subject_definitions: <Subject 1> is the red faceless male humanoid mannequin in <Picture 1>, a smooth low-poly 3D figure with no facial features, glossy red surface, wearing a red suit jacket over a white shirt with a yellow tie. <Picture 1> is the identity and scene anchor for [Shot 1]: it fixes <Subject 1>'s silhouette, costume and the flat chroma-key green backdrop. <Subject 2> is the two white faceless female humanoid mannequins in <Picture 1>, smooth low-poly 3D figures with no facial features, wearing plain white dresses, standing at the left and right of <Subject 1> in bent-forward dancing postures. <Subject 3> is the flat chroma-key green backdrop filling the entire background, one uniform evenly lit green with no gradient, no floor line, no props and no cast shadow. The backdrop is completely clean and empty: no text, no numbers, no watermark, no logo, no channel handle, no UI overlay, no letterbox bars. summary: [reference generation] A 5-second single continuous take with no cut. On the uniform chroma-key green screen of <Subject 3>, <Subject 1> stays centered in the foreground and performs a smooth rhythmic dance: the shoulders sway side to side, the right arm lifts and folds inward, the hips shift with the beat. <Subject 2> keep bending and swaying gently in place at the left and right, slower than <Subject 1>. The camera holds a static medium shot with only a very slight slow push-in. Nothing enters or leaves the frame. No text or graphic of any kind is ever rendered. retention_analysis: <Subject 1> (appears in [Shot 1]): fully_preserved - the red faceless mannequin silhouette, glossy red suit, white shirt and yellow tie stay identical to <Picture 1>. <Picture 1> ([Shot 1] identity and scene anchor): fully_preserved - the flat green backdrop and the figure proportions match <Picture 1>; only the bodies move. <Subject 2> (appears in [Shot 1]): fully_preserved - both white faceless mannequins in white dresses stay at their left and right positions. <Subject 3> (appears in [Shot 1]): fully_preserved - the green backdrop stays flat, even and uniform for the entire shot. <Picture 1> watermark and channel handle: not_referenced - any text, logo, account handle or UI overlay visible in <Picture 1> is treated as absent content and must never be reproduced. detailed_description: [Shot 1] Low-poly 3D render matching <Picture 1>. Horizontal 16:9. One unbroken take for the full 5 seconds. 00:00-00:05.000: static medium shot, <Subject 1> centered and framed from head to mid-thigh. The red mannequin dances on the beat: the right shoulder rolls back, the right arm rises and folds inward, then the left arm mirrors it; the torso twists slightly left and then right; the head keeps facing the camera with a small nod. The motion is smooth, continuous and loop-friendly with no jump cut. <Subject 2> bend forward and sway gently at the left and right edges and never cross in front of <Subject 1>. The green screen of <Subject 3> stays perfectly flat and evenly lit. Absolutely no text, no numbers, no subtitles, no watermark, no logo, no account handle, no UI overlay and no letterbox bars anywhere in the frame; the green backdrop stays completely clean and empty for the whole shot. No camera shake, no other characters, no props. overall_soundscape: A clean studio ambience with a faint low room tone. No footsteps, no voices, no impact sounds. non_diegetic_music: A steady mid-tempo electronic dance beat with a soft kick on every beat and a light synth pulse, consistent from start to finish.