subject_definitions:
<Subject 1> is the person in <Picture 1>, whose facial identity, body proportions, physique, defining features, and all distinctive reference objects must remain consistent throughout the target video. The only intentional wardrobe modification is that the bunny suit and bunny ears are black; no other physical characteristic or clothing detail is changed.
<Subject 2> is the person in <Picture 2>, whose facial identity, body proportions, physique, defining features, and all distinctive reference objects must remain consistent throughout the target video. The only intentional wardrobe modification is that the bunny suit and bunny ears are white, <Subject 2> isn't wears shirt; no other physical characteristic or clothing detail is changed.
<Subject 3> is the visual style of <Picture 3>, providing the target video's overall rendering style, color treatment, lighting character, and cinematic presentation without replacing the identities or physical features of <Subject 1> or <Subject 2>.
<Picture 1> is the identity and appearance reference for <Subject 1>.
<Picture 2> is the identity and appearance reference for <Subject 2>.
<Picture 3> is the visual-style reference for the target video.

summary:
[reference generation] The target video shows <Subject 1> and <Subject 2> performing a synchronized dance in a dark nighttime park. <Subject 1> stands on the left and wears the same identity-accurate appearance from <Picture 1>, with only the bunny suit and bunny ears changed to black. <Subject 2> stands on the right and wears the same identity-accurate appearance from <Picture 2>, with only the bunny suit and bunny ears changed to white. The video follows a clear progression from still anticipation, through restrained body movement, to dynamic dancing, and finally a shared bow. The visual presentation follows the style of <Picture 3>, while both subjects' identities, physiques, clothing details, distinctive objects, and spatial continuity remain stable.

retention_analysis:
<Subject 1> (appears in [Shot 1], [Shot 2], [Shot 3]): partially_preserved - the person's facial identity, body proportions, physique, defining features, and distinctive reference objects from <Picture 1> are fully retained; only the bunny suit and bunny ears are intentionally recolored black.
<Subject 2> (appears in [Shot 1], [Shot 2], [Shot 3]): partially_preserved - the person's facial identity, body proportions, physique, defining features, and distinctive reference objects from <Picture 2> are fully retained; only the bunny suit and bunny ears are intentionally recolored white.
<Subject 3> (applies to [Shot 1], [Shot 2], [Shot 3]): weak_reference - the visual style, lighting character, color treatment, and cinematic presentation of <Picture 3> guide the target video without altering the identities or physical attributes of the two subjects.
<Picture 1> (identity and appearance reference for <Subject 1>): partially_preserved - its subject identity and physical appearance are retained, with only the specified bunny-suit and bunny-ear color change.
<Picture 2> (identity and appearance reference for <Subject 2>): partially_preserved - its subject identity and physical appearance are retained, with only the specified bunny-suit and bunny-ear color change.
<Picture 3> (visual-style reference): weak_reference - its style informs the rendering and presentation rather than being copied as a literal scene or replacing the nighttime park setting.

detailed_description:
The target video uses the visual language of <Subject 3>, with coherent nighttime contrast, controlled cinematic lighting, and a stable, polished presentation. The setting is a typical public park at night: dark trees and low grass form the background, with several streetlights creating pools of warm illumination while preserving the overall darkness. The camera remains frontal and at the subjects' eye level, keeps both full bodies visible, and maintains a stable spatial relationship between them while gradually moving closer.

[Shot 1] The video opens with a full-body frontal composition in the quiet nighttime park. <Subject 1>, whose identity and physical appearance match <Picture 1>, stands on the left side of the frame in the black bunny suit and black bunny ears. <Subject 2>, whose identity and physical appearance match <Picture 2>, stands on the right side in the white bunny suit and white bunny ears. Their faces, body proportions, physiques, defining features, clothing construction, and distinctive reference objects are stable and clearly visible. Both subjects are in a poised waiting state immediately before beginning the dance. They hold a natural ready posture with only subtle breathing and minimal preparatory swaying. The camera is level, frontal, and distant enough to show both subjects from head to toe. Focus initially favors <Subject 2> while keeping <Subject 1> legible.

[Shot 2] At 00:04.000, without a jump in camera position or background continuity, both subjects begin dancing. The first movement is restrained: a gentle rhythmic sway through the shoulders, torso, and hips, followed by small coordinated steps. The rhythm feels practiced and relaxed rather than stiff. The camera performs a slow, continuous dolly-in while maintaining a full-body composition. The performance then develops into dynamic choreography with larger arm gestures, clear weight shifts, turns of the upper body, and energetic synchronized footwork. <Subject 1> moves with visible enjoyment, confident timing, and fluid expressive accents. <Subject 2> follows the same choreography with slightly awkward secondary movements and a mildly tentative quality, while still staying coordinated with <Subject 1>. The awkwardness is limited to secondary motion and does not alter <Subject 2>'s identity, physique, costume structure, or facial features. Focus gradually transitions from <Subject 2> toward <Subject 1> as the camera approaches, with both subjects remaining sharp enough to preserve their identities.

[Shot 3] At 00:10.000, the dynamic sequence resolves cleanly. Both subjects complete the final dance phrase together, settle their feet, and face directly toward the camera. They pause briefly in synchronized stillness, then perform a shared respectful bow at the same time. <Subject 1> retains an enjoyable, comfortable finish, while <Subject 2>'s finish remains slightly less polished but clearly synchronized. The camera completes its gentle approach without a sudden zoom, pan, reframing jump, or focus snap. The final frame holds both full bodies and the consistent nighttime park, streetlights, trees, and grass. Preserve reference identity, body shape, facial structure, wardrobe details, black and white color assignments, and distinctive objects through the final bow. Avoid extra limbs, malformed hands, body deformation, costume morphing, ear-shape changes, background replacement, lighting flicker, temporal identity drift, unwanted cuts, and any camera discontinuity.

overall_soundscape:
Quiet nighttime park ambience continues throughout: a soft outdoor room tone, distant insects, a faint breeze moving through leaves and grass, subtle fabric movement from the costumes, and restrained footstep sounds during the choreography. The ambience remains spatially consistent and never overwhelms the dance.

non_diegetic_music:
An upbeat, polished dance track with a clear steady beat begins softly during the waiting state, builds from restrained rhythmic pulses into a more energetic section for the dynamic choreography, then resolves with a brief clean ending as both subjects bow. The music is audience-only, with no spoken dialogue and no abrupt changes that would disrupt the practiced rhythm.