Color Grading logo

Color Grading

Organization
palmier-io
color-grading

Color grade clips the way a professional colorist does — normalize, balance, match, then build a look — using inspect_color scopes and apply_color (exposure, white balance, wheels, curves, hue curves, LUTs). Use when the user asks to color, grade, match shots, fix white balance, fix exposure, neutralize a cast, or apply a look/film stock.

Overview

Publisherpalmier-io
Repositorypalmier-skills
Skill namecolor-grading
Stars
82
Forks
10
Bundled files
Instructions only
LicenseApache-2.0
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • Self-contained

    Everything the model needs lives in the instructions — no extra files to sync.

  • Open source

    Published by palmier-io on GitHub. Read the source before you install it.

Installation

Install the Color Grading AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/palmier-io/palmier-skills/tree/main/skills/color-grading
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/palmier-io/palmier-skills.git /tmp/palmier-skills
mkdir -p .claude/skills
cp -r /tmp/palmier-skills/skills/color-grading .claude/skills/color-grading
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Color Grading in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Color Grading on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Color Grading is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Grade in the colorist's order, scopes-first — never skip to a look, since a look on top of a cast just bakes it in. One apply_color call merges onto the current grade (only the params you pass change), so build up across calls and verify with inspect_color between stages. Ranges live in the reference tables below; prose names the knob, not the number.

Workflow

- [ ] 1. inspect_color — read scopes BEFORE touching anything
- [ ] 2. Normalize — exposure + white balance to neutral
- [ ] 3. Primary — black/white points, contrast, saturation
- [ ] 4. Match — pull each shot toward the reference (if matching)
- [ ] 5. Secondary — hue curves for skin/sky/foliage
- [ ] 6. Look — wheels split-tone and/or LUT
- [ ] 7. inspect_color again — confirm scopes moved as intended

1. Read first. inspect_color(clipId). Check luma.black/luma.white (near 0/1? clipLowPct/clipHighPct flag crush/clip), balance.warmCool + balance.greenMagenta (nonzero = a cast to neutralize), zones (where the cast lives), and hueHistogram12 (orange bins 1–2 = skin, cyan/blue bins 6–8 = sky).

2. Normalize. exposure for global brightness (prefer it over gamma). temperature to correct warmCool — higher = warmer, so lower it if too warm. tint to correct greenMagenta, pushing opposite the cast. Re-inspect: both balances near 0 before moving on.

3. Primary. blacks/whites to set the points, contrast + highlights/shadows to shape roll-off, saturation for overall sat — reach for vibrance instead when you want to boost muted color without clobbering skin.

4. Match shots. Grade the hero fully, then for each other shot call inspect_color(clipId, reference: <hero>), read the returned gap and its hints (e.g. "warmer than ref → cooler temperature"), apply the nudges, and repeat until the gap is small. Close it numerically — don't eyeball it.

5. Secondary. Adjust specific hues via hueCurves.targets (each targetHue affects ~±22°): hueShift rotates, satScale mutes/pops, lumShift darkens/brightens — e.g. deepen a sky without touching skin. Per-channel redCurve/greenCurve/blueCurve are tone-selective: tame a blown sky with blueCurve: [[0,0],[0.7,0.7],[1,0.85]].

6. Look. Split-tone with the wheels (*Hue+*Amount per zone, e.g. teal/orange = shadowsHue:210,shadowsAmount:0.2 + highsHue:30,highsAmount:0.15), add a film toe via masterCurve ([[0,0.06],[1,0.95]]), or fold in a lut:{path,strength} (.cube, absolute path; copied into project storage; pass strength alone to re-blend). For named stocks, build by hand — see Film looks below.

7. Confirm. Re-inspect_color: points near 0/1 without heavy clipping, neutral balance unless intentionally toned, skin in the orange bins.

apply_color reference

Merges onto the current grade; reset:true starts neutral. clipIds required (video/image only).

GroupParamsRangeNotes
Toneexposure / contrast−3…3 EV / 0.5–1.51 = neutral contrast
highlights shadows blacks whites−1…1points & roll-off
White balancetemperature / tint2000–11000 K / −100…100higher = warmer; +tint green, −magenta
Saturationsaturation / vibrance0–2 / −1…11 = neutral; vibrance protects skin
Wheels (per zone)shadowsHue/midsHue/highsHue + matching *Amount0–360° / 0–1hue push + strength
shadowsLum / midsGamma / highsGain−0.5…0.5 / 0.5–2 / 0.5–1.5per-zone brightness
CurvesmasterCurve redCurve greenCurve blueCurve[[x,y],…] 0–1master preserves chroma
Hue curveshueCurves.targets[] = {targetHue, hueShift, satScale, lumShift}±30° / 0–2 / ±0.5~±22° selectivity
LUTlut = {path, strength}strength 0–1.cube, abs path

Hues: 0 red · 30 orange/skin · 60 yellow · 120 green · 180 cyan · 210 sky-blue · 240 blue · 300 magenta.

inspect_color(clipId | mediaRef, atFrame?, reference?) — clipId reads a clip's graded look, mediaRef a raw asset. Returns scopes (luma, meanRGB, zones, saturation, balance, hueHistogram12, colorfulPct) + the frame; with reference, also a gap and actionable hints (the backbone of shot matching).

