Sora 2
Create a 10-second vertical 9:16 ultra-realistic cinematic smartphone video. Scene: A busy city sidewalk at sunset. People are walking, cars passing in the background. The footage feels completely spontaneous, as if someone suddenly pulled out their phone to record something strange. 0–2 seconds: A tiny stray kitten is sitting alone in the middle of the sidewalk, looking directly into the camera. The kitten looks extremely realistic, slightly confused and vulnerable. People casually walk past without noticing. 2–5 seconds: A huge black raven suddenly lands next to the kitten. The raven slowly looks at the kitten, then turns its head and stares directly into the camera. The camera subtly zooms in, creating suspense and making the viewer wonder what is about to happen. 5–8 seconds: The raven unexpectedly pulls out a tiny red flower from underneath its wing and gently places it in front of the kitten. Several pedestrians stop and stare in disbelief. The moment feels emotional, surreal, and completely unexpected. 8–10 seconds: The kitten picks up the flower with its mouth and looks directly into the camera. The camera rapidly pushes toward the kitten’s eyes. Inside the reflection of the kitten’s eye, we see the exact same kitten and raven scene happening again. The final frame must visually match the opening frame perfectly, creating an extremely smooth seamless loop that makes viewers watch the video twice without realizing it has restarted. Style: photorealistic, cinematic but believable, authentic smartphone footage, natural sunset lighting, realistic animal fur and feathers, physically accurate animal movement, natural human reactions, shallow depth of field, subtle handheld camera shake, realistic exposure and autofocus changes, immersive city ambience. No captions, no subtitles, no text, no logos, no watermark, no CGI look, no cartoon style. Make it feel like a completely real viral video accidentally captured on a phone.