subject_definitions: <Subject 1> is the adult woman whose… | Deepfake AI视频

生成此视频所用的文字转视频提示词

subject_definitions: <Subject 1> is the adult woman whose facial identity, short dark undercut hairstyle, East Asian features, warm skin tone, athletic build, and facial structure are defined by <Picture 2>; in the target video she retains that identity consistently while appearing in the waterfall setting from <Picture 1>. <Subject 2> is the moss-covered rock from <Picture 1>; its dark rounded surface, green moss, and the visible word "MOSS" remain clear at the start, then the rock is crushed in her hand and breaks into small fragments and rubble. <Subject 3> is the waterfall creek setting from <Picture 1>; the waterfall, shallow stream, rocks, and green forest surroundings provide the environment. <Picture 1> is the concrete opening frame and composition anchor for [Shot 1]. summary: [reference generation + keyframe completion] The target begins on the exact opening composition of <Picture 1>, with <Subject 1> holding <Subject 2> in <Subject 3>. She gives a confident smirk, then squeezes the rock in her hand until it crumbles, causing small bits of rubble and debris to fall downward while the waterfall continues moving behind her. retention_analysis: <Subject 1> (appears in [Shot 1]): fully_preserved - the woman retains the facial identity, hairstyle, skin tone, and athletic proportions defined by <Picture 2>. <Subject 2> (appears in [Shot 1]): partially_preserved - the mossy rock begins intact with readable "MOSS" lettering, then is visibly crushed and broken apart into fragments and falling rubble. <Subject 3> (appears in [Shot 1]): fully_preserved - the waterfall, creek, rocks, and forest background remain consistent and visibly active. <Picture 1> ([Shot 1] first frame): fully_preserved - the opening pose, framing, wardrobe, hand position, rock placement, and overall waterfall composition match the connected image at the start of the clip. detailed_description: Naturalistic live action with controlled contrast, soft outdoor daylight, and measured framing. The clip should preserve the identity from <Picture 2> while beginning from the exact visual arrangement of <Picture 1>. [Shot 1] The video opens on the exact composition of <Picture 1>: <Subject 1> faces the camera in a steady medium close-up, holding <Subject 2> up beside her face in the waterfall setting of <Subject 3>. The camera remains static throughout, with no cut, no zoom, and no pan. After a brief opening beat, <Subject 1> gives a small, confident smirk into the camera. Without changing the overall framing, she tightens her grip on <Subject 2> and visibly squeezes it in her hand. The force of her grip crushes the rock rather than merely cracking it. The rock compresses and breaks apart in her hand, with small chunks, grit, and rubble falling downward from between her fingers. The falling debris is clearly visible and drops naturally out of frame. She maintains composure and control during the crush, keeping the moment readable and grounded. Her expression stays self-assured and faintly amused as the fragments continue to fall. The waterfall and stream behind her continue moving naturally throughout the shot. She finishes holding the crushed remains in her hand with a calm, satisfied look. overall_soundscape: Constant waterfall and creek ambience with soft forest air throughout. A subtle hand-tightening rustle leads into a crisp stone-crushing sound, followed by small bits of rock and rubble falling and lightly scattering downward. No speech. non_diegetic_music: N/A