Which Okabe-Ito colors should I pick when I only have two or three categories?

For two categories, pick orange (`#E69F00`) and blue (`#0072B2`) — they give the widest warm-cool separation and survive every common color vision deficiency. For three, add bluish green (`#009E73`). For four, add vermilion (`#D55E00`). Reserve yellow (`#F0E442`) for last; on a white background it has only a `1.32:1` WCAG contrast and disappears in thin lines. The Recommended subsets section above shows the full ordering up to seven groups.

Why does yellow (#F0E442) look so faded on Okabe-Ito plots?

Yellow's WCAG contrast against white is only `1.32:1`, far below the `4.5:1` AA threshold, so thin strokes and small labels lose definition even on a calibrated screen. Two fixes work: give yellow shapes a dark border (`#000000` at 1 px is enough), or swap yellow for a darker Okabe-Ito color when the geometry is thin (Vermilion `#D55E00` or Blue `#0072B2` are usually fine drop-ins).

Is the Okabe-Ito palette built into matplotlib, seaborn, or base R by default?

Not really. **matplotlib** and **seaborn** ship their own defaults (`tab10`, `deep`), so you have to set the cycle yourself with `plt.rcParams["axes.prop_cycle"]` or `sns.set_palette()` as shown in the code sections above. **Base R 4.0+** does include it: `palette.colors(palette = "Okabe-Ito")` returns the eight HEX codes. **ggplot2** does not ship a built-in `scale_color_okabe_ito()`; use `scale_color_manual()` with the vector, or install the `ggokabeito` package.

How do I verify that my Okabe-Ito figure is actually readable under color blindness?

Run the exported figure through a CVD simulator before submission. The most-used tools are **Color Oracle** (free, macOS / Windows / Linux desktop overlay), **Coblis** (web), and **Sim Daltonism** (macOS). All three simulate protanopia, deuteranopia, and tritanopia in real time. The Okabe-Ito palette is designed to survive these simulations, but only if you keep marker shapes, line styles, or labels varying alongside color — pure-color encoding can still fail when two categories happen to land on similar luminance.

What is the difference between the Okabe-Ito palette and Viridis?

Okabe-Ito is **categorical** — eight discrete hues optimized for distinguishing groups. Viridis is **sequential** — a continuous gradient optimized for ordered numeric values like expression intensity or density. Using Okabe-Ito for a heatmap makes continuous data look segmented; using Viridis for unordered groups suggests a magnitude that doesn't exist. Pick by data type, not by appearance.

Can I use the Okabe-Ito palette for volcano plots?

Yes — the Okabe-Ito palette is a strong default for volcano plots. The conventional mapping uses vermilion (`#D55E00`) for upregulated genes, blue (`#0072B2`) for downregulated, and a neutral gray (`#999999`) or black for non-significant points. All three pairs stay separable under deuteranopia and protanopia, which is hard to achieve with the typical red / green / gray scheme. See our 3-minute volcano plot guide or generate one directly with the volcano plot tool.

Is the Okabe-Ito palette the same as the Wong colorblind palette?

Yes — the two names refer to the same eight HEX codes. Masataka Okabe and Kei Ito published the palette under the Color Universal Design framework; Bang Wong then featured it in *Nature Methods* 8(6):441 (2011), which is why "Wong colorblind palette" became a common search term in life-science circles. There is no separate "Wong" set with different colors.

← Back to Blog

Scientific DesignMay 23, 2026· 10 min read

Okabe-Ito Palette HEX Codes for Scientific Figures

Copy all 8 Okabe-Ito / Wong colorblind palette HEX and RGB codes, with Prism, Python, R, MATLAB, CSS, contrast notes, and figure-ready subsets.

Colorblind-safe palette

Okabe-Ito Palette

#E69F00

#56B4E9

#009E73

#F0E442

#0072B2

#D55E00

#CC79A7

#000000

Ready to create your first figure?

Describe what you need. Get a publication-ready figure in minutes.

Generate a data plot with Okabe-Ito in FigCanvas