media-use

Resolve, generate, and operate on media assets for HyperFrames video compositions.

1|Updated Jan 10, 2024
One-click install
npx skills add https://github.com/noemdb/cfla --skill media-use-noemdb
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: media-use
Source: https://github.com/noemdb/cfla/tree/main/.agents/skills/media-use
Command: npx skills add https://github.com/noemdb/cfla --skill media-use-noemdb

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve? Building video compositions requires sourcing background music, sound effects, images, icons, logos, voiceovers, and color grades from scattered catalogs and tools, then tracking them across projects. This Skill unifies every media need into one resolve verb that returns frozen local files plus a ledger record, eliminating context noise and manual asset hunting. ## Core Features & Use Cases - Unified media resolution: Resolve BGM, SFX, images, icons, brand logos, TTS voices, color grades, and LUTs into frozen local files with one command, with generation fallback via TTS/music/image models when catalogs miss. - Audio engine: Produce voiceovers, transcriptions, captions, background removal, and GSAP volume-ducking keyframes through a shared audio pipeline. - Media operations and treatments: Cut, reframe, dither, and apply source-aware color treatments (VHS, camcorder, halftone, ASCII, film memory) with deterministic persistence and verification. - Cross-project reuse: Adopt existing project assets into a searchable ledger and reuse frozen assets across projects. - Use Case: While building a product promo video, resolve a music bed, generate a TTS voiceover with word-level captions, apply a restrained product-polish grade to the footage, and auto-duck the music under the narration. ## Quick Start Ask the agent to resolve a background music track and a TTS voiceover for your HyperFrames project, then run the resolve script with --doctor first to verify the heygen CLI is installed and signed in.

Frequently Asked Questions about media-use

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

FAQPage Schema
How do I resolve background music or sound effects for a video project?▼

Run the resolve script with --type bgm or --type sfx, an --intent description, and your --project directory. It searches the HeyGen catalog and bundled library, freezes the chosen file locally, and records it in the project ledger.

How do I add a TTS voiceover with captions to a composition?▼

Use the shared audio engine with an audio_request.json listing your lines; it produces voice files plus word-level timestamps in audio_meta.json for captions. HeyGen CLI auth unlocks the free TTS path, with local Kokoro as an option.

Does the color grading workflow support HDR or LOG footage?▼

The realtime treatment path is Rec.709/SDR only and must not silently process HDR, HLG, PQ, or camera LOG sources. The dither script explicitly rejects PQ/HLG inputs and requires tone-mapping to Rec.709 first.

Can I reuse media assets across multiple projects?▼

Yes. Resolved assets are frozen locally and recorded in a ledger; before resolving fresh, list reusable candidates with --candidates and judge fit. You can also adopt existing project assets into the index with --adopt.

Why should I never read a .cube LUT file directly?▼

A 3D LUT contains roughly size-cubed lines of raw numbers (about 36k lines at size 33), which bloats context without legible signal. Use grade-compare to see it rendered or cube-validate for a one-line check instead.

What happens when the media catalog has no matching asset?▼

When the catalog misses, the skill falls back to generation through TTS, music, or image models. LUT entries can also resolve offline via deterministic buildCube params when a CDN download fails.