ComfyUI IPadapter V2 update fix old workflows #comfyui #controlnet #faceswap #reactor

28 Mar 202404:44

TLDRIn this tutorial, the presenter guides users through updating to the latest version of the IP adapter V2, thanking Mato for his creation. The video demonstrates how to update the IP adapter node in Comfy, organize the required models in the correct folder, and adapt existing workflows to the new version. It also showcases an experiment using the IP adapter with the reactor to integrate a character's face into a new image, proving the technique's compatibility with the updated tool.


  • ๐Ÿ”ง Thank Mato, the creator of the IP adapter V2, for his contribution to the community.
  • ๐Ÿ’ป Start by updating the Comfy manager and the IP adapter node for a seamless transition to the new version.
  • ๐Ÿ”— If manual update is needed, a link to the GitHub page is provided in the video description for accessing the required models.
  • ๐Ÿ“‚ Ensure that the models are placed in the correct folder: 'com ui/models/ip adapter' for the new version to recognize them.
  • ๐Ÿš€ Explore the two main nodes available in IP adapter V2: 'IP adapter Advanced' and 'IP adapter Tiled', each with its unique functionalities.
  • ๐Ÿ–ผ๏ธ Understand the difference between the nodes: 'Tiled' is for panoramic images and includes image sharpening, while 'Advanced' is more general.
  • ๐Ÿ”„ Delete the old IP adapter node and reconnect the workflow to the new one to resolve any errors during the update process.
  • ๐ŸŽจ Test the new version's effectiveness by using a previously worked on workflow and adapting it to the updated tools.
  • ๐ŸŒŸ Experiment with integrating character faces into new images using a combination of the IP adapter and reactor for enhanced results.
  • ๐Ÿ“น The video description will be updated with workflows compatible with the new IP adapter V2 for viewer convenience.
  • ๐Ÿค– Encourage viewers to engage with the content, ask questions, and enjoy the learning process.

Q & A

  • What is the main topic of the video?

    -The main topic of the video is about updating and using the new version of the IP adapter, specifically version 2.

  • Who is Mato and what is his contribution mentioned in the video?

    -Mato is the creator of the IP adapter. His contribution is acknowledged in the video as he has made many of the tools and techniques used in the video possible.

  • What is the first step in updating to the new version of the IP adapter?

    -The first step is to update Comfy and enter the manager, then click on 'update comfy UI'.

  • What can be done if the automatic update through the manager doesn't work?

    -If the automatic update doesn't work, one can manually update by visiting the GitHub page provided in the video description and downloading the necessary models.

  • What is the new folder location for the models in version 2 of the IP adapter?

    -In version 2, the models should be located in the folder 'com ui/models/ip adapter'.

  • What are the two main nodes available for the IP adapter in the new version?

    -The two main nodes available are 'IP adapter Advanced' and 'IP adapter Tiled'.

  • What is the difference between 'IP adapter Advanced' and 'IP adapter Tiled'?

    -The main difference is that 'IP adapter Tiled' is used for images that are more panoramic and not square. It also prepares the images for the IP adapter, eliminating the need for a 'prepare image to clip Vision' node.

  • How does one resolve the error encountered when transitioning from the old to the new IP adapter?

    -To resolve the error, one should delete the old IP adapter node and then press 'Q' again to continue working with the new node.

  • What technique was demonstrated in the video using the IP adapter and reactor?

    -The technique demonstrated involves integrating the face of a certain character into a new image using a combination of the IP adapter and the reactor.

  • What is the final result of the experiment conducted in the video?

    -The final result is an image where the character's face is integrated in a way that is very closely related to the reference image, showcasing the effectiveness of the new version of the IP adapter.

  • What does the video speaker encourage viewers to do at the end of the video?

    -The speaker encourages viewers to ask questions, subscribe to the channel, like the video if they enjoyed it, and most importantly, to have fun with the tools and techniques discussed.



