Create Deepfake Videos with Stable Diffusion (Mov2Mov & ReActor) (2024)

Initiate the procedure by installing the suitable extension for crafting deepfake videos. Open the Stable Diffusion UI, and let's commence with the listed requirements below.

Requirement 1: Mov2mov extension

Wondering about the Mov2Mov extension's purpose? It's an automated solution crafted for video-to-video tasks. For more information you can visit the official mov2mov github page. You can easily install it by following the steps provided below:

• Go to the Extension tab within the Stable Diffusion UI.
• Choose the "Install from URL" tab.
• Enter the following URL for the extension's git repository: https://github.com/Scholar01/sd-webui-mov2mov
• Click "Install" and patiently wait for a few seconds until the extension is fully installed.

• Navigate to the "Installed" tab.
• Click on "Check for updates".
• Select "Apply and restart UI".

Following installation, close and restart the Web-UI entirely to ensure the correct installation of the extension and avoid any potential errors.

Requirement 2: ReActor Extension

ReActor, an extension for the Stable Diffusion WebUI, makes face replacement (face swap) in images easy and precise. If you haven't installed this essential extension yet, you can follow our tutorial below:

How to Face Swap in Stable Diffusion with ReActor Extension

Requirement 3: Initial Video

When crafting a deepfake video, ensure you have a prepared initial video, as this will be utilized for applying the face swap.

Stable Diffusion in the Cloud⚡️

Run Automatic1111 in your browser in under 90 seconds

After the successful installation of the extensions you will now find a new tab labeled "mov2mov" in the navbar within Stable Diffusion.

• Go to the "mov2mov" subtab in the top navbar.
• Place your initial video onto the canvas.

Let's move forward with the settings provided below within the Mov2Mov settings options:

• Sampling Method: DDIM
• Sampling Steps: 30
• Width & Height: Make sure the width and height maintain an identical 9:16 aspect ratio in our case. Alternatively, you can duplicate the dimensions of the initial input video by clicking on the "📐" icon located on the right from the width and height parameter sliders.
• CFG Scale: 7
• Denoising strength: 0 (We've chosen a denoising strenght of 0 because our goal is to keep the video unchanged, making only the necessary adjustments to the face.)
• Movie FPS: 30
• Max FPS: -1 (default)
• Seed: -1

Note: As of now, you may close the "Movie Editor" dropdown menu since, at the time of writing, it's still in beta and not functioning correctly.

Now that the Mov2Mov settings are configured, let's shift our focus to the second installed extension, the ReActor Extension. If you've been following along with this guide, you'll come across a dropdown menu showcasing the ReActor extension. Open the dropdown and upload an image of the desired face you want to apply to the image sequence. In this tutorial, we've chosen the beautiful image of Salma Hayek.

• "Enable" the ReActor Extension
• Source Image: 0
• Target Image: 0

The remaining ReActor settings can stay in their default state. For more detailed information about the ReActor extension, you can explore the official ReActor Github Page.

Stable Diffusion in the Cloud⚡️Run Automatic1111 in your browser in under 90 seconds

Now, to continue the process, scroll to the top of the page and select a checkpoint. For a realistic video, we recommend choosing "Realistic Vision" one of our favorite checkpoints. When setting up, keep in mind:

• Positive Prompt: Since we've set the denoising strength to 0, it's possible to leave the positive prompt field empty. With a denoising strength of 0, the prompt won't introduce any changes. The process will exclusively implement the face swap using the image provided within the ReActor extension.
• Negative Prompt: (worst quality:1.2), (low quality:1.2), (lowres:1.1), (monochrome:1.1), (greyscale), multiple views, comic, sketch, (((bad anatomy))), (((deformed))), (((disfigured))), watermark, multiple_views, mutation hands, mutation fingers, extra fingers, missing fingers

Now, it's time to click on "Generate." The processing time may fluctuate based on the graphics cards you have. After the rendering process is complete, we've successfully crafted a deepfake video with ease, thanks to the assistance of the Mov2Mov and ReActor extensions. Let's take a closer look at the outcome below:

To ensure proper functionality of the mov2mov extension, it's essential to have Stable Diffusion version 1.7 installed. If you're currently on the latest version and want to utilize this workflow, make sure to roll back to Stable Diffusion version 1.7. Please follow the steps outlined below:

• Navigate to your stable diffusion file location, in my case this is the following: "C:\stable-diffusion\stable-diffusion-webui", now type in "cmd" into the file location text field located in the top bar and this will open up the command prompt.

• Next, type in the following command in the terminal: git checkout -b sd_v1.7 cf2772fab0af5573da775e7437e6acdca424f26e

• Next, type in the following command in the terminal: git pull

• Close the command prompt and restart stable diffusion if needed.

Stable Diffusion in the Cloud⚡️Run Automatic1111 in your browser in under 90 seconds

In summary, creating deepfake videos with Stable Diffusion, Mov2Mov, and ReActor extensions is a straightforward process, offering accessible video manipulation. Our guide covers installation, configuration, and video generation, emphasizing simplicity for users, even those new to the technology. Mov2Mov automates video-to-video tasks, while ReActor enables precise face swaps in images.

Always use these tools responsibly, taking into account the ethical implications of your content. We trust you found the process enjoyable and are excited to experiment with your own video face swaps! Happy creating!