Novel-to-Viral: Auto-Generate Thrilling Captions and Storyboards

The Problem It Solves

In the era of short videos, converting lengthy novel chapters into fast-paced, algorithm-friendly viral posts is a core challenge for content creators. Manual rewriting is time-consuming and often fails to nail the precise "pop density," while generic AI tools can produce verbose, high-compliance-risk, or storyboard-misaligned text. There is a need for a tool that automatically scans source text, extracts core conflicts, and outputs ready-to-use "narrative captions + storyboard scripts."

How It Works

The skill's core is a complete pipeline from text scanning to multimodal output, with capabilities and key steps as follows:

Core Capabilities:
* Intelligent Highlight Scanning: Instead of uniform compression, it automatically identifies key plot points like "conflicts/reversals/humiliation moments/power crushes" and strings them together based on "pop types" (e.g., identity reveal, power碾压, information bomb).
* Genre-Adaptive Emotion Modeling: Based on the input genre (isekai, urban, xuanhuan, etc.), it automatically matches different emotional rhythm models. For instance, "face-slapping" genres follow a chain of anticipation → conflict → explosion → dominance.
* Hard Compliance Constraints: Built-in content red lines automatically avoid violations such as gore, sexual suggestion, and follow-baiting. It uses emotions and results to substitute sensitive descriptions.
* Linked Storyboard Generation: Narrative paragraphs and storyboard shots are strictly one-to-one, separated by blank lines, ensuring the visual cutting rhythm syncs with the copy.

Key Workflow Steps:
1. Input & Validation: Receives novel text (single chapter or collection, up to ~20k characters). The skill first performs legality checks, prompting for empty input or non-novel text.
2. Highlight Extraction & Narrative Creation: From a first-person perspective, it crafts oral-style posts following a five-act rhythm: Hook Opening → Background Setup → Conflict Build-up → Emotional Peak → Cliffhanger Ending. It strictly adheres to the hard rule of "at least one pop/reversal every 200-300 characters."
3. Storyboard Script Generation: After narrative completion, it generates a shot description for each blank-line-separated narrative unit. Descriptions include key details like character actions, environment, and camera angles for direct editor reference.
4. Self-Check & Output: Internally executes a 12-item quality checklist, covering narrative perspective, pop-rhythm, storyboard count consistency, and content compliance, before outputting structured text.

Boundaries & Key Notes

• Capability Limits: Excels at mainstream web novel genres like isekai, urban, xuanhuan, suspense, ancient romance. Not suitable for pure literature, essays, or poetry (non-narrative text). It handles a maximum of ~15 chapters (~20k characters) per run and cannot process an entire book at once.
• Input Requirements: The original novel text must be provided. Providing only the book title won't work. Text that is too short (<200 characters) may lead to incomplete output. For chapters consisting purely of dialogue or narration, it will generate but include a warning.
• Compliance Red Lines: Strictly prohibits glorifying gore, sexual suggestion, follow-baiting (e.g., "please follow"), and superstition. For example, fight scenes only describe momentum and outcome ("as the blade fell, the entire hall fell silent"), not gory details.
• Output Style: Enforced first-person narration throughout. Internal monologue (e.g., "I thought") is forbidden; emotions are conveyed through actions and dialogue. Language must be colloquial, e.g., changing "he articulated" to "he said."