๐Ÿ”ง Introduction to Updating and Using IP Adapter V2

The video begins with an introduction to the new version of the IP adapter, expressing gratitude to Mato, the creator of the tool. It emphasizes the importance of Mato's contributions and encourages viewers to check out his channel. The video then outlines the process of updating the Comfy manager and IP adapter node, including manual update instructions through the GitHub page. It also explains the new file structure for the models in the updated version and demonstrates how to organize them properly.



๐Ÿ’กIP Adapter V2

IP Adapter V2 is the latest version of a tool used in the video for image processing. It is an upgrade from the previous version and is designed to improve the efficiency and quality of image-related tasks. In the video, the creator thanks Mato for developing this tool and guides the audience on how to update to this new version, emphasizing its importance in their workflow.


Comfy is a software or platform where users can manage and update their tools, such as the IP Adapter node. In the context of the video, it is used to update the system to ensure compatibility with the new version of the IP Adapter. It is an essential part of the setup process.


GitHub is a web-based hosting service for version control and source code management, often used by developers to store and share their projects. In the video, GitHub is mentioned as a place where viewers can manually update the IP Adapter node if the automatic update through the manager doesn't work.


In the context of the video, models refer to the files or algorithms used by the IP Adapter for processing images. These models are essential for the functionality of the tool and need to be correctly placed in the designated folder for the IP Adapter to work properly.


A workflow in this context refers to a series of steps or procedures followed to complete a task or project, such as image processing using the IP Adapter. The video demonstrates how to adapt previous workflows to the new version of the IP Adapter, ensuring a smooth transition and continued productivity.

๐Ÿ’กK Sampler

The K Sampler is a component or tool used in the video's workflow for image processing. It is connected to the IP Adapter and is part of the process of generating and refining the final image output. The video shows how to reconnect the K Sampler to the updated IP Adapter node.


The Reactor is a tool or technique used in combination with the IP Adapter to achieve specific image processing results, such as integrating a character's face into a new image. It is showcased in the video as a method that continues to work effectively with the updated IP Adapter.

๐Ÿ’กAdvanced and Tiled Nodes

Advanced and Tiled Nodes are options within the IP Adapter V2 that offer different functionalities for image processing. The Advanced node provides general image processing capabilities, while the Tiled node is specifically for panoramic images and includes an option for image sharpening. These nodes are used to tailor the workflow to the specific needs of the project.

๐Ÿ’กExtra Model Paths.yaml

Extra Model Paths.yaml is a configuration file that allows users to specify additional locations for their model files. This is useful when the models are not in the default folder location or when using multiple versions of models. In the video, it is mentioned as a way to link to the old model location if needed.

๐Ÿ’กFace Restoration

Face Restoration is a technique used in image processing to enhance or reconstruct facial features in an image. In the video, it is demonstrated as part of the process to achieve a final result that closely matches the reference image, using the IP Adapter and Reactor in conjunction.


The term 'Community' in this context refers to the group of individuals who share a common interest in the tools and techniques being discussed in the video, such as the IP Adapter. The video creator expresses gratitude to Mato for his contributions to the community, highlighting the collaborative and supportive nature of this group.


Introduction to the new version IP adapter V2

Gratitude expressed to Mato, the creator of the IP adapter

Importance of updating Comfy and IP adapter node through the manager

Manual update instructions provided through a GitHub link

Explanation of the new folder structure for models in V2

Demonstration of loading a workflow from the previous version

Differences between IP adapter Advanced and IP adapter Tiled options

Resolution of an error message by deleting the old IP adapter node

Updating workflows for compatibility with the new IP adapter V2

Experiment showcasing the integration of a character's face into a new image

Description of a simple workflow adapted to LCM

Step-by-step guide to building the IP adapter with the new version

Organization of the workflow by grouping elements for clarity

Second pass technique to refine the image results

Integration of the Reactor for final image enhancement

Final result closely resembling the reference image

Encouragement to update old workflows with the new IP adapter V2

Closing remarks with thanks to Mato and an invitation for further engagement