OpenZhenZhen AI Prompt Gallery · MiniMax H3 · video

Golden Hour Through the Skyscraper Canyon

Golden Hour Through the Skyscraper Canyon — explore this MiniMax H3 creation on OpenZhenZhen, with its original prompt and a reusable generation example.

Golden Hour Through the Skyscraper Canyon — AI video creation
Golden Hour Through the Skyscraper Canyon

Prompt

Chinese prompt

20秒、16:9、写实电影感的第一人称FPV飞行:黄金时刻、晴朗天空中穿越由玻璃与钢铁摩天楼组成的密集现代城市。全程保持一条连续路径,不剪辑、不出现第三人称视角、不显示无人机机身、无HUD或文字;自然镜头响应,避免鱼眼畸变。0–3秒:在屋顶高度沿两座反光玻璃塔之间向前巡航,低角度暖阳在立面上形成长长的金色反射条纹,城市在下方远处展开。3–13秒:加速进入平滑且受控的自由式峡谷飞行,先绕塔角向左急转,再向右快速穿过狭窄玻璃建筑间隙,随后俯冲并在狭窄的垂直街道峡谷中穿梭。每次转弯都伴随自然倾斜,玻璃反射、钢铁边缘、耀斑和方向性运动模糊从镜头旁掠过,速度逐渐提升至高速。13–17秒:继续在密集建筑间进行快速而精准的峡谷飞行,保持居中的前向视角和连贯运动。17–20秒:从峡谷中用力向上拉升至城市天际线之上,展现延伸至地平线的完整城市;最后几秒稳定为高空滑行,暖色晴空开阔,极远处仅有轻微大气雾霭。

English prompt

20-second 16:9 photoreal cinematic FPV flight through a dense modern city of glass-and-steel skyscrapers at golden hour under a clear sky. One continuous first-person path, no cuts, no third-person view, no visible drone, no HUD or text, natural lens response without fisheye distortion. 0–3s: Cruise forward at rooftop height between reflective glass towers; warm low sunlight creates long golden streaks on the façades, with the city far below. 3–13s: Accelerate into a smooth controlled freestyle canyon run. Whip-turn left around a tower corner, snap right through a tight gap, then dive and weave down a narrow vertical street canyon. Bank and lean naturally with each turn; glass reflections, steel edges, sun flare, and directional motion blur stream past while the speed rises toward a fast run. 13–17s: Continue the fast, precise canyon flight through dense buildings, maintaining a centered forward view and coherent motion. 17–20s: Pull hard upward out of the canyon above the skyline, revealing the full city stretching to the horizon. Settle into a stable high-altitude glide for the final seconds, with warm clear sky and only slight haze in the extreme distance.

Topics and styles

View API model

Explore more AI image and video prompts · Try image generation · Try video generation · Create on the AI canvas