What problem does it solve? Turning storyboard frames and reference images into short AI-generated video clips through fal.ai requires careful handling of uploads, endpoint selection, prompt conventions, and output tracking. This Skill wraps that entire workflow in a single CLI so agents can run Seedance 2.0 reference-to-video and image-to-video generations with dry-run validation, local file uploads, MP4 downloads, and JSON receipts. ## Core Features & Use Cases - Reference-to-video (ref2v): Animate up to 9 image references on Seedance 2.0 full or fast endpoints, with @ImageN prompt handles, 480p-1080p resolution, and multi-action timed-segment prompts. - Image-to-video (i2v): Deterministic start-to-end frame morphs across Seedance and Kling O3/V3 endpoints, with automatic parameter stripping for Kling. - Safe iteration workflow: Dry-run every prompt before spending credits, verify credentials with validate and doctor --remote, and keep a JSON receipt next to every downloaded MP4 for reproducibility. - Use Case: Animate a storyboard frame into a 4-second portrait canary using two extra identity/style anchor images, then compare receipts across runs to measure drift. ## Quick Start Ask the agent to dry-run a Seedance canary that animates your storyboard frame with a restrained motion prompt, then remove the dry-run flag to render the MP4 into your project folder.
