- 成片時長:
- 15s
- 成片尺寸:
- 1280 × 720
藍色影棚里的超現實時裝
藍色影棚里的超現實時裝的原生視頻效果,原作者在 X 上完整公開了生成 Prompt。 原生視頻規格為 15 秒 · 2560×1440。原帖完整公開了 Prompt,本頁按原文保留
- 影片模型
- MiniMax H3
- 成片時長
- 15s
- 畫面比例
- 16:9
- 生成模式
- 未指定
提示詞
[Upload reference ] + Use la ragazza in reference as a strict identity reference for the model and her yellow sunglasses. Preserve face, hair, and the exact frame shape and colour throughout. 15 seconds, 16:9. Premium fashion commercial, photoreal live action. Infinite black void, polished black floor. She is the only lit element: hard key from one side, second hard light from the other, deep shadow between, strong speculars on the yellow frame and lenses. She walks slowly toward the camera in a black high-neck top and long black coat, chin level, eyes on the lens. Camera tracks backward at her pace, holding her size in frame. All type is heavy condensed uppercase sans-serif in vivid yellow, rendered as physical objects in 3D perspective with real thickness, catching the same light as she does and casting faint shadows on the floor. [0-4s] LOOK rises out of the floor ahead of her in deep perspective, grows as she approaches, and sweeps past on both sides out of frame. [4-7s] SEE enters from the right at high speed, streaks behind her with heavy motion blur, exits left. [7-10s] FOCUS appears large and fully out of focus, resolves into razor sharpness over half a second, holds, breaks apart. [10-12s] Four thin yellow lines draw a rectangle around her and FRAME assembles along its lower edge. Throughout, passing type reflects across the glossy lenses as it moves near her. [12-15s] She stops. All type collapses inward and vanishes into black. The camera continues into a tight close-up on her face and the yellow sunglasses. She holds completely still, eyes on the lens. On the black beneath her the title composes in the same yellow type: "SHIOBA LENS", and below it, smaller: "MADE TO BE SEEN". Hold. All text clean, legible, correctly spelled English capitals, each word once only. No garbled or duplicated letters, misspellings, subtitles, watermarks, or other text or logos. Audio: low sustained bass, Composition: model centered, all type within the central third of the frame
準備素材與修改建議
- 先檢查產品輪廓與文字,再修改燈光。
- 一次只修改一個動作或鏡頭方向,其他條件保留。
- 這些是修改與檢查建議;來源影片不能證明 Studio 按這些修改生成了相同結果。