Colors & palettes

Colors appear throughout the Felt Style Language — in paint (color, strokeColor), in label (color, haloColor), and in legends. There are three ways to express a color:

1. A literal color — a hex, HSL, or RGB string.

2. A smart color — the keyword "auto", which Felt resolves to a contrasting color at render time.

3. A palette shortcut — a named, multi-color ramp like @galaxy, used by data-driven (categorical, numeric, heatmap, H3) visualizations.

Literal colors

Any of these string formats are accepted wherever a single color is expected:

{
  "color": "#3B82F6",
  "strokeColor": "hsl(217, 80%, 40%)",
  "haloColor": "rgb(38, 113, 0)"
}

rgba(...) is also supported and is handy for fully-transparent fills (for example, a transparent "Other"/background class in a categorical raster: "rgba(0, 0, 0, 0)").

Smart color: "auto"

"auto" computes a contrasting color automatically, based on the current map theme and the colors around it. It keeps strokes and label halos legible without hard-coding a value that might clash on a dark basemap or against a particular fill.

Prefer "auto" for:

• strokeColor on points and polygons

• color and haloColor on labels

…unless a specific color has been requested.

{
  "paint": {
    "color": "#8F7EBF",
    "strokeColor": "auto",
    "strokeWidth": 1
  },
  "label": {
    "color": "auto",
    "haloColor": "auto",
    "haloWidth": 1.5
  }
}

Palette shortcuts

For any data-driven visualization, prefer a named palette over a hand-built color array. Palettes are referenced with an @ prefix (for example "color": "@galaxy") and Felt expands them to the right number of colors for your classes or categories.

Only the palette names listed below are valid. Use sequential palettes for single-direction data (population, elevation), and diverging palettes for data with a meaningful midpoint (change from a baseline, temperature anomaly). Reserve literal hex/HSL colors for simple (uniform) styles or when a specific color is requested.

Vector palettes

Used by categorical, numeric, heatmap, and H3 visualizations on points, lines, and polygons.

Sequential (low → high)

Palette	Description
@galaxy	Purple to yellow (popular default)
@ylRed	Yellow to red
@ylGrn	Yellow to green
@purpYl	Purple to yellow
@pinkYl	Pink to yellow
@lightning	Lightning gradient
@copper	Copper gradient
@spruce	Spruce green
@riverine	Water gradient
@neptune	Blue gradient
@violet	Violet gradient
@purple	Purple gradient

Diverging (two directions from a center)

Palette	Description
@bluRd	Blue to red
@tealOr	Teal to orange (colorblind-safe)
@bluBr	Blue to brown
@grnOr	Green to orange
@purGrn	Purple to green
@weath	Weather gradient

Categorical (distinct colors)

Palette	Description
@catPalette1 … @catPalette7	Categorical color schemes
@catPalettePT1	Paul Tol categorical palette 1
@catPalettePT2	Paul Tol categorical palette 2

Sequential and diverging vector palettes also come in numbered variants that request a specific number of colors — for example @galaxy7, @galaxy8, @galaxy9. Variants exist for 7, 8, and 9 colors (@lightning also has a 6-color variant). The unnumbered name lets Felt pick the right count for you, which is usually what you want.

Heatmap palettes

Used by heatmap visualizations, ordered from least to most density.

Palette	Description
@geyser	Green → beige → orange → red (default)
@redOrHeat	Yellow → orange → red → magenta
@purpYlHeat	Light yellow → coral → pink → purple
@lightningHeat	Purple → teal → green → yellow
@ylGrnHeat	Dark teal → green → yellow
@bluRdHeat	Blue → gray → red (diverging feel)
@tealRedHeat	Teal → green → yellow → orange → pink
@purpYlPink	Purple → teal → yellow → orange → red/pink

Raster palettes

Used by numeric, hillshade, and raster-algebra visualizations.

Sequential

Palette	Description
@feltGrays	Grayscale gradient
@cbBlues	ColorBrewer blues
@cbPurples	ColorBrewer purples
@feltPinks	Pink gradient
@cmOceanGreens	Ocean greens gradient
@pattFeltOcean	Teal ocean gradient
@cmOceanDeep	Deep ocean gradient
@mpaInferno	Matplotlib inferno
@mplPlasma	Matplotlib plasma
@mplVirdis	Matplotlib viridis — perceptually uniform
@cividis	Cividis — colorblind-safe
@feltYlRed	Felt yellow to red
@veg	Vegetation gradient

Diverging

Palette	Description
@feltWeath	Felt weather gradient
@feltHeat	Felt heat gradient
@feltVibrant	Felt vibrant gradient
@nclBlOrRed	NCL blue-orange-red diverging
@cbRedYlGrn	ColorBrewer red-yellow-green diverging
@cbBrBG	ColorBrewer brown-blue-green diverging

Terrain

Palette	Description
@terrain	Terrain elevation gradient
@rTerrain	R terrain gradient
@feltHypso	Felt hypsometric tints
@wikiTerrain	Wikipedia terrain gradient

Vegetation

Palette	Description
@nasaNDVI	NASA NDVI vegetation index

For raster categorical layers, use the vector categorical palettes above (@catPalette1 … @catPalette7, @catPalettePT1, @catPalettePT2).

Raster palettes also support numbered variants to request a specific number of steps — for example @feltGrays2, @feltGrays3, … @feltGrays9.

How colors map to classes

When a palette or color array drives a data-driven visualization, the number of colors should line up with the number of classes or categories:

• Categorical: one color per category. With showOther: true and an explicit categories array, provide N + 1 colors — the extra one styles the "Other" bucket.

• Classed numeric: N break points produce N − 1 classes, so an explicit color array needs N − 1 entries.

• Continuous: provide 2 or more colors and Felt interpolates smoothly between them in the HCL color space.

A palette shortcut handles all of this for you — Felt expands @galaxy to exactly the colors it needs. Reach for explicit arrays only when you want precise control over each color.