Seaborn Color Palettes: Interactive Explorer and Selection Guide

Name

Soren Atelier

Updated on 8/19/2026

A seaborn color palette is an ordered list of colors that seaborn hands to your plots — one color per category for bar and line charts, or a continuous ramp for heatmaps. You change it in one line:

import seaborn as sns
 
sns.set_palette("colorblind")            # global default for every plot after this
sns.barplot(data=df, x="day", y="total", hue="sex", palette="deep")   # one plot only
sns.heatmap(corr, cmap="vlag", center=0)  # continuous data uses cmap, not palette

Three things to remember before the details:

• palette= takes discrete colors and belongs on categorical plots (barplot, boxplot, lineplot with hue).
• cmap= takes a continuous colormap and belongs on heatmap, kdeplot, and other density plots.
• Seaborn's default is deep. If the figure is going to be published or shared, switch it to colorblind.

The explorer below shows every named palette with its actual hex values. Click a color bar to copy the call.

Quick syntax reference

The decision table

This is the part people usually get wrong: matching the palette type to the data, not just picking a palette that looks nice.

Two traps worth calling out:

Diverging without centering. sns.heatmap(corr, cmap="vlag") on data from -0.2 to 0.9 puts the neutral color at 0.35, so weak positive correlations look "neutral". Always pass center=0:

sns.heatmap(corr, cmap="vlag", center=0, vmin=-1, vmax=1, annot=True)

Sequential for categories. palette="Blues" on a five-category bar chart tells the reader that the darkest bar is "most" of something. If the categories have no order, use deep or colorblind.

The six named qualitative palettes

Seaborn's deep, muted, pastel, bright, dark, and colorblind all use the same hue order. Only saturation and lightness differ, so you can swap between them without remapping which category gets which hue.

import seaborn as sns
 
for name in ["deep", "muted", "pastel", "bright", "dark", "colorblind"]:
    print(name, sns.color_palette(name).as_hex()[:3])

deep       ['#4c72b0', '#dd8452', '#55a868']
muted      ['#4878d0', '#ee854a', '#6acc64']
pastel     ['#a1c9f4', '#ffb482', '#8de5a1']
bright     ['#023eff', '#ff7c00', '#1ac938']
dark       ['#001c7f', '#b1400d', '#12711c']
colorblind ['#0173b2', '#de8f05', '#029e73']

Practical selection:

• deep — the default. Fine for exploration and internal decks.
• muted — many overlapping series, or large filled areas where full saturation is tiring.
• pastel — filled shapes that carry dark text or markers on top.
• bright — projectors and dark backgrounds.
• dark — thin lines on a light background.
• colorblind — anything you publish. Derived from the Okabe-Ito set, so it survives the common forms of color vision deficiency.

Seaborn's own continuous palettes

Beyond Matplotlib's colormaps, seaborn registers four of its own:

The flare / crest distinction matters more than it sounds: rocket and mako bottom out near black, which is fine for a filled heatmap cell but unreadable as a thin line on a white page.

import seaborn as sns
 
# heatmap: the dark end is fine
sns.heatmap(matrix, cmap="rocket")
 
# ordered lines: use flare so no series is nearly black
sns.lineplot(data=df, x="month", y="value", hue="cohort", palette="flare")

Building a custom palette

When brand colors or a specific hue are required, build the ramp instead of hunting for a named match.

import seaborn as sns
 
# White -> your color
sns.light_palette("seagreen", as_cmap=True)
 
# Near-black -> your color
sns.dark_palette("#69d", as_cmap=True)
 
# Custom diverging map from two hue angles (0-359)
sns.diverging_palette(220, 20, as_cmap=True)
 
# Grayscale-safe sequential ramp
sns.cubehelix_palette(start=.5, rot=-.75, as_cmap=True)
 
# Explicit list — full control
brand = ["#0f4c81", "#e8743b", "#19a979", "#945ecf"]
sns.set_palette(sns.color_palette(brand))

To pin specific categories to specific colors — so "churned" is always red across every chart in a report — pass a dict:

status_colors = {"active": "#0173b2", "churned": "#d55e00", "trial": "#de8f05"}
sns.barplot(data=df, x="month", y="users", hue="status", palette=status_colors)

Common errors

Related Guides

• Matplotlib Colormaps: Interactive Explorer — the full Matplotlib gallery, since every Matplotlib colormap name also works in seaborn.
• Seaborn Heatmap — where cmap and center decide whether the figure is readable.
• Seaborn Barplot — the most common home for a qualitative palette.
• Seaborn Lineplot — multi-series color with hue and palette.
• Seaborn vs Matplotlib — how the two libraries divide styling responsibilities.
• Matplotlib Colormap: Complete Guide — normalization, colorbars, and custom colormap construction.