Okabe-Ito Palette Hex Codes & 5 Colorblind-Safe Palettes for Science (2026)

Colorblind-Safe Scientific Palettes: Copy-Ready Hex Codes

Short on time? Scroll to the table just below for the Okabe-Ito colorblind-safe palette hex codes, copy them straight into your figure, and move on. The rest of this guide unpacks Viridis, Paul Tol, Wong, and the other Nature-endorsed schemes, and walks through how to match a palette to the kind of data you are actually plotting.

Scientific Color Palette Generator

Generate colorblind-safe palettes for your research figures. Export hex codes, RGB values, and more.

Try it free →

Okabe-Ito Palette: Full Hex Code Table

When researchers ask which categorical scheme to trust for publishing, the Okabe-Ito palette is the usual answer. Masataka Okabe and Kei Ito assembled these colors with color vision deficiency in mind, Nature Methods points to the same set (there it carries the "Wong palette" label), and Claus Wilke ships it as the built-in default throughout "Fundamentals of Data Visualization." That combination is why it has become the reference point for distinguishing unordered groups in a figure.

Complete Okabe-Ito Hex Codes at a Glance

What Makes the Okabe-Ito Palette Reliable

The eight swatches were chosen so that none of them collapse into one another under the major types of color vision deficiency (CVD) (protanopia, deuteranopia, and tritanopia). Since something on the order of one in twelve men sees color atypically, treating accessibility as optional means quietly losing part of your audience on every figure.

Why the set earns its reputation:

• Accessibility was a design constraint up front, not a patch applied after the colors were already chosen
• The hues differ in brightness as well as tint, which keeps them readable once a figure is printed in black and white
• It carries the backing of Nature, Science, and other leading peer-reviewed journals
• Drop-in support exists across Python (matplotlib), R (ggplot2), MATLAB, and essentially any plotting library you are likely to reach for

Code Snippets for Immediate Use

Python (matplotlib):

okabe_ito = ['#E69F00', '#56B4E9', '#009E73', '#F0E442',
             '#0072B2', '#D55E00', '#CC79A7', '#000000']

R (ggplot2):

okabe_ito <- c("#E69F00", "#56B4E9", "#009E73", "#F0E442",
               "#0072B2", "#D55E00", "#CC79A7", "#000000")
ggplot(data) + scale_color_manual(values = okabe_ito)

CSS / Web:

--okabe-orange: #E69F00;
--okabe-sky-blue: #56B4E9;
--okabe-green: #009E73;
--okabe-yellow: #F0E442;
--okabe-blue: #0072B2;
--okabe-vermillion: #D55E00;
--okabe-purple: #CC79A7;
--okabe-black: #000000;

Why Color Decisions Shape Scientific Outcomes

Treating color as decoration is a mistake. The palette you pick governs how fast and how accurately a reader extracts your finding, and occasionally it is the difference between a manuscript that passes review and one that comes back for revision.

Reviewers and readers tend to look at the figures before they read a single sentence of the body text. Work in PLOS ONE and eLife has linked papers carrying clear, legible figures to higher citation counts down the line.

A signaling pathway laid out with an even, accessible palette. The colors pull each element apart without breaking the figure's overall balance.

Where the palette decision actually shows up:

The Three Categories of Scientific Color Palettes

The first decision is not which colors to use but which family of palette your data belongs to. Get that wrong and even beautiful colors will misrepresent the numbers. Three families cover almost everything.

1. Sequential Palettes

Ideal for: Quantities that move in one direction only (temperature, concentration, signal intensity)

These ramp steadily from pale to deep, or from washed-out to saturated, so the visual progression mirrors a value climbing from its low end to its high end. Reach for them whenever the data has an inherent order.

Examples: Viridis, Magma, Inferno, Plasma, Blues, Greens

Typical jobs:

• Heatmaps of gene expression
• Choropleths showing population density
• Fluorescence intensity in microscopy
• Essentially any single-variable ramp from low to high

2. Diverging Palettes

Ideal for: Quantities that pivot around a meaningful zero (gain vs. loss, above vs. below a baseline)

Here a muted color sits at the middle and two opposing hues stretch out to either side, so the reader instantly sees which way, and how far, a value departs from its reference point.

