Fitness Girl Vlog: Hip Abduction + RDL
Duration: 15秒 Aspect ratio: 9:16 人物参考: 图片1 Audio: 节奏感健身音乐、器械轻响、杠铃落位声;无对白、无字幕 Overall style: 真实手机拍摄的高级健身Vlog,快速卡点剪辑,轻微手持,健身房自然窗光 人物一致性: 全程只有一名成年女主。五官、发型和服装完全沿用@图片1,不额外改变。身体呈专业健美女性体态…
オープンソース の素晴らしい動画プロンプト集 · 5,900件収録
CAMERA: Single unbroken handheld boyfriend vlog take throughout, 30 seconds total. The camera feels like a real boyfriend holding a small mirrorless camera or phone, not a professional production. Natural handheld movement, imperfect framing, occasional camera shake, spontaneous reactions. Realistic autofocus adjustments, slight motion blur and natural exposure changes.
CHARACTER: Use the woman from the reference image as the main character. Maintain her exact facial identity, hairstyle, facial features, body proportions and overall appearance throughout the entire video. She must remain the same person in every shot.
Most importantly, she doesn't perform for the camera.
She sometimes notices it. Sometimes forgets it's there. Sometimes looks back. Sometimes just keeps walking.
That's what makes the footage feel personal.
STORY:
0–5s - Morning at a small Tokyo apartment. The boyfriend casually walks into the room while she fixes her hair and gets ready for the day. She notices the camera, smiles, laughs and playfully tells him to stop filming. Nothing important happens. That's intentional.
5–10s - Walking through Tokyo.
She leaves the apartment and walks through a quiet neighborhood. Shops are opening, bicycles pass, locals go about their morning.
She stops at a convenience store and asks the person behind the camera which drink or snack she should get.
10–18s - Local food.
They walk through a small alley into a cozy restaurant. She orders ramen, picks up the chopsticks and takes a bite.
The food is hotter than expected.
She reacts naturally and laughs while the boyfriend laughs behind the camera.
18–25s - Afternoon.
They explore a lively neighborhood. She looks through small shops, takes photos and occasionally turns back toward the camera.
The camera isn't always focused on her face.
Sometimes it's her hands. Sometimes the street. Sometimes the people passing by.
25–30s - Night.
Tokyo lights come on.
She walks slightly ahead, turns back and smiles.
They get on a train. She sits beside the window watching the city lights pass outside.
The camera slowly moves closer.
Cut. WHY THIS WORKS The prompt doesn't tell the model to create beautiful cinematic shots.
It tells the model to behave like someone who happens to be filming his girlfriend.
That's a completely different instruction.
The imperfect framing matters.
A professional camera operator would constantly correct the composition. A boyfriend wouldn't.
The pauses matter too.
Real people don't react perfectly on cue. They hesitate, look away, laugh at the wrong moment and sometimes forget they're being filmed.
And the camera shouldn't feel invisible.
Small shakes. Autofocus hunting. Exposure adjustments. Slight motion blur.
Those imperfections are what sell the realism.
The biggest instruction might actually be:
She does not pose for the camera
Because the moment the subject starts performing, the footage stops feeling like a memory and starts feeling like an advertisement.
AI video is very good at making things look cinematic.
The harder test is making 30 seconds of completely ordinary life feel like someone actually lived it.