Try a prompt such as:
“Nighttime car interior filmed from the dashboard or windshield.
Two characters sit in the front seats. The main character sings
passionately with exaggerated facial expressions, clear mouth movement,
gentle body swaying, occasional closed eyes, head movement, and expressive
hand or paw gestures. The second character slowly looks over with an
unimpressed, confused side-eye. Start with a two-person medium shot,
cut to the reaction close-up, then the singer close-up, and return to
the two-person shot. Natural cabin lighting, subtle handheld movement,
realistic candid comedy movie feel.”