Chicos, estaba experimentando con Seedance 2.0 para un nuevo avatar IA. El primer video es solo este prompt: "woman recording a selfie video saying te camera: "Hi! I’m Mila, an AI creator, cosplayer, gamer, and professional overthinker. I make nerdy, cute content. Your support helps me improve my setup and create even more for you. Welcome to my world!" Pero la voz es muy "falsa", se siente demasiado IA. Entonces le pedí a Claude con su Skill de Seedance 2.0 de Higgsfield y me dio este prompt. Escucha el segundo video como es más natural: SCENE CONTEXT A young woman films a selfie-style video of herself sitting at her gaming desk setup, holding the phone at arm's length, talking directly to the camera in a casual, warm, slightly self-deprecating tone. ACTIVE REFERENCES @[milapc](3e2d742b-ad7f-48d3-9fed-9ce60633a9a7): young woman, red wavy hair, round black glasses, freckles across nose and cheeks, navy blue t-shirt, warm smile, seated selfie pose with arm extended toward camera. 100% matches the reference. LOCATION MAP Foreground: her face and shoulders fill most of the frame, arm extended toward the lens. Midground: dark desk behind her with a monitor showing a torii-gate landscape wallpaper, two small figurines, keyboard with backlit keys, purple LED strip glow along the desk edge. Background left: bed with pink linens, cherry blossom branches, a small candle. Background right: window with warm sunlight through blinds, a clothing rack. Camera is her own phone, held at arm's length, slightly above eye level. FIRST FRAME/BLOCKING She is centered in frame exactly as in the reference, phone-selfie framing, head and upper chest visible, eyes looking directly into the lens, a small warm smile already forming before she speaks. FORMAT MODE One continuous shot, the camera does not cut on its own. Handheld selfie framing held throughout. OPTICS MCU (medium close-up), FOV 29°, natural portrait compression, shallow depth of field with the desk and monitor softly blurred behind her, no drift for the full duration.