Videos get downsampled to 800 pixels (on the longest side), 5 second duration, and 12 fps. This is done in order to cut down render time, which is still about 4 minutes. Please ❤️ this Space. tony.assi.media@gmail.com for business inquiries.
Reference Mode: Uses the first frame of the video to determine the face to swap. Choose direction to prioritize which face.
Many Faces Mode: Swaps all faces. Use gender selection to filter which ones.