Examples: Red-Blue, Purple-Green, Brown-Teal

Typical jobs:

• Values sitting above or below a cutoff
• Correlation matrices spanning positive and negative
• Deviations measured against a control condition
• Climate anomalies relative to a long-run mean

3. Qualitative (Categorical) Palettes

Ideal for: Telling apart groups that have no inherent ranking

These rely on hues that read as clearly separate yet imply no order, with every swatch carrying roughly equal visual weight so no single group steals focus.

Examples: Okabe-Ito, Paul Tol, Set1, Paired

Typical jobs:

• Separate treatment arms in an experiment
• Distinct cell types, species, or tissues
• Categorical splits in bar and scatter plots
• Legend entries that each need their own identity

A cell diagram leaning on categorical colors to keep the organelles apart. Every structure gets its own hue, so a reader can name it at a glance.

5 Additional Palettes Recommended for Scientific Work

Okabe-Ito gets cited most, yet it is far from the only carefully engineered choice. The schemes below are just as defensible, and each comes with hex values you can drop straight into your code.

1. Viridis Family (Top Choice for Sequential Data)

Stéfan van der Walt and Nathaniel Smith built the Viridis color maps for Matplotlib to solve a specific set of problems at once:

• Perceptually uniform: equal jumps in the underlying value look like equal jumps in color
• CVD-friendly: they stay legible across the range of color vision types
• Grayscale-safe: nothing important is lost when the figure goes to black and white

You will find all of these shipped with Python (matplotlib), the R viridis package, MATLAB, and the large majority of current plotting tools.

2. Scientific Colour Maps by Fabio Crameri

Fabio Crameri's Scientific Colour Maps were designed from the start around the way scientific data should be shown. The set runs to over 50 maps, and every one of them clears the same four bars:

• A perceptually even progression
• No breakdown under color vision deficiency
• Still legible after a grayscale conversion
• Free to download for whatever software you run

Common picks from the collection:

• batlow: a Blue to Yellow to Red ramp that works as an all-round sequential map
• roma: a diverging map that stays CVD-safe end to end
• vik: Blue to White to Red, for diverging data

Packaged downloads cover MATLAB, Python, R, GMT, QGIS, Paraview, Illustrator, and plenty of other tools.

3. Paul Tol's Color Schemes

Written by a working scientist for the same crowd, Paul Tol's schemes span every one of the three families in a single, consistent system. The lineup includes:

• Bright: a punchy categorical set of 7 colors
• Vibrant: a more saturated categorical set, also 7 colors
• Muted: a calmer categorical set running to 9 colors
• Light: a pastel categorical set, 9 colors
• Sunset / BuRd: a pair of diverging choices
• YlOrBr / Iridescent: sequential maps for gradient data

4. ColorBrewer Palettes

Born in cartography, ColorBrewer crossed over into scientific figure work years ago and is now a default reference. What it offers:

• A live web tool with toggles for CVD safety and print suitability
• Coverage of all three families, neatly separated into sequential, diverging, and qualitative
• Export in a handful of formats

Dependable ColorBrewer choices:

• Sequential: YlGnBu, PuBuGn, BuPu
• Diverging: RdBu, PiYG, PRGn
• Qualitative: Set2, Paired, Dark2

5. Bang Wong / Nature Methods Palette

The palette Bang Wong published in Nature Methods is, color for color, the same eight values as the Okabe-Ito set covered earlier, and Nature journals name it outright as their recommendation for any scientific figure. Rather than repeat the values here, see the full hex and RGB listing in the Okabe-Ito section near the top of the page.

A reaction mechanism where well-separated colors let the reader trace electron movement step by step, with no guesswork.

Designing Figures That Work for Colorblind Readers

Color vision deficiency touches roughly 8% of men and 0.5% of women. Put fifty scientists in a seminar room and around four of them will not read a red-green encoding the way you meant it.

Color Pairings to Avoid

Safe Combinations That Work Across CVD Types

These pairings hold their distinction for nearly every type of color blindness:

1. Blue + Orange: the most dependable pairing there is, clear to essentially everyone
2. Blue + Red: plenty of contrast and reliable across most CVD profiles
3. Blue + Yellow: a wide brightness gap keeps it CVD-compatible
4. Purple + Yellow: bold to look at and still accessible

Adding Redundant Encoding for Many Categories

Once a figure carries more than three or four categories, lean on redundant encoding: back up each color with a second visual channel so the message survives even with the color stripped out.

