Start with the data type, then inspect the chart in context.
See the colors in a chart
Preview the palette on numbered sample data. Compare chart types and viewing conditions with the controls.
The last valid result is still shown. Correct the inputs before copying or downloading.
Sample data; numbers identify colors. Swatches show the selected vision model, while copied values and exports retain the original colors.
An input is outside sRGB. This tool maps it to sRGB before generation; previews, checks and exports use the same 8-bit values.
Color alone cannot communicate every distinction. Keep direct labels, markers or patterns and inspect the actual chart.
Inspect the closest pair
Minimum Euclidean OKLab D65 distance among all pairs, native scale. Machado severity 1.0 simulations are clipped and rounded to sRGB; no universal visibility threshold is applied.
Normal vision
Protanopia model
Deuteranopia model
Tritanopia model
Colors against the chart background
3:1 is a reference for meaningful graphical boundaries. This table checks only each original color against the chosen background, not all adjacencies or text.
CSS exports variables; JSON includes the original palette and diagnostics. ECharts and Chart.js exports are palette integration snippets, not complete chart configurations or continuous-scale implementations.
Color calculations run in your browser. The website still loads assets, advertising and analytics.
Give your data the colors it needs.
Start with the data type, then inspect the chart in context.
Data palette type
A category, an ordered value and a change around a midpoint tell different stories. Choose the palette type first, then check the order and appearance in a bar chart, line chart or heatmap with numbered sample data.
Colors against the chart background
Colors need to work with the chart background, labels and markers. Inspect simulated views and the closest pair without treating a single difference value as a universal safety certificate.
How to choose chart colors
Start with the data type, then inspect the chart in context.
01
Data palette type
Select categorical for groups, sequential for ordered values or diverging for a meaningful midpoint. Set the color count and source colors.
02
See the colors in a chart
Try the real chart background and sample layouts. Inspect original-color contrast, simulated views and the closest pairs, then review labels and markers.
03
Use the result
Copy individual original HEX values or download code and diagnostics. Add your own data and any continuous scale configuration in the chart library.
Continue your color workflow
Explore a palette or check a real foreground and background.
Color palette generator
Build five-color harmonies, lock favorites and export the palette.
Generate light and dark UI color themes from a brand color. Edit semantic roles, inspect contrast pairs and export CSS, Tailwind v4 or DTCG color tokens.
Choose the data type, compare the chart and keep the export context.
When should I choose categorical, sequential or diverging?
Use categorical for unordered groups, sequential for low-to-high values and diverging for values around a meaningful midpoint. The data type determines the color task; these are not interchangeable decoration styles.
How many colors can I generate?
Categorical and sequential support 2–8 colors. Diverging uses 3, 5 or 7 so there is one center swatch. More categories become harder to distinguish; direct labels, grouping and markers may help more than additional hues.
Will the starting color appear exactly in the result?
Not necessarily. It seeds hue and chroma; generated lightness and sRGB mapping can change the value. Diverging uses both endpoints. These are original OKLCH-based rules, not ColorBrewer, viridis or other named preset implementations.
Why do some colors fall below 3:1 against the background?
Light sequential steps or a diverging midpoint can have low contrast with a light background. The table is a reference for each original color against the backdrop, not a complete WCAG assessment of adjacent areas, labels or an entire chart.
Does the vision simulation guarantee color-blind safety?
No. It uses Machado 2009 severity 1.0 endpoint models in linear sRGB, followed by clipping and 8-bit rounding. Real perception varies. Closest-pair OKLab distances have no universal pass threshold; use labels, markers and patterns too.
Why check sequential lightness?
Ordered data should have an interpretable progression. The check reports whether exported colors have monotonic OKLab lightness after mapping and rounding. It is a numeric diagnostic, not proof of equal visual steps in every context.
Are copied colors affected by the simulated preview?
No. The chart and swatches show the selected model, while HEX copies and every export contain the original palette. Numbered sample data are illustrations, not analysis of your own dataset.
What do the exports include, and are colors uploaded?
CSS provides variables, JSON saves settings and diagnostics, and ECharts/Chart.js snippets show palette integration. You supply data, series and any continuous value mapping. Calculations are local; valid settings persist in the full URL for reloads and language changes. The site still loads assets, advertising and analytics.