This tool will help you to use AI to create deepfakes any photo/video with the help of Google Collab.

I’ve tried to make it as simple and accessible as possible, requiring just 7 simple steps and you can do this on any platform with access to a modern web browser.

1. A google account
2. Access to an undisrupted internet connection
3. A source image
4. A target video (the lesser things over the face the better)
5. Patience

1. Open the Collab link in a new tab and connect it with Google’s T4 GPU

2. Start the step-1 of the project, you will see a little green check mark after it’s done with the tasks. This one might take a while.

3. Start step-2 of the process. This one is rather faster and would be done in under 2 minutes, after which you’ll see a green check mark again.

4. Click on the menu icon on the top left and click on “Show File Browser”.

5. Upload both, your input image and target image/ video here, you’re free to rename it at this point.

6. In the step-3 command, rename “vid.mov” to your target video (making sure if capitalisation and file type) and “clown.png” to your input picture (or don’t, you’ll see the video with Pennywise if not :P)

-> you can edit the arguments to reduce/ increase the workload like adding “– skip audio”(that doesn’t process audio) or removing “– many faces”(when your video just has one face to work with) among many others. refer to the original ROOP documentation for more.

7. Execute step-3 and wait until it shows a check mark as well. Upon completion, you’ll find your deepfake named “Swapped.mp4” in the file browser, which you can download.