Three Cities, One Take · Sleeve and Palm Wipe Transitions

MiniMax H3TravelTransitionUrbanHandheld CameraRealisticVlogStructured PromptLip SyncText to Video15s
Fifteen seconds to cover Seoul, Hanoi and Marrakech, and the trick isn't editing — it's occlusion: a sleeve wipes across the lens, a passing cyclo cuts the frame, a palm covers the camera, and each blackout swaps the city. The prompt writes out three 2.5-second beats and their exact transition triggers, while locking the same face and hairstyle throughout.

PROMPT · Video Prompt

Create with AI
Create a 15-second ultra-realistic cinematic travel vlog filmed entirely as natural handheld smartphone footage. Use the same young woman throughout all three locations, maintaining identical face, facial proportions, skin tone, long dark-brown hair, hairstyle, and natural expressions. No beauty filters, no subtitles, no music. Fast but smooth transitions created by objects covering the camera lens. 0:00–0:02.5 — SEOUL, SOUTH KOREA Cold winter morning with light snow falling. The woman walks through a busy Seoul street, filming herself in selfie mode while holding a takeaway coffee. Seoul Tower is visible in the background. She smiles naturally and says: “Seoul mornings hit different!” Realistic traffic, pedestrians, footsteps, wind and soft city ambience. 0:02.5–0:05 — TRANSITION She turns the phone away from herself, revealing the Seoul street and N Seoul Tower. She raises her sleeve toward the camera until the fabric completely covers the lens, creating a natural seamless transition. 0:05–0:07.5 — HANOI, VIETNAM The sleeve reveals a completely new location: lively Hanoi street filled with motorbikes, Vietnamese flags, street vendors and old buildings. The woman is now wearing a traditional Vietnamese conical hat and casual light clothing. She holds a fresh bánh mì close to the camera, takes a bite and says naturally: “Best sandwich ever!” 0:07.5–0:10 — TRANSITION She quickly tilts the camera upward toward colorful old buildings, tangled electrical wires and Vietnamese flags. A passing cyclo moves directly across the lens, completely blocking the frame and creating the transition. 0:10–0:12.5 — MARRAKECH, MOROCCO The cyclo clears to reveal a warm, atmospheric Marrakech souk. Golden lanterns, colorful textiles, narrow market alleys and traditional Moroccan architecture surround her. She holds a glass of mint tea toward the camera, smiles and says: “This place is magical.” Rich warm lighting and authentic market ambience. 0:12.5–0:15 — FINAL TRANSITION / ENDING She turns the camera around and begins walking deeper through the crowded souk. She looks back toward the camera, reaches her hand forward, and places her palm directly over the lens until the screen becomes completely covered. Cut to black. Visual style: photorealistic live-action, authentic travel vlog, natural smartphone camera movement, realistic skin texture, natural lighting, documentary realism, subtle handheld shake, genuine facial expressions, believable environments, fast-paced travel montage. Audio: natural location sounds only — Seoul footsteps, traffic, light snow and wind; Hanoi motorbikes, vendors and street voices; Marrakech market chatter, lantern ambience and distant calls. No music, no subtitles, no artificial sound effects.

✍️ Editor’s Notes

The core trick here is writing transitions as physical occlusion instead of editing instructions — a sleeve wipes across the lens, a rickshaw cuts through frame, a hand covers the camera. The model doesn't need to understand editing terminology at all, it just needs to generate one natural blackout following an object's motion. Each of the three cities gets exactly one line of dialogue plus one prop locked down — coffee, a baguette, mint tea — using the held object as the beat marker, while local ambient sound gets its own separate Audio section — sound cues lock down a sense of place better than visual description does. The ending has a hand come into frame directly in place of a “cut to black” instruction, wrapping things up clean.

Create with AI

Related Works

׋›
↑