Video Face Swaps: Roop & NextView Extension (Stable Diffusion)

Are you interested in effortlessly creating deepfake video face swaps using stable diffusion technology? You've come to the right place! Our tutorial guides you through the straightforward process of achieving this with ease.

This guide will assist you in navigating and installing the necessary extensions for achieving stunning video face swaps. Don't hesitate any longer, let's get started right away!

Prior to diving into the creation of remarkable video face swaps, it's essential to have specific extensions installed. The following extensions are required:

Requirement 1: Roop Extension

If you haven't installed the Roop extension yet, there's no need to worry. We've provided a detailed guide below on how to install and utilize the Roop extension:

How to Face Swap in Stable Diffusion with Roop Extension

Requirement 2: NextView Extension

If you want to efficiently transform an original video into an image sequence and subsequently convert the face-swapped images back into a video, all without exiting the stable diffusion interface, the NextView extension is your ideal choice. More information can be found on the NextView Github Page.

• Open your Stable Diffusion Web UI.
• Click on the "Extensions" tab.
• Navigate to the "Install from URL" subsection.
• Paste the following URL: https://github.com/NextDiffusion/next-view
• Click on the "Install" button.
• After installation go to the "Installed" tab and click on "Apply and restart UI".

Note: Ensure that ffmpeg is installed on your system. Uncertain about the installation process for ffmpeg on your device? We suggest referring to this tutorial for guidance on ffmpeg installation.

Requirement 3: Initial Video

Make sure you possess an initial video to initiate the face swap process. In this tutorial, we'll be utilizing the video provided below for our demonstration, this video is downloadable, allowing you to access it and follow the tutorial step by step.

Before we proceed, it's crucial to ensure that specific global settings are correctly configured. These global stable diffusion settings play a vital role in guaranteeing the accuracy of the names for the face-swapped image sequence. This precision is essential for the seamless utilization of the image sequence in the NextView Extension. Furthermore, it's imperative to have the image sequence in PNG format for optimal results. These measures are pivotal in the creation of a stunning video.

• Navigate within Stable Diffusion to the "Settings" tab.
• Click on "Saving images/grid" in the left sidebar.
• Confirm that the "File format for images" is set to png.
• Ensure the "Images filename pattern" field is left empty.
• Make sure the option for "Adding a number to the filename when saving" is unchecked.

After configuring the settings correctly, select "Apply settings," followed by clicking on "Reload UI."

Once all the specified requirements are successfully installed, the procedure will commence by accessing the NextView Extension.

• Navigate to the NextView Extension within Stable Diffusion.
• Place the video onto the canvas within the NextView Extension.
• Click on the "Generate Image Sequence" button.

After the process is complete, you will be given a file location containing all the generated images extracted from the video. You can now proceed to copy this file location, which will enable you to utilize the image sequence for batch processing within the Img2Img tab.

After copying the file location containing the image sequence extracted from the video file using the NextView Extension, you can move on to the Img2Img "batch" tab.

Img2img batch settings

• Copy the file location containing the image sequence extracted from the video file using the NextView Extension.
• Navigate to the Img2Img tab.
• In the Img2Img "batch" subtab, paste the file location into the "input directory" field.
• Choose a suitable location for the "output directory." (This is where all the batch face-swapped images will be stored)

Note: The location you entered in the "output directory" will be used later to convert the images into a video with the NextView Extension.

Global Img2Img settings:

• Checkpoint: Realistic Vision
• Sampling Method: Euler a
• Sampling Steps: 30
• Denoising Strength: 0 (We've set denoising to 0 because our intention is to preserve the image sequence as is, without altering anything except for the face)
• Seed: -1
• Width & Height: Make sure that the video ratio aligns with the image sequence you supplied earlier, in our case, this ratio is 9:16.

Now that we have these settings in place, we can transition to the following stage, which entails employing the Roop extension for face swapping. Hopefully, you already have a fantastic face in mind for the face swap! Let's move forward.

If you've been following this tutorial, you'll find a dropdown menu featuring the Roop extension. Open this dropdown and upload an image of the beautiful face you'd like to apply to the image sequence. For the sake of this tutorial, we've selected the stunning image of Salma Hayek as our choice. And, of course, remember to "Enable" the Roop extension as shown below:

Now, click the "Generate" button and patiently await as your image sequence goes through the face-swapping process using the face image you've provided on the Roop extension canvas.

Once everything has finished rendering, proceed by copying the output directory file location where the face-swapped images are stored.

• Copy the output directory location.
• Access the "NextView" extension in the top navbar within Stable Diffusion.
• In the NextView Extension, go to the "Image Sequence 2 Video" tab.
• Paste the file location into the "Image Sequence Location" text field.
• Click "Generate Video" to convert the face-swapped image sequence into a video.

Note: The output location of the video will be the following: "stable-diffusion-webui>extensions>next-view>output_videos"

If you've followed this tutorial accurately, you should have successfully created a face-swapped video featuring the beauty of Salma Hayek, as shown below.

After following this tutorial, you should now have created an impressive face-swapped video, as illustrated in our example showcasing the beauty of Salma Hayek. We trust that this guide has provided you with valuable insights and assistance in your exploration of the potential of deepfake technology. Embrace your creative ventures and enjoy the journey!