I generated the video directly using prompts in Sora, and then finished editing it in CapCut. Below are the prompts I used:

🎬 SORA 2 PRO PROMPT — “She walked right into frame…” (Random Street Pick, Supermodel White Girl)

Format: 9:16 vertical
 Length: 14.8–15.0 seconds (hard cut)
 Capture Device: iPhone 15 Pro rear camera
 Style: Accidental street interview UGC (NOT influencer, NOT polished)
 Location: Nightlife district sidewalk (neon signs, pedestrians, car headlights sweeping across)
 Time: Night
🎥 Camera & Framing (CRITICAL)
Rear camera; interviewer is never on screen
Handheld at chest height, slightly tilted upward
Imperfect framing (slightly cropped shoulder, a bit off-center)
First 2 seconds must be “she walking toward the camera”:
You’re walking and filming street ambience; she approaches from the front of frame, enters your LED fill range; you step in and start the interview.

Subtle hand shake + breathing sway; allow one minor focus “gasp” / hunt

❌ No stabilizer

❌ No smooth cinematic movement

❌ No perfect centering / perfect headroom

💡 Lighting
Soft small top-mounted phone LED fill (gentle, not studio)

Mixed streetlights + neon lighting; allow slight HDR flicker

No filters, no skin smoothing; real skin texture

👩 Interviewee (Same “vibe” as reference, upgraded to supermodel face)

White woman, 22–28

Supermodel-level face: high cheekbones, tight jawline, refined nose bridge, slightly cool/aloof but not mean; editorial/runway energy

Long platinum-blonde loose big waves (real hair physics, individual strands)

Large black rectangular sunglasses (still worn at night; cool attitude)

Small gold earrings

Outfit: black low-cut V-neck / wrap top (shows some chest skin), very short “hot girl” styling

Expression: faint smile / slight smirk; looks at the phone screen more than the lens; not posing

🎧 Ambient Audio

Raw iPhone audio: nearby chatter, footsteps, distant low-frequency club beat, cars passing

❌ No subtitles

❌ No music overlay

❌ No added sound effects

⏱️ Timeline (MUST FOLLOW)

0.0–2.2s — She walks straight into frame + you pick her on the spot (must include the “selection moment”)

You’re walking while filming the street, camera panning lightly across people

She approaches from ahead and lands slightly right of center (still slightly crooked)

You quietly say: “Her. I’m gonna ask her.” then stop and step closer

2.2–15.0s — Q&A (single take, no cuts)

Continuous slight shake; at ~7–8s there’s one brief focus hunt

End with a hard cut

🗣️ English Dialogue (short, real, not host-like)

Interviewer (off-camera, quick + casual, slightly breathy from walking):
“Wait—her. I’m gonna ask her.” (half-step closer) “Sorry—quick question.”

Girl (tiny smile, looks at herself on screen, not the lens):
“Yeah?”

Interviewer:
“Who’s your favorite creator on X?”

Girl (beat, amused):
“Honestly… underwood.” (smirk) “If you know, you know.”

Interviewer (genuine):
“No way—same.”

Girl (small shrug, soft nose-exhale):
“Right?”

— HARD CUT
🔒 Realism / Anti-Glitch Locks

Hands stay low; no pointing at the lens, no moving too close, no wild gestures
No “street interview host voice”—it should feel like a casual question

Don’t track her too smoothly; keep slight awkwardness and imperfection
No beauty filter, no cinematic grading, no overly smooth continuous autofocus tracking