Late Autumn Street: Scarf and White Breath

A late-autumn street: a grey scarf up to her chin, white breath dissolving in the cold air, the camera following quietly. 12s landscape, 720P.

Watch on YouTube

Original Video Prompt

EN

Original prompt (verbatim)

[SPEC] 12 seconds, 16:9 landscape, 24fps, 1080p, cinematic street portrait.

[CHARACTER LOCK] A clearly adult 24-year-old East Asian woman with distinctly East Asian features: naturally pretty, long loose straight black hair, soft polished everyday makeup; a dark brown coat, straight-leg jeans, ankle boots, a chunky oatmeal-grey knit scarf looped at her neck and a canvas tote. Keep the same face, outfit, hairstyle and body proportions for the entire clip.

[SETTING & TIME] An old city street at early-winter dusk, plane-tree leaves piled along the curb, warm yellow light spilling from small shops, damp reflections from the previous night on the pavement. Cold enough that her breath is clearly visible.

[SHOT LIST]
0-3.2s: medium tracking shot from behind and to the side, she walks along the sidewalk with hands in her coat pockets, one corner of the scarf lifting in the wind.
3.2-6.4s: cut to the front, camera slowly walking backward with her. Head lowered, she exhales a puff of white breath, looks up at the lens, calm and relaxed.
6.4-8.8s: close-up, she pulls the scarf up over her chin, fingertips resting briefly on the knit texture.
8.8-12s: she passes under a streetlamp, the light sweeping over her head and lighting her hair and the scarf edge; the camera stops and she walks out of frame.

[TEXTURE] Real handheld, shallow depth of field, mixed cool ambient and warm streetlamp light, slight grain; breath must dissipate naturally, not like a smoke effect.

[NEGATIVE PROMPT] influencer-template face, Western/Caucasian bone structure, non-Asian face, over-smoothed skin, plastic skin, breath looking like smoke, scarf clipping, deformed fingers, pedestrians clipping, drifting features, saturated grading, subtitles, watermark, text.

The badge above shows the language of the original prompt. Models can interpret languages differently, so the same idea may render differently if you write it in another language.

Copy this prompt and generate your own video, free and online

Generation settings

Task type
Simple video
Model
agnes-video-2.5-flash
Duration
12s
Resolution
1280x720
Prompt language
EN
Published
2026-11-09

Made something with Agnes? Share it

Post your video and the prompt behind it on GitHub Discussions. Your video stays on your own channel — we only feature it here (credited, with a link back) and may add it to the official Community Showcase playlist.

Submit on GitHub Discussions

Submissions are curated by hand. Nothing is re-uploaded to our channel.

More video prompt ideas