• Shapes: circles, squares, triangles, diamonds
• Line styles: solid, dashed, dotted
• Fill patterns: solid, hatched, crosshatched
• Point sizes: larger and smaller markers as an extra cue

Do this and the figure stays readable in grayscale and for readers anywhere on the CVD spectrum.

A photosynthesis diagram showing color hierarchy at work: the core processes carry the bold colors while supporting parts sit back in gentler tones.

What Major Journals Require

Nature Journals

Pulling from Nature's figure guidelines:

Color Mode:

• Hand in figures as RGB, not CMYK
• RGB covers a wider gamut for on-screen viewing
• The journal takes care of the CMYK conversion at the print stage

Accessibility Standards:

• Red-green pairings are called out as something to steer clear of
• Rainbow scales are advised against
• The stated preference is an accessible scheme such as Wong 2011
• For fluorescence, swap red + green for green + magenta

Practical Specifications:

• Single column: 89 mm wide
• Double column: 183 mm wide
• Type set at 5 to 7 pt once the figure is at final size
• A floor of 300 DPI for raster images

Science Magazine

Science lands in much the same place:

• It expects colorblind-friendly palettes
• Continuous data is meant to ride on perceptually uniform maps
• Raw red versus green should never be the only thing carrying a critical comparison

Cell Press

Cell titles stress a slightly different short list:

• Plenty of contrast where colored regions meet
• One color scheme held steady across every figure in the manuscript
• Legends that are clearly labeled and include the actual color swatches

Common Palette Mistakes and How to Fix Them

Mistake 1: Using Rainbow or Jet Color Maps

What goes wrong: A rainbow ramp (red through yellow, green, and on to blue) is not perceptually even. Its yellow and cyan stretches read brighter than what sits beside them, so the eye perceives sharp "bands" in data that is actually smooth.

How to fix it: Drop in Viridis, Magma, or any other perceptually uniform map instead.

Mistake 2: Using Too Many Distinct Colors

What goes wrong: Push past ten separate colors and readers simply cannot keep track of which color stands for which category.

How to fix it:

• Hold categorical colors to six or under whenever you can
• Fold related items together under one color
• Lean on shades of a single hue when you have subcategories to show

Mistake 3: Inconsistent Colors Across Figures

What goes wrong: If "Control" is blue in Figure 1 but red in Figure 2, the reader has to relearn the legend every time they turn the page.

How to fix it: Lock in a color key at the outset of the project and reuse it for every figure in the paper.

Mistake 4: Insufficient Luminance Contrast

What goes wrong: Pale colors against white, or neighboring hues of similar brightness, blur together in print and on some displays.

How to fix it:

• Preview the figure in grayscale before you call it done. If shapes blend together there, some readers will hit the same wall in full color.
• Confirm there is enough luminance gap between any two adjacent elements.

Mistake 5: Colored Text Labels

What goes wrong: Text in color is harder to read than plain black, and a color shift on press can leave a label barely legible.

How to fix it: Keep the text black and tie each label to its data point with a colored connector line or a bordered box.

Tools for Palette Selection and CVD Testing

Palette Selection Resources

Colorblind Simulation Tools

Software-Specific Implementation Notes

Python:

import matplotlib.pyplot as plt
# Viridis is the default; specify alternatives as needed
plt.imshow(data, cmap='viridis')  # or 'magma', 'plasma', 'inferno'

R:

library(viridis)
ggplot(data) + scale_color_viridis()

library(RColorBrewer)
display.brewer.all(colorblindFriendly = TRUE)

GraphPad Prism: Right-click on the graph, select "Define color scheme," then choose a colorblind-safe option.

