed or translated into another image, often with specific modifications or alterations. This can involve changing certain elements within the image, such as the face, while keeping others intact. In the video, 'image to image' is used to describe the task of replacing a face in an image with another face, using the IP adapter model.


A 'prompt' in the context of image generation models is an input or instruction that guides the model in creating an output. It can be a text description, a sample image, or any other form of data that provides the model with the necessary information to generate the desired image. In the video, a prompt is used in conjunction with the image to image process to ensure that the generated image aligns with the user's specifications.


In the context of the video, 'settings' refer to the adjustable parameters within the model manager or image editing software that allow users to customize the behavior and output of the models. These settings can include options like denoise strength, blend mode, and image resolution, which can be tweaked to achieve the desired results in image processing tasks.


Variations in this context refer to the different outputs or results produced by the image processing models when different inputs or parameters are used. By adjusting settings like denoise strength, users can generate a range of images with varying degrees of similarity to the original input or prompt. These variations allow users to explore different possibilities and select the most suitable outcome for their needs.

💡unified canvas

The 'unified canvas' is a term used in digital art and image editing to describe a workspace where multiple elements or layers are combined into a single cohesive image. In the video, the unified canvas is used to integrate the processed image, such as the one with the swapped face, into a larger project or to make further adjustments to the image.


The process of installing models from is described.

Users can browse models categorized under Concepts, Styles, and Tools.

The importance of having the correct version for the models is emphasized.

The IP adapter plus face model is recommended for generating similar faces.

Instructions on how to add a model using the website interface are provided.

The model manager's functionality for importing models is discussed.

Details on how to sync models through settings are mentioned.

The process of loading images with specific dimensions for use with the IP adapter is outlined.

The use of prompts and settings for image-to-image functionality is explained.

Adjusting denoise strings affects the variation in generated images.

A method for fine-tuning the IP adapter to match a specific face is described.

The impact of blend mode on the accuracy of the generated face is discussed.

Instructions on how to send an image to the unified canvas for further adjustments.

The process of changing the face in an image while keeping other elements intact is detailed.

The significance of adjusting image generation settings for optimal results is highlighted.

A step-by-step guide on achieving accurate facial generation using the IP adapter is provided.