Index of every page on this site: llms.txt
All skills
Run it yourself

Grade short-form vertical video with ffmpeg only: one locked chain (deband, grain, S-curve with a 0.94 highlight shoulder, saturation), frozen into a versioned Hald LUT, with a numeric QC gate (ffprobe plus signalstats) and a face-cam exception. Use at the final encode of a video, after the edit is assembled losslessly and before delivery.

  • 4 files
  • 23.5 KB
  • Updated 5 days ago
  • GitHub
Use this Skill: https://skilld.dev/gh/jokerhtml/needcraft-skills/color-grading

Nothing lands on disk. Nothing to clean up.

Fork this Skill

Edit a local copy. It keeps the author and licence.

README.md

≈648 tokens on demand. Your agent reads this file only when SKILL.md points to it.

Color Grading (ffmpeg-only)

Grade short-form vertical video with ffmpeg only: one locked chain, frozen into a versioned Hald LUT, verified numerically.

What it is

A repeatable final-encode method: render clean, assemble losslessly, grade once with a locked filter chain (deband, grain, S-curve with a 0.94 highlight shoulder, saturation), freeze the chain into a Hald LUT for future videos, and verify the delivery file with ffprobe plus signalstats. It includes a face-cam exception and a worked QC example with real probe output from a synthetic palette chart.

Who it is for

An editor or agent doing the final encode of a video built from rendered graphics or screen content, who wants the same look on every video and a numeric check instead of eyeballing it.

How to use

  1. Read SKILL.md — prerequisites, input spec, master chain, LUT procedure, QC gate.
  2. Assemble your edit losslessly (libx264 -crf 0 or ffv1) at delivery resolution and frame rate.
  3. Run the master chain at final encode (copy the command from SKILL.md; substitute your input/output filenames).
  4. Run the QC gate commands on the delivery file. All four colour tags must read correctly and the signalstats numbers must match your reference row.
  5. See examples/qc-worked-example.md for the full worked example — the exact commands used to build a synthetic test chart, the real probe output, and what each number means.

Example

The QC worked example (examples/qc-worked-example.md) shows a real run on ffmpeg 8.1.2: ungraded input YMIN=25, YMAX=224 → graded output YMIN=17, YMAX=217, all four colour tags reading bt709/tv. Those numbers are an illustration of the procedure on a synthetic chart, not targets to copy — your reference row comes from your own first approved video.

Install

Copy this folder into your skills directory and reference it by name at final encode. Requires ffmpeg 8.1.2 or later (with ffprobe). No build step, no API key, no account.

License

MIT — see the license in the skill frontmatter.

Disclaimer

The chain is tuned for graphics-led short-form video with a dark palette. It does not colour-correct raw or log camera footage, and the grain/shoulder rationale in the skill is the author's working rationale, not a cited published finding. ffmpeg behaviour varies by version — re-verify the colour-tag behaviour on your own build. YouTube bitrate figures are cited from YouTube's published "recommended upload encoding settings" help page (https://support.google.com/youtube/answer/1722171), checked 2026-11-16.

Source: SKILL.md on GitHub

skilld matched fixed text patterns in SKILL.md and file names. Patterns miss obfuscated code.

skilld run checks every file with the same patterns. It asks for approval before it loads a Skill with a behavior marked Needs approval.

No third-party reports yet.

Signed by skilld at 90f40b1. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub 4 days ago.

Activeupdated 5 days ago
Other metadata
metadata
{
  "version": "1.0.0",
  "date": "2026-11-16",
  "author": "Muhammad Saad Ranjha / NeedCraft"
}

README badge

README badge for jokerhtml/needcraft-skills/color-grading