hbg-life-simulation

Convert Chinese life-story scripts into narrated comic videos with synchronized captions and QA checks.

137|26|Updated Aug 1, 2026
One-click install
npx skills add https://github.com/Mr-funny/hbg-life-simulation --skill hbg-life-simulation-mr-funny
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: hbg-life-simulation
Source: https://github.com/Mr-funny/hbg-life-simulation
Command: npx skills add https://github.com/Mr-funny/hbg-life-simulation --skill hbg-life-simulation-mr-funny

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires edge-tts, and includes scripts (resource) and references (resource) and assets (resource) components.

What problem does it solve? Turning a long Chinese "模拟人生 / 人生副本" story into a finished narrated video normally requires manual character consistency, TTS timing, caption splitting, storyboard planning, camera motion, audio mixing, and final MP4 verification. This Skill automates that entire pipeline while enforcing identity anchors, caption semantics, storyboard density, and encoded-video quality gates. ## Core Features & Use Cases - Consistent comic IP: Generates and approves character identity anchors before scene art, repeating immutable facial traits in every prompt so the protagonist never changes face. - HBG opening module: Builds a rapid multi-life flash opening with ratchet SFX, alternating zoom motion, and a complete selected-life title reveal. - Audio-driven storyboard: Uses continuous Edge TTS with real VTT timing to align scenes and semantic Chinese captions, targeting 4–8 second stills. - Rendering and QA: Assembles HyperFrames compositions with zoom/pan motion, mixes fixed-level BGM, supports low-disk streaming FFmpeg rendering, and verifies the final encoded MP4 for black frames, silence, loudness, and caption safe areas. - Use Case: Paste a 10-minute Chinese life-story script and receive a verified 1920×1080 MP4 with a unified comic protagonist, synchronized short captions, and stable background music. ## Quick Start Use the hbg-life-simulation skill to turn my Chinese life-story script into a landscape narrated comic video with a consistent protagonist, synchronized captions, and a final quality-checked MP4.

Frequently Asked Questions about hbg-life-simulation

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I turn a Chinese story script into a narrated comic video?

Provide the story text and invoke the skill; it archives the source, locks a comic character identity, generates continuous Edge TTS narration with VTT timing, builds a dense storyboard, and renders a verified MP4. Landscape 1920×1080 is the default output.

How does the skill keep the same character across all generated images?

It generates an identity anchor image first and requires approval before scene art. Every later prompt repeats the immutable facial traits, age, hairstyle, and clothing rules defined in CHARACTERS.md, and identity-drifted sheets are regenerated.

Does the skill support vertical 1080×1920 video?

Yes, portrait output is supported but only when explicitly requested. The style system initializes a separate portrait HBG_STYLE.json with its own caption safe area, and landscape parameters are never copied into portrait projects.

Can I use my own account for image generation instead of built-in ImageGen?

Yes, through a localhost-only CLIProxyAPI setup, but only after explicit per-project consent. The user completes OAuth login personally, credentials are never read or stored in the project, and built-in ImageGen is the fallback when declined or unavailable.

Why does the skill reject some generated images even when they look good?

It enforces reality checks on anatomy and props: hands are traced fingertip-to-forearm at full resolution, phones must face the correct direction, and physically impossible objects are rejected. High-risk shots like hands and phones are generated as standalone images, not grid panels.

What are the limitations for rendering long videos?

Videos over four minutes require a disk preflight with at least 12 GiB free, style validation, and chunked encoding. A streaming FFmpeg fallback renders per-scene segments to avoid materializing the full timeline when disk space is constrained.