Skip to main content
ponderloop

Color Blindness Simulator (Palettes and Images)

#2ca02c and #ff7f0e differ by ΔE 55.2 with typical vision but only 1.4 under protanopia. Of the 10 pairs in the sample palette, 7 get hard to tell apart (ΔE 2000 below 10) under at least one of protanopia, deuteranopia, tritanopia and achromatopsia. The simulation applies the Machado, Oliveira and Fernandes (2009) matrices in linear RGB, so it is an approximation of how these colors look, not a diagnosis or a medical test. Paste your own colors, or open an image on your device, to see yours.

Updated Runs in your browser: nothing you enter is uploadedHow this tool works

This is an approximation. It models an average observer on an sRGB screen using published matrices; real color vision varies from person to person, and this is not a test or a diagnosis.

Your colors

    100% (full: protanopia, deuteranopia, tritanopia)

    How each color looks

    5 colors, 10 pairs: 7 pairs get hard to tell apart under at least one type (ΔE below 10).

    • #d62728

      • Typical vision#d62728
      • Protanopia#615725
      • Deuteranopia#8b7c1f
      • Tritanopia#ec002b
      • Achromatopsia#6f6f6f
    • #2ca02c

      • Typical vision#2ca02c
      • Protanopia#a39119
      • Deuteranopia#968838
      • Tritanopia#009b89
      • Achromatopsia#8b8b8b
    • #1f77b4

      • Typical vision#1f77b4
      • Protanopia#5a79b7
      • Deuteranopia#456cb3
      • Tritanopia#00868d
      • Achromatopsia#727272
    • #ff7f0e

      • Typical vision#ff7f0e
      • Protanopia#a59100
      • Deuteranopia#c4ae05
      • Tritanopia#ff616d
      • Achromatopsia#a3a3a3
    • #9467bd

      • Typical vision#9467bd
      • Protanopia#5279c0
      • Deuteranopia#5d7bbb
      • Tritanopia#8e7589
      • Achromatopsia#7b7b7b

    Pairs that get hard to tell apart

    ΔE is the CIEDE2000 difference between two colors. Below 10 they are hard to tell apart; below 5 they look nearly the same.

    • #d62728 and #1f77b4 differ by ΔE 48.6 with typical vision, but only 1.2 under achromatopsia (nearly the same).
    • #2ca02c and #ff7f0e differ by ΔE 55.2 with typical vision, but only 1.4 under protanopia (nearly the same), 7.8 under achromatopsia (hard to tell apart).
    • #1f77b4 and #9467bd differ by ΔE 26.4 with typical vision, but only 1.8 under protanopia (nearly the same), 6.1 under deuteranopia (hard to tell apart), 3.6 under achromatopsia (nearly the same).
    • #d62728 and #9467bd differ by ΔE 36.3 with typical vision, but only 4.8 under achromatopsia (nearly the same).
    • #d62728 and #2ca02c differ by ΔE 71.8 with typical vision, but only 4.8 under deuteranopia (nearly the same).
    • #2ca02c and #9467bd differ by ΔE 51.8 with typical vision, but only 5.9 under achromatopsia (hard to tell apart).
    • #2ca02c and #1f77b4 differ by ΔE 52.6 with typical vision, but only 9.6 under achromatopsia (hard to tell apart).

    An image

    Open a chart, map or screenshot to see it as each type does. The image stays on your device.

    Sample palette as it appears with each type

    Simulated at full severity from 5 colors. Each cell is the 8-bit sRGB color the model produces.
    OriginalProtanopiaDeuteranopiaTritanopiaAchromatopsia
    #d62728#615725#8b7c1f#ec002b#6f6f6f
    #2ca02c#a39119#968838#009b89#8b8b8b
    #1f77b4#5a79b7#456cb3#00868d#727272
    #ff7f0e#a59100#c4ae05#ff616d#a3a3a3
    #9467bd#5279c0#5d7bbb#8e7589#7b7b7b