Palette Decision Guide

Not sure which family fits? Follow the branches below and the dataset will point you to its palette type:

Is your data categorical (named groups or types)?
├── YES → Use a Qualitative palette (Okabe-Ito, Paul Tol)
│         └── More than 6 categories? Add shapes or line patterns.
│
└── NO → Is your data continuous?
         ├── Does it have a meaningful center point?
         │   ├── YES → Use a Diverging palette (RdBu, vik)
         │   └── NO → Use a Sequential palette (Viridis, Magma)
         │
         └── Must it survive grayscale printing?
             └── YES → Prefer high-contrast sequential options (Viridis, Cividis)

Putting These Principles into Practice

A manuscript figure, a poster panel, and a slide for a lecture all answer to the same color rules, so what you learn here carries from one format to the next.

If you would rather not hand-pick every swatch, Figviz starts you off with scientifically grounded color schemes by default, which keeps the output legible and accessible without extra fiddling. There is a wider gallery to browse in our diagram tools collection.

Want the full picture on building figures that are ready for submission? Our companion piece, How to Make Scientific Diagrams for Research Papers, goes deeper.

Summary

Once a handful of ideas click into place, picking colors for a figure stops feeling like guesswork:

1. Let the data dictate the family: Sequential, Diverging, or Qualitative
2. Bake in accessibility from the first draft: reach for CVD-safe sets like Okabe-Ito or Viridis
3. Check before you submit: run a CVD simulation and look at the grayscale version
4. Honor the journal's rules: no red-green pairs, no rainbow maps
5. Keep the count low: two or three lead colors carry most figures comfortably

Where to start by data type:

• Categorical data: Okabe-Ito, equivalently the Wong palette
• Continuous data: anything in the Viridis family
• Diverging data: ColorBrewer RdBu or Crameri's roma

The whole point is to make your science readable for everyone who looks at it, and deliberate color choices are what get you there.

A ROC curve drawn in a clean, professional palette, proof that even a dense technical plot reads better when the color is chosen on purpose.

Related Resources

• How to Make Scientific Diagrams for Research Papers
• Best Free BioRender Alternatives for Teachers
• Best Free Adobe Illustrator Alternatives for Teachers
• Science Diagram Tools

FAQ

Q: What is the best color palette for scientific figures? A: It depends on the data. For unordered categories, Okabe-Ito is the go-to benchmark, being both colorblind-friendly and backed by Nature journals. For data that ramps continuously, the Viridis family (Viridis, Magma, Plasma, Inferno) leads the field thanks to its even perception and wide accessibility.

Q: Why should I avoid red and green together? A: Red-green color blindness (protanopia or deuteranopia) affects roughly 8% of men, and for them those two hues merge into one. Switch to blue-orange or blue-yellow and the figure reads correctly for the whole audience.

Q: What color palette does Nature recommend? A: Nature points authors to the Wong palette from its 2011 Nature Methods paper, which is the very same eight colors as Okabe-Ito. The same guidance steers you away from rainbow scales and toward green-magenta in place of red-green for fluorescence imaging.

Q: How many colors should I use in a scientific figure? A: Six or fewer categorical colors is the rule of thumb. Once you cross about eight, readers lose the thread of which color maps to which group. If you genuinely need more categories, let shapes, patterns, or line styles share the load alongside color.

Q: Should I use RGB or CMYK for my figures? A: Go with RGB. Nature and most journals ask for it because RGB spans a wider gamut for on-screen reading, and they convert to CMYK for the press themselves. If you want to see the printed result ahead of time, do a CMYK conversion on your own copy before you submit.

Q: How do I test whether my colors are colorblind-friendly? A: Push the figure through a simulator such as Color Oracle (desktop), Coblis (web), or Viz Palette. Each one renders the figure as different CVD types would see it. If the distinctions you care about still hold up, the palette is safe to use.

Q: What are the Okabe-Ito palette hex codes? A: There are eight: Orange #E69F00, Sky Blue #56B4E9, Bluish Green #009E73, Yellow #F0E442, Blue #0072B2, Vermillion #D55E00, Reddish Purple #CC79A7, and Black #000000. Masataka Okabe and Kei Ito put the set together with CVD accessibility as the leading design goal.

Q: What is wrong with rainbow or jet color maps? A: They are perceptually uneven. The yellow and cyan portions read brighter than their neighbors, which fabricates visual "bands" with no basis in the actual data and can steer readers toward conclusions that are not there. A perceptually uniform map like Viridis removes the problem outright.