Storyboard Concept: Reference-to-Video from Text to Image

Demonstrates how the "text-to-image" storyboard concept is implemented in a reference-to-video manner: first use two concept images of the same scene as reference materials (one wide shot to set the tone, one medium shot to establish the subject), then specify camera movement, pacing, and lighting direction in the prompt, so the model completes the transformation from static storyboard to dynamic shot within a single 8-second, 16:9 output. In terms of composition, base contains only non-asset input parameters: prompt, duration, resolution, aspect ratio, native audio toggle, and smart storyboard toggle; all media inputs are generated by assets and bound to the image_urls array and the elements' element_input_urls, avoiding any externally fabricated URLs. Applicable scenarios: early-stage storyboard previsualization for film and animation, shot concept validation for advertising creative proposals, and rapidly expanding existing concept images into playable shot drafts. The reference images serve as anchors for style and subject consistency, while the prompt handles camera language and pacing control; the two roles are separated, making it easy to iterate item by item.