Use the uploaded video as the base motion reference. Use the baby as the identity reference for the girl in the video, and use the pet as the identity reference for the cat in the video.
Replace the two characters in the original video with the two uploaded characters while keeping everything else the same:
the baby replaces the girln only
the pet replaces the cat only
preserve each person’s recognizable facial features, hairstyle, skin tone, age appearance, and overall identity
keep the original body positions, movements, timing, camera angle, scene composition, and background
keep the same actions, gestures, expressions, and interaction between the two characters
keep the same outfit style if possible, but adapt the clothing naturally to fit the baby and the pet
maintain realistic proportions, natural head movement, and accurate face tracking
ensure the face swap is stable in every frame
avoid face flickering, warping, identity mixing, or facial distortion
do not switch sides
do not merge the two identities
do not change the choreography or motion
do not alter the background or camera movement
Final result: a natural, realistic video where the left character is replaced by the baby and the right character is replaced by the pet, while the original video’s motion and scene remain unchanged.