Late-Night Car Park Conversation

The full template video becomes the reference for its original scene, motion, and timing; you only upload one clear character image. The live backend will keep the same prompt, model, aspect ratio, duration, and camera, plus seed and resolution where the provider supports them. AI will reproduce the result as closely as possible, but cannot guarantee a frame-for-frame match.

Template model Kling 3.0 isn't connected yet — this submission runs on Pika H3 2K with the same scene reference, prompt, ratio, duration and camera where supported.