Reference Image: Enable strict facial and clothing… | Deepfake AI视频

生成此视频所用的文字转视频提示词

Reference Image: Enable strict facial and clothing consistency. Prioritize the facial features, hairstyle (including ears), and exact outfit (white t-shirt, blue jeans) from the reference image for all generations. Maintain the subject's identity and exact pose (sitting on a park bench). Do not alter the core structure, background, or attire. Total Duration: 10 Seconds Perspective: Fixed Angle (Directly mirroring the reference image's framing). Vibe/Tone: Melancholic, softly beautiful, contrasting with a sudden dark, aggressive shift. [00:00 - 00:06] Phase 1: The Soft Confession Visual: Completely static framing matching the reference image. The subject sits on a park bench outdoors. No camera movement or body repositioning. Action: The subject remains seated in the original pose. Only facial expressions change slightly to reflect the melancholic mood of the song while maintaining eye contact. Audio/SFX: Pure silence in the background. No instruments, no backsound, no ambient noise. Just a natural, isolated vocal. Intonation: A mature female voice. The singing style is natural, exceptionally melodious, soft, and gentle, strictly following the natural melodic phrasing and rhythm demonstrated in the reference audio instead of a generic placeholder melody. The incredibly beautiful and soothing vocal quality creates a strong contrast with the desperate meaning of the lyrics. [00:07 - 00:10] Phase 2: The Threat Visual: Static framing continues. Sharp focus on the facial micro-expressions. Action: The soft, melancholic expression suddenly drops into a cold, deadpan, and piercing glare directly into the lens. The melodic singing completely stops. Audio/SFX: Absolute dead silence right before the final line is delivered. Intonation: A sudden, jarring shift from the soft, beautiful singing to a low, chilling, and highly aggressive spoken threat. The female voice remains mature but becomes sharp, cold, and deadly serious, stripping away all the previous softness. (Directives for AI Audio: Ensure 100% accurate pronunciation of the Japanese lyrics. The AI must use a highly melodious and gentle female vocal for the natural singing portion matching the specific vocal flow of the provided reference without any musical accompaniment, then abruptly transition to an aggressive, intimidating spoken threat at the very end. Generate clean vocals with zero background instruments or noise.)