返回案例库

品牌广告与电影感 · MiniMax H3 (Hailuo 03)

Telescope Peek At MiniMax Fashion

A vintage telescope view reveals MiniMax in a stylish, Wes Anderson-inspired sequence with subtle text overlays.

0 次阅读2026/08/14 发布按照提示词中的镜头与时间轴执行
生成结果来源:AtlasCloudAI / MiniMax 官方展示 · CC BY 4.0已保存到本站本地媒体库

完整提示词

Image 1 to Image 4 continuous keyframes, simulating an old telescope finding the MINIMAX device. Start with out-of-focus handheld shake then quickly rack focus to Image 1, transitions between images via telescope panning (whip pan, motion blur, optical trailing, exposure flicker) switching at the most blurred point and quickly stabilizing to refocus. Maintain a fixed dual circular lens mask throughout (black feathered vignette, position, size, edge blur absolutely consistent without distortion or drift) with only internal image movement. In Image 2, fabric flutters gently in the wind, MINIMAX text undulates with the fabric surface remaining legible; Image 3 features a person walking naturally like an unintentionally captured fashionable passerby, turning and swinging arms; Image 4 shows the person adjusting glasses or slightly lifting chin, displaying a cool, relaxed fashion campaign attitude. Red text loads subtly with focus, initially slightly out-of-focus with low opacity, fading in to become clear within 0.3–0.5 seconds (allow minor vertical slide-in or letter-spacing breathing), fading out before switching or being taken away by motion blur. No rotation, bouncing, or large fly-ins/fly-outs. Wes Anderson-style peeping 35mm film texture (slight grain, soft highlight bloom, restrained colors, red text accents), minimalist, sophisticated, slightly quirky and lively. Strictly prohibit adding new people, vehicles, buildings, or logos. Maintain the core composition of the original images and the MINIMAX device unchanged.

为什么这样写

AtlasCloudAI 收录的 MiniMax H3 Brand Film & Cinematic案例。建议先保留主体、参考素材顺序、镜头运动与声音结构,再替换角色、产品或环境变量。

模型与出处

使用模型

MiniMax H3 (Hailuo 03)

提供方

MiniMax / Hailuo AI

AtlasCloudAI / MiniMax 官方展示 · CC BY 4.0

提示词是依据官方公开指南与示例整理的学习版本;不同模型结果会随版本、输入素材、随机种子和参数变化。