Viral Video Teardown & Short Video Remake

The Problem: What Do You Learn From A Viral Video?

When content creators encounter a high-performing short video, they face a dilemma: either superficially imitate the visuals and script, leading to homogeneity, or perceive "structure" as too abstract to grasp. The core pain point is how to extract the underlying structure—the hook, pacing, and functional allocation of shots—and systematically transplant it to one's own product, topic, or account, moving beyond mere surface replication.

How The Skill Works: Structural Extraction & Remaking

The viral-video-teardown-remake skill provides an end-to-end technical pathway from analysis to reproduction. Its workflow is not a simple "reskin," but a structural rebuild.

1. Structured Teardown: The reference video (as a link, file, or screenshot) is dissected into a hook, mid-section shots, and a conversion CTA, with precise start and end times marked. Success is attributed across three dimensions: the structural design itself, the weight of the content (topic/information), and the craft of presentation (camera/editing), scored across six criteria. The output is a "blueprint" of why the video worked.

2. Topic Remake & Storyboarding: The original video's shot count, function allocation, and overall temporal shape are preserved, but all content is replaced with the user's topic. The key output is a storyboard, where each shot is explicitly split into visual and voiceover fields—a necessary condition for later synthesis. The voiceover script must be edited to fit the time limits of the target real-time model card (e.g., text_to_speech) during this authoring stage to prevent truncation later.

3. Limited Generation & Confirmation: After the free stages, storyboard images (beatra.images.generate) and voiceover audio (beatra.speech.synthesize) are generated, allowing for preview of real assets. The final video generation requires explicit approval. The workflow enforces pre-generation confirmation covering: the selected video model, video duration based on the actual voiceover length, canvas ratio (default 9:16), and estimated cost. The final video duration is strictly constrained by the selected model card's capabilities, and only the opening frame is animated; other approved shots are delivered as stills for user editing.

4. Delivery & Review: Final deliverables include the teardown report, storyboard, all still images, voiceover audio, and the generated video. Key metadata for each generation task is included, such as task ID, returned media links, model used, actual duration, and charged credits.

Key Constraints & Suitability

• Input-Driven: This skill strictly requires a reference video to tear down. Without a specific benchmark video—only an original idea or a product image—other workflows like "storyboard script" or "product video" are more suitable.
• Hard Duration Constraint: The total video duration cannot exceed the maximum supported by the selected real-time video model card (image_to_video). The voiceover script must be written within this limit, an irreversible constraint for the workflow.
• Payment & Confirmation Nodes: The entire process is free until the final storyboard is approved. Every subsequent generation step (images, speech, video) is a paid operation, each requiring independent confirmation of parameters and cost before execution. Video generation is the most expensive step.
• Creative Scope: This skill delivers an original short film built upon a successful video's structure, not a replica of the original's visuals, music, or on-screen talent. Requesting the latter falls outside this workflow's capability.