Film looks (no LUT)

A LUT just freezes a set of curve/wheel/saturation moves — build any of them by hand, in this order:

  1. Tone curve (toe + shoulder). The biggest tell: film never hits pure black or clips white. Lift the masterCurve toe and roll off its shoulder with gentle mid-S, e.g. [[0,0.04],[0.25,0.22],[0.75,0.82],[1,0.97]].
  2. Per-channel crosstalk. Hue shifts with brightness — what separates real emulation from "S-curve + a tint." Lift the blue toe and pull its shoulder → teal shadows + warm highlights (blueCurve: [[0,0.05],[0.5,0.5],[1,0.92]]); nudge green mids for Fuji.
  3. Split-tone the wheels (shadows→teal, highlights→amber).
  4. Saturation sculpting — vibrance over flat saturation, plus hueCurves satScale to mute/boost individual hues.
  5. Hue rotation — small hue-vs-hue shifts (≤8°): Kodak warms oranges, Fuji greens.

The recipes below are this method with the dials pre-set — apply onto an already-normalized clip, then re-inspect_color and tune toward the shot (scale *Amount and curve deviations toward neutral for less intensity). If the user has the real .cube, prefer lut:{path,strength:0.85}; these approximate a stock's character, not its science.

Kodak 2383 — Hollywood print (warm, contrasty; dense blacks, teal shadows, amber highlights):

json
{ "contrast": 1.25, "saturation": 1.15, "vibrance": 0.1, "temperature": 6800, "blacks": -0.1, "whites": 0.05,
  "shadowsHue": 195, "shadowsAmount": 0.12, "highsHue": 40, "highsAmount": 0.12, "masterCurve": [[0,0.02],[0.25,0.21],[0.75,0.83],[1,0.98]] }

Kodak 2389 — low-con print (clean, neutral; filmic roll-off, little tone):

json
{ "contrast": 1.08, "saturation": 1.03, "temperature": 6550, "blacks": -0.04, "highsHue": 45, "highsAmount": 0.06, "masterCurve": [[0,0.015],[0.3,0.29],[0.7,0.72],[1,0.99]] }

Fuji Eterna / 3513-like (soft, green-pastel; milky lifted blacks). The lifted toe is the signature — don't flatten it:

json
{ "contrast": 1.05, "saturation": 0.95, "vibrance": 0.12, "temperature": 6300, "blacks": 0.08, "shadows": 0.1, "highlights": -0.15, "whites": -0.04,
  "midsHue": 120, "midsAmount": 0.08, "shadowsHue": 185, "shadowsAmount": 0.1, "masterCurve": [[0,0.06],[0.25,0.27],[0.8,0.82],[1,0.95]] }

Kodak Portra — warm negative (soft, skin-flattering). The −tint + orange curve flatters skin — confirm the orange bins grew, not the reds:

json
{ "contrast": 1.05, "saturation": 1.0, "vibrance": 0.15, "temperature": 6900, "tint": -3, "blacks": 0.06, "highlights": -0.1, "highsHue": 38, "highsAmount": 0.08,
  "hueCurves": { "targets": [ { "targetHue": 30, "satScale": 1.08, "hueShift": -4 } ] } }

Bleach bypass (desaturated, high-contrast, metallic). Scale saturation 0.4–0.7 for intensity:

json
{ "contrast": 1.35, "saturation": 0.55, "temperature": 6300, "tint": 5, "blacks": -0.15, "whites": 0.1, "masterCurve": [[0,0.0],[0.2,0.13],[0.8,0.9],[1,1.0]] }

Teal & orange — blockbuster (cool shadows, warm skin). Keep the orange curve modest so faces don't go radioactive:

json
{ "contrast": 1.2, "saturation": 1.1, "vibrance": 0.15, "shadowsHue": 210, "shadowsAmount": 0.2, "highsHue": 30, "highsAmount": 0.15,
  "hueCurves": { "targets": [ { "targetHue": 30, "satScale": 1.15 } ] } }

Vintage / faded (milky blacks, low sat, warm-yellow age). The high toe is the whole look — blacks never reach 0:

json
{ "contrast": 0.95, "saturation": 0.8, "temperature": 6700, "tint": 4, "blacks": 0.12, "highlights": -0.1, "masterCurve": [[0,0.1],[0.5,0.52],[1,0.92]] }

Notes

  • apply_color is the colorist path and folds in LUT + curves + hue curves as live, editable effects. apply_effect is for generic, non-color FX — don't grade with it.
  • get_timeline exposes each graded clip's color object in the same vocabulary — pass that object as apply_color({ clipIds, color }) to copy a whole grade (replaces; mutually exclusive with knobs/reset).
  • Mutations return timeline deltas (including updated color) — patch from those; don't re-read the whole timeline after every nudge.
  • Trust scopes over the preview for balance — monitors and JPEGs lie about subtle casts; use the frame for composition/look judgment.
  • lumShift is an additive HSV value shift in display space, not a linear luma divide, so per-hue darkening won't amplify shadow macroblocks.
  • Curve points are piecewise-linear, clamped flat outside their range — include endpoints near [0,0]/[1,1] unless you mean to clip.

Frequently asked questions

What does the Color Grading AI skill do?

Color grade clips the way a professional colorist does — normalize, balance, match, then build a look — using inspect_color scopes and apply_color (exposure, white balance, wheels, curves, hue curves, LUTs). Use when the user asks to color, grade, match shots, fix white balance, fix exposure, neutralize a cast, or apply a look/film stock.

Why use Color Grading on TypingMind?

Because you install it once and use it with any model. Color Grading is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Color Grading in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/palmier-io/palmier-skills/tree/main/skills/color-grading. TypingMind reads its SKILL.md and installs it as a skill you can enable per chat.

Which AI models can use Color Grading?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Color Grading?

As many as you like. As long as a model supports skills, you can use Color Grading with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Color Grading AI skill free?

Yes. It is published on GitHub by palmier-io under the Apache-2.0 license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