European Bus Interior — A JSON Scene Definition

Veo 3Structured PromptRealisticUrbanHandheld Camera10s
A JSON block that even writes out its own model and duration_seconds fields, describing an everyday scene aboard a European bus. An engineering approach that treats an API request body itself as the prompt.
PROMPT · Veo 3 提示词
Create with AI
{ "model": "Veo3Fast", "duration_seconds": 9, "aspect_ratio": "16:9", "prompt": "Interior of a realistic European city bus in daytime. Blue patterned seats and yellow vertical poles. Two human passengers sit mid-front, facing camera, composed and slightly awkward. Several small macaque monkeys are seated like commuters and hold onto poles, with subtle idle motions such as blinking and gentle head turns. Tone is dry comedic but respectful. Lighting is soft daylight from side windows. Start with a rear-aisle view and slowly dolly forward to a medium two-shot of the two humans, keeping the monkeys visible in surrounding seats. Keep movements natural and small. No text on screen.", "camera": { "movement": "slow dolly in from rear aisle to mid-bus two-shot", "focal_length": "35mm", "aperture": "f/2.8", "frame_rate": 24, "motion_strength": "low" }, "style": { "look": "cinematic realism", "color": "neutral with gentle contrast", "grain": "light" }, "audio_notes": { "sfx": [ {"timecode": "00:07.5", "cue": "short bus PA chime"} ], "voiceover": { "language": "nl-NL", "voice": "male driver, calm", "timecode": "00:08.0", "text": "Volgende halte Apendrecht." }, "ambience": "quiet bus engine hum and subtle city noise" }, "constraints": { "no_text_on_screen": true, "no_violence": true, "no_chaos": true }, "negative_prompt": "cartoonish acting, aggressive behavior, jumping or running, shaky cam, low quality, blur, overexposed highlights, exaggerated expressions, text overlays, meme captions" }
✍️ Editor’s Notes

指定「Veo3Fast」这个具体子模型挺少见,说明作者对速度和成本有明确考量,愿意牺牲一部分细节精度换取生成效率。猴子当乘客坐公交这个设定靠「像通勤族一样扶着扶手、偶尔眨眼转头」这种细微动作描述去压住违和感,不追求猴子做什么夸张表演,反而是这种克制让画面显得更真实滑稽。荷兰语报站台词精确到了时间码,还标注了「冷静的男性司机音」,这类语音细节对Veo系列这种自带音画同步能力的模型是刚需信息,少了这些模型很容易配一个风格不搭的声线。

Create with AI
Related Works
×