Bulin 47 AI 视频生成器

One continuous 6-second photorealistic vertical handheld medium-close freestyle performance outside the red-lit restaurant entrance. @Image1 is the ONLY identity and COMPLETE outfit source for the central adult performer. Preserve that uploaded face, hair, age appearance and full everyday clothes, replacing the featureless gray mannequin in every movement reference. Ignore the portrait background. Reference people are distinct from the performer. @Image2 is the opening pose at 0.0s: face front/SCREEN-RIGHT listener, screen-right hand pinched near chin. @Image3 is the turn toward SCREEN-LEFT at 1.0s. @Image4 is the wider open-palm gesture toward SCREEN-LEFT at 1.5s. @Image5 is the return toward the front/right listener at 2.5s. @Image6 is the emphatic raised open palm held forward with a clear gap from the face, with a slight forward lean at 4.0s. @Image7 is the lowered bent forearm sweeping sideways across the lower chest and closer framing at 5.5s. These are six ordered poses of ONE performer, not different people. Synthesize continuous moving full-screen footage, never a collage, gray figure, labels or slideshow. Keep the same red columns, glass doorway and three background onlookers. The SCREEN-LEFT foreground listener wears black with two white sleeve stripes; that listener's arm occludes the lower torso. The SCREEN-RIGHT foreground listener wears muted purple and holds an upright smartphone beside ONE round portable black speaker. Preserve their roles, not their identities as the central performer. Keep the uploaded person's head and body coherent behind these foreground occlusions. 0.0–0.8s: a small screen-right-hand pinch beside chin, looking toward right listener, mouth articulating a fast phrase. 0.8–1.8s: rotate head and shoulders SCREEN-LEFT, open both palms outward at chest height, matching @Image3 then @Image4. 1.8–2.8s: rotate back toward front/right listener, raise an open palm and answer with the other hand, matching @Image5. 2.8–3.8s: alternating chest-level hand beats while mouth performance continues; camera drifts slightly closer. 3.8–4.8s: lean slightly forward, raise the open hand toward the listener in front of the upper chest, fingers relaxed and spread, matching @Image6. Leave a clear gap between hand and face: never touch the cheek, ear or forehead. 4.8–6.0s: LOWER that hand beneath shoulder height, bend the elbow and sweep the relaxed hand sideways across the lower chest toward SCREEN-LEFT while continuing rapid animated mouth movement, matching @Image7. Finish talking with a loose low hand gesture, not a stop palm and not an arm extended at the camera. Do not touch the face and do not loop back to the first gesture after 3 seconds. The right foreground head/shoulder grows slightly larger as camera and performer lean closer. End still in performance, no freeze or bow. A single handheld medium-close shot, slight forward drift only, no cut, orbit, sudden zoom-out or full-body dance. No microphone or stage. Natural hands and elbows; no extra fingers, fused arms, identity swaps, logos, text, lyrics or watermarks. Silent output; no audio track. Mouth movement is visual performance, not a promise of synchronization to any soundtrack.