AI 拥抱视频：用两张照片送出一个真正的拥抱

Create a 5-second natural, affectionate hug video using exactly two uploaded identity-reference images. Image 1 defines person A; image 2 defines person B. Keep each person's own facial features, age, skin tone, hair, body proportions and clothing distinct throughout. Do not merge faces, swap identities or invent a third person. Place both people together in a simple, softly lit setting inspired by image 1. Match the lighting without changing either identity. Show both from at least the waist up with their arms visible. Use one steady eye-level medium shot, no cuts, zooms or photo transitions. Seconds 0–1.5: already standing close, they turn gently toward each other. Seconds 1.5–4: both slowly raise their arms and settle into a calm, comforting hug. Seconds 4–5: hold the hug with quiet breathing and a small smile; no dramatic gestures. Keep faces slightly offset at three-quarter angles so neither face passes through the other. Hands rest naturally on the shoulders or upper back, with no extra fingers, limbs or intersecting arms. No kissing, talking, fantasy transformations, angel wings, captions, logos or watermarks. Silent video, no music and no speech.