First-frame-driven native-audio short film: Vidu Q4 Preview image-to-video

Using the image-to-video capability of Vidu Q4 Preview (the author-track image-to-video route) as its subject, this is a hands-on demonstration with minimal parameters but complete information: a single first-frame image drives the visuals while the native audio switch is enabled, so dialogue and sound effects are produced within the same generation pipeline. On the author route, the only required field is image_url, so this entry focuses on the division of labor in which "the first frame determines composition and aspect ratio, while the prompt determines motion and sound": base explicitly sets duration=8, resolution=1080p, and audio=true, covering the full default-value semantics of the schema; the prompt describes only camera movement, character performance, and sound events, and does not re-describe the visual content already fixed by the first frame. The first-frame material is generated by the production line (one image asset) and bound to image_url. Use cases: anyone who wants to quickly verify what kind of short film with sound can be produced from "one image + one motion description"; developers who need to run multi-option comparisons within the 3–16 second range; and production teams that want to understand how this model family performs on character expressions, camera movement, and synchronized sound effects. Note that on this route a single generation is capped at 16 seconds, so long shots must be split and stitched, and the output aspect ratio follows the first-frame image's proportions (1:4 to 4:1).