返回案例库

音乐与节奏 · MiniMax H3 (Hailuo 03)

MiniMax H3 案例 061 · @ivy432hz

How the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.…

0 次阅读2026/08/07 发布按提示词中的镜头与时间轴执行
生成结果来源:Cheerself AI 收录 · @ivy432hz X 原帖使用来源站预览

完整提示词

How the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.39-second mark of the target video; Picture 3 (from Shot 1) aligns with the 9.10-second mark of the target video. integrated_multimodal_description: [Shot 1] 16:9 digital-animation music video, contemporary mixed-media collage with anime illustration, watercolor paper, editorial cutouts, Xerox grain, halftone dots, CMYK misregistration, scanner offsets, nested mattes, and controlled psychedelic instability. Collage is active inside the image at all times, not only during transitions: maintain one dominant subject plus two to five simultaneous cutout, window, reflection, miniature, or texture states in the same frame. Begin with the exact full-body composition, cream negative space, black suit, purple ponytail, gold "AI" hair clip, bouquet, shadows, and proportions of <Picture 1>. From 00:00.000 to 00:04.900 the camera performs an almost imperceptible push in while her cast shadow moves in the wrong direction, the bouquet wrapping breathes like thin paper lungs, separate petals rotate at incompatible frame rates, miniature crops of her eye and shoes drift inside the bouquet, and tiny duplicate edges appear one beat late; the face, body, and hands remain clean. At 00:05.387, the first clustered impact tears the cream background into irregular watercolor windows, the black suit separates into floating lapel cutouts, and the bouquet unfolds outward as a multi-plane botanical stage, landing for several crisp frames on the exact teal clothing, purple-haired woman, pale peonies, watercolor texture, and visible "AI" and "AITuber" text of <Picture 2>. Do not replace <Picture 1>: keep a small full-body suited version walking across the lower paper margin, preserve cream-background shards as moving windows, and let black lapels frame the watercolor portrait. On the impacts at 00:05.611 and 00:05.845, the camera whip-pans left, snaps to a low angle, then rolls clockwise with large amplitude as peony petals become offset contact-sheet tiles of her eyes, hands, ponytail, suit, and bouquet. At 00:07.637 and 00:07.861 the tiles collide, reverse depth, and print a blue sky through her silhouette while watercolor peonies continue blooming inside the black-suit cutouts. From 00:08.864 the camera accelerates through a circular tear in her iris; at 00:09.099 the layers resolve sharply into the exact close-up, blue pop-art sky, sunflower, face, ponytail, gold "AI" clip, and halftone composition of <Picture 3>. From 00:09.100 to 00:15.000, keep all three references concurrently alive: place the original full-body figure as a rotating paper figurine inside the sunflower center, use the watercolor peony portrait as translucent cloud-shaped windows across the sky, and retain the pop-art close-up as the dominant plane. Animate the sunflower as a rotating offset-print shutter, duplicate its petals into different perspective planes, slice the blue sky into horizontal strips that travel at contradictory speeds, and orbit the camera around her profile through abrupt 20-degree angle jumps without breaking directional momentum. End at exactly 00:15.000 on a sharply legible synthetic tail frame: her purple-haired profile is centered inside the dark pupil of an enormous yellow sunflower, a miniature black-suited version and pale watercolor peonies remain visible within separate pupil layers, three red hibiscus-shaped paper apertures are entering from screen left, the camera is still rolling clockwise, and torn blue and yellow layers remain suspended mid-collision. No fade, dissolve, generic crossfade, new text, dialogue, or lip movement.

为什么这样写

来自 @ivy432hz 的公开 MiniMax H3 实测案例。建议先复用主体、时间轴、镜头与音频结构,再替换角色、商品或场景变量。

模型与出处

使用模型

MiniMax H3 (Hailuo 03)

提供方

MiniMax / Hailuo AI

Cheerself AI 收录 · @ivy432hz X 原帖

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。