
Create an ultra-realistic cinematic night video using the up...
Prompt
Create an ultra-realistic cinematic night video using the uploaded image as the exact reference. The result must look like a real camera recording, not an AI-generated video. Preserve the person's exact identity and appearance from the reference image. Do not alter his face, facial proportions, eyes, nose, lips, jawline, beard, hairstyle, skin texture, skin tone, age, body proportions, clothing, or expression. No face replacement, no beautification, no facial enhancement, no identity drift. Keep the exact same nighttime location and composition from the reference image. The city buildings, street lights, trees, rooftop, background lights, shadows, colors, and perspective must remain consistent and physically realistic. Do not generate a different location or rearrange the background. The person stands naturally in the exact same position. Add only subtle realistic human movements: natural breathing, very slight body movement, realistic blinking, and tiny natural shifts in posture. For the first 7 seconds, he calmly looks toward the camera while standing naturally with his hands together. During the final 3 seconds, he slowly raises one hand upward in a natural, relaxed gesture, like greeting someone. The hand and arm movement must follow realistic human anatomy and natural physics. No sudden movement, no stiff fingers, no distorted hands, no extra fingers. Use realistic low-light camera exposure, natural skin texture, realistic shadows, subtle lens characteristics, natural depth of field, realistic motion blur, accurate light reflections, and slight natural camera movement. Avoid excessive sharpness, artificial skin smoothing, plastic-looking faces, glowing edges, flickering lights, warped buildings, unnatural body motion, or exaggerated cinematic effects. The final video should look like an ordinary high-quality smartphone or professional camera recording captured at night in the real location. Keep everything stable and physically consistent from the first frame to the last frame.