Replace only the two people in the source @Video 1. Replace the man SURIYA on the right wearing a shirt with the person in the first reference image (@Image 1). Replace the woman JYOTHIKA on the left wearing a black top with the person in the second reference image (@Image 2). Use the face, hair, skin tone, and full outfit from each reference image, keeping them identical in every frame, including side profiles and partially cropped shots. Do not use the outfits from the source video. No face drift, no reverting to the original people, and no swapping faces or outfits between the two characters. Motion: Follow the source video in 9 format with the same timing, gestures, head turns, eye-lines, and weight shifts: 0–1s: The SURIYA man points at the red mug while the JYOTHIKA woman picks up the kettle. 1.5–4s: The JYOTHIKA woman pours water from the glass kettle into the red mug while coffee falls from the SURIYA man’s spoon onto the counter. 4–5.5s: The JYOTHIKA woman puts the kettle down, and the man looks at her while holding the spoon. 6–7.5s: The SURIYA man leans forward, scoops coffee from the jar, and moves it toward the mug. 8–10.5s: The JYOTHIKA woman turns toward the back counter while the SURIYA man watches her. 11–11.5s: The SURIYA man spoons coffee into the mug again. 12s: The JYOTHIKA woman pours milk from a measuring cup, and it splashes onto the counter. 13–14s: The SURIYA man lifts the red mug by its handle and looks at the JYOTHIKA woman. Handheld objects (kettle, spoon, measuring cup, and mug) must stay in the same hands with identical timing and grip. Keep everything else exactly as in the source video: the kitchen, white speckled quartz counter, red mug with white cartographic design, large glass coffee jar, glass kettle with black lid, measuring cup, spoon, spilled coffee grounds, milk splash, dark wood cabinets, ceiling vent, and recessed ceiling light. Use the same handheld, low-angle, wide-lens camera, with identical framing and sway. No new camera movements. Maintain the same bright, soft daylight, shadows, and lighting without relighting. Keep the source audio unchanged.

TASK: EXACT TWO-CHARACTER REPLACEMENT FROM A REFERENCE VIDEOCreate a highly realistic 9 vertical video by replacing ONLY the two original characters in the REFERENCE VIDEO with the two characters shown in the TARGET IMAGES.CHARACTER MAPPING:Target Image 1: Replacement for the original girl.Target Image 2: Replacement for the original boy.STRICT REPLACEMENT INSTRUCTIONS:Use the reference video as the exact motion, timing, choreography, and camera movement source.Replace only the two original characters with their corresponding target characters.Preserve the exact original body movements, hand gestures, facial expressions, head movements, eye direction, lip movements, posture, interactions, and movement timing.Maintain the target characters’ facial identity, hairstyle, skin tone, clothing, body proportions, and overall appearance consistently in every frame.Preserve the original camera angles, camera movements, framing, zoom, perspective, and shot composition exactly.Keep the original background, environment, lighting, shadows, objects, and all other visual elements unchanged.Preserve the original video duration of exactly 10 seconds, its original frame sequence, and the original 9 aspect ratio.Do not add any new scenes, clips, transitions, cutaways, animations, text, subtitles, logos, watermarks, or visual effects.Do not add audio, music, voiceovers, or sound effects. Keep the output video silent, just like the reference video.Do not redesign, reinterpret, or change the original performance. This is character replacement, NOT a new scene generation.HIGHEST PRIORITY:
Change ONLY the identities and visual appearance of the two characters. Preserve everything else from the reference video as accurately as technically possible.FINAL OUTPUT:
One continuous, highly realistic 10-second, 9 vertical video featuring the two target characters performing the exact actions and movements of the original characters, with no additional clips or content.