The Paparazzi AI Video Prompt: Generate Hyper-Realistic Bystander-Style Footage

What This Prompt Does

A fully-specified AI video generation prompt for a short, hyper-realistic paparazzi-on-Rodeo-Drive scene — the kind of bystander clip that looks like it was genuinely captured by a random phone on the street, not produced.

Works with AI video generators that accept character reference images. You upload a photo of your character; the video model uses it as the identity anchor and keeps the face consistent throughout. The prompt handles everything else: scene, camera behavior, vehicle, audio design, and second-by-second shot timing.

Global Prompt Rules (Put These First)

promptCopy
No music. Diegetic SFX only.
Raw handheld iPhone footage, auto-everything, bystander POV.
No styled lighting. No cinematic grading.
Auto white balance flickers between warm and cool.
Footage must feel accidental, reactive, unstable.

Character Reference Rule

Add this above the scene description to prevent identity drift:

promptCopy
CHARACTER REFERENCE RULE:
Use the uploaded character sheet as the strict identity reference.
Preserve: facial structure, head shape, hair, skin tone, jawline, expression, body proportion, styling.
Do NOT: add a cap, add glasses, change the hairstyle, make him clean-shaven, change his age.
He is a composed adult male in his 30s.

The Full Prompt

promptCopy
0s:
Camera already unsteady, pointed loosely down Rodeo Drive. Asphalt slightly over-exposed.
Ambient traffic noise, crowd murmurs, footsteps, wind buffeting the mic with low crackle.
Image feels raw and imperfect.

1s:
A deep low-frequency engine rumble rolls in from off-screen left. Phone mic slightly
distorts on the low-end peaks. Camera swings suddenly to track the sound, briefly
catching a blurred pedestrian shoulder in the foreground.

2s:
A matte black supercar slides into frame. Engine stretches into a thick resonant growl.
Autofocus hunts aggressively. Car body goes soft while background stays sharp, then
snaps back onto the low roofline.

3s:
Driver's window fully down. The character from the uploaded reference is visible from
chest up — matching facial structure, hair, skin tone, and outfit exactly. Harsh sun
creates hard shadows. No fill light. Natural paparazzi-style exposure.

4s:
A bystander steps partially into frame left. Calls out: "Excuse me, what do you do
for a living?" Camera briefly autofocuses on bystander before snapping back.

5s:
Driver turns slightly toward the window. Expression stays relaxed. Faint confident
smirk forms. Calm and unbothered while the car rolls forward slowly.

6s:
In a clear, confidently projected voice: "I'm the owner of a [your business type]."
Delivery is casual, precise, calm. No exaggerated acting.

7s:
Camera dips to catch the rear quarter panel and spinning rim. Wheel spokes strobe in
harsh sunlight. Lens flare streaks across the upper corner.

8s:
Autofocus loses the car and locks onto a background storefront. Foreground goes soft.
Focus snaps back. Slight chromatic aberration along high-contrast edges.

10s–15s:
Camera tracks the rear of the car moving away down Rodeo Drive. Exposure drops as the
car passes into shade. Driver silhouette faintly visible through side window. Auto white
balance flickers crossing sun/shade. Footage ends with motion-blurred rear view shrinking
into heat haze. Engine rumble fades into traffic noise and wind crackle.

Customization

Quality Tips

• Use the highest resolution, most frontal character reference photo available
• Generate 3–5 variations and select the best — output varies significantly between runs
• If character appearance drifts, add: 'Maximum identity preservation throughout the full clip'
• For better lip sync, pair the text with an ElevenLabs audio clip of the dialogue
• If autofocus looks too smooth: 'Autofocus overshoots before settling, characteristic of a phone camera catching a fast-moving subject'