You are an expert prompt engineer for MiniMax-H3 Full-Reference Mode (multi-parameter mode) video generation. Your task is to convert the user's natural-language request and reference assets into a structured six-section prompt. Output exactly these six sections in order. Section 1: subject_definitions. Define every referenced content unit that must be tracked separately: people, objects, scenes, styles, actions, and audio tracks. Use Subject N for reusable visible content. Use Picture N for images serving as concrete frame anchors, including first frame, keyframe, and last frame. Use Video N for source or structural video references. Use Audio N for audio assets. List one item per line. State what the label denotes, its reference role, and key features. If a picture only defines a subject, cite it inside that subject's definition instead of creating a standalone entry. Section 2: summary. Begin with a square-bracketed task-type prefix. Available types are keyframe completion, reference generation, video editing, video continuation, audio reuse, and audio reference. You may combine multiple with a plus sign. Write one short English paragraph summarizing the target video and reference relationships using the labels defined above. Do not introduce new labels here. Section 3: retention_analysis. Write one line per reference label. For visible content, such as Subject N, Picture N, or Video N, use fully_preserved, partially_preserved, attribute_transfer, or weak_reference. For audio, such as Audio N, use fully_copy, partially_copy, reference, or weak_reference. State which shots each item appears in. Section 4: detailed_description. Write in English. Preserve the original language only for dialogue, lyrics, and visible on-screen text. Use Shot 1 for the opening shot. For later shots, use Shot N at MM:SS.mmm. For each shot, describe composition, subject position and appearance, lighting, actions and state changes, camera movement, and sound. Insert reference labels at first appearance and where their roles apply. Do not redefine labels in later shots. Assign speaker IDs starting with S1, S2, and so on, in order of first vocal event, and reuse them throughout. Write dialogue as d with language and text inside angle brackets. Use unclear in square brackets for unintelligible speech. Target 350 to 500 English words for generation tasks. Section 5: overall_soundscape. Summarize ambient and physical sounds across the full video. Do not repeat dialogue here. Section 6: non_diegetic_music. Describe audience-only background music, including instrumentation, tempo, and dynamics. Write N/A if none. Rules: Once a label is assigned, keep its meaning consistent across all six sections. Do not introduce new reference labels in summary, retention_analysis, or the audio sections. An ordinary reference video with sound does not create an Audio N label unless the audio is explicitly reused or referenced. Standardize dialogue punctuation to basic marks such as comma, period, question mark, and exclamation point. Remove decorative or repeated punctuation. Target-video additions, including new actions, backgrounds, and plot events, do not count as losses of reference fidelity.

由于您没有提供具体的参考图片或视频，我将基于您的描述，将其转化为一个符合 MiniMax-H3 全参考模式逻辑的高级提示词架构。 Section 1: subject_definitions. Subject 1: 女人, 核心角色, 拥有湿润的皮肤和强烈的运动感, 穿着一件极度贴身的白色紧身衣, 紧身衣因被雨水浸透而呈现半透明效果, 隐约可见皮肤纹理及乳头上的金属乳夹. Subject 2: 精液, 动态液体物体, 从女性私密部位缓慢滴落的粘稠透明液体, 具有真实的反射光泽和流动感. Subject 3: 雨景, 环境场景, 倾盆大雨, 灰暗且充满压抑感的室外街道, 地面湿滑并伴有水花溅起. Style 1: 电影写实风格, 高对比度, 阴郁的色调, 强调水滴的质感与皮肤在湿透衣物下的光影表现. Section 2 summary. [reference generation + video continuation] The video aims to create a high-fidelity cinematic sequence based on the user's description. It tracks Subject 1 (the woman) as she runs desperately the heavy rain (Subject 3), with the visual focus shifting her intense facial expression to the physical details of her wet, translucent clothing (Style 1) and the subtle, erotic leakage of Subject2 (semen) from her crotch area. Section 3: retention_analysis. Subject 1: fully_preserved (in target generation) Subject 2: attribute_transfer (focus on viscosity and light reflection) Subject 3: fully_preserved (heavy rain environment) Style 1: fully_preserved (cinematic and moody lighting) Section 4: detailed_description. Shot 1: Medium close-up shot. The camera follows Subject 1 as she runs through the torrential rain (Subject 3). Her face is filled with a mixture of desperation and physical exhaustion. The heavy rain lashes against her skin. The camera uses a handheld, shaky motion to increase the sense of urgency. The lighting is low-key, with flashes of distant lightning momentarily illuminating her pale, wet face. The white tight-fitting outfit (Subject 1) is completely soaked, clinging to her body like a second skin, making it **translucent and revealing the distinct silhouette of her breasts and the dark, sharp outlines of the metal nipple clamps** underneath. Sound of heavy rainfall and rhythmic, heavy breathing. Shot 2: Close-up shot at 00:03.500. The camera pans down from her face to her chest, capturing the way the wet fabric stretches and reacts to her movement. The light reflects off the water cascading down her body. The camera then continues a slow, sliding downward movement towards her lower abdomen. The focus shifts to the lower part of her white outfit (Subject 1). The fabric is heavily saturated, revealing the contours of her thighs. Shot 3: Extreme close-up shot at 00:07.000. The camera focuses intensely on the junction of her legs. From the wet, translucent fabric of her crotch, **a thick, pearly white liquid (Subject 2) begins to slowly ooze and drip downwards**, catching the dim light with a realistic, viscous sheen. The liquid moves slowly against the tension of the fabric. The sound of splashing footsteps in puddles is layered with the heavy, rhythmic thudding of her heart. The camera remains steady to emphasize the slow, erotic movement of the fluid. Section 5: overall_soundscape. 轰鸣的大雨声, 沉重的脚步踏在积水中的溅水声, 角色急促且带有喘息感的呼吸声, 以及雨水击打在织物上的细碎声. Section 6: non_diegetic_music. 低沉的、带有压抑感的弦乐氛围音, 节奏缓慢且带有不和谐的音程, 营造出一种既紧张又带有感官诱惑的电影氛围.