    The Machado et al. (2009) matrices at severity 1.0

    Each row of a matrix multiplies the linear red, green and blue of a color to give the new red, green and blue. Intermediate severities interpolate between tabulated steps of 0.1.
    TypeNew redNew greenNew blue
    Protanopia0.152286, 1.052583, −0.2048680.114503, 0.786281, 0.099216−0.003882, −0.048116, 1.051998
    Deuteranopia0.367322, 0.860646, −0.2279680.280085, 0.672501, 0.047413−0.011820, 0.042940, 0.968881
    Tritanopia1.255528, −0.076749, −0.178779−0.078411, 0.930809, 0.1476020.004733, 0.691367, 0.303900

    Types of color vision deficiency

    TypeMilder formWhat is affectedHow this page shows it
    ProtanopiaProtanomalymissing or weak long-wavelength (L, "red") conesMachado matrix, severity slider for protanomaly
    DeuteranopiaDeuteranomalymissing or weak medium-wavelength (M, "green") conesMachado matrix, severity slider for deuteranomaly
    TritanopiaTritanomalymissing or weak short-wavelength (S, "blue") conesMachado matrix, severity slider for tritanomaly
    AchromatopsiaIncomplete achromatopsiano working cones, so only brightness is seenRelative luminance as a gray

    Pure red (#ff0000) at different severities

    Severity 0% is typical vision; 100% is the dichromat. Achromatopsia does not depend on severity.
    SeverityProtanDeutanTritanAchromatopsia
    0%#ff0000#ff0000#ff0000#7f7f7f
    30%#d04a00#d66300#f42e1f#7f7f7f
    60%#a75900#bb7d00#ff0004#7f7f7f
    100%#6d5f00#a39000#ff000f#7f7f7f

    How to see a palette or image as a color-blind person might

    1. Enter your colors. Type or paste colors into the box, one per line or separated by spaces or commas: hex (#d62728), rgb(), hsl() or CSS names. Up to 12 distinct colors are used.
    2. Choose a severity. Leave the slider at 100% for the full dichromat types (protanopia, deuteranopia, tritanopia). Move it down to see the milder anomalous trichromacies: protanomaly, deuteranomaly and tritanomaly.
    3. Compare the cards and read the warnings. Each color shows its typical look and how it appears with each type. The warnings list pairs that are clearly different with typical vision but have a CIEDE2000 difference under 10 once simulated. Under 5 they look nearly the same.
    4. Try an image. Choose an image file or drop one on the page. It is read and drawn in your browser only, and shown next to its simulated versions. Use it to check charts, maps, and UI screenshots.

    Frequently asked questions

    How accurate is a color blindness simulator?
    It is an approximation. Real color vision varies from person to person, with the exact cone sensitivities, the viewing conditions and the screen all playing a part, and a simulation on a typical sRGB display cannot reproduce an individual's experience. The Machado et al. model is a well-cited physiologically based model, and the results are good enough to see which colors become confusable, which is what matters for design. Treat it as a design check, not a diagnosis; only an eye-care professional can test color vision.
    Which matrices does this use?
    The pre-computed matrices from Machado, Oliveira and Fernandes, "A Physiologically-based Model for Simulation of Color Vision Deficiency", IEEE Transactions on Visualization and Computer Graphics 15(6), 2009, pages 1291 to 1298. At full severity, the first row of each matrix (the one that makes the new red) is [0.152286, 1.052583, −0.204868] for protanopia, [0.367322, 0.860646, −0.227968] for deuteranopia and [1.255528, −0.076749, −0.178779] for tritanopia; two more rows make the new green and blue. They are applied to linear RGB, not to gamma-encoded sRGB values.
    Why linear RGB instead of the sRGB numbers?
    The matrices were derived for linear light, where adding and mixing colors is physically meaningful. sRGB values are gamma-encoded, so multiplying them directly would give wrong brightness. The page decodes each channel to linear light, multiplies by the matrix, clips the result to the 0 to 1 range and encodes it back. For pure red (#ff0000) that gives #6d5f00 for protanopia, #a39000 for deuteranopia and #ff000f for tritanopia.
    What do the severity slider and "anomalous" mean?
    Dichromats lack one cone type entirely (protanopia, deuteranopia, tritanopia). Anomalous trichromats have all three cone types, but one is shifted, which gives a weaker deficiency: protanomaly, deuteranomaly or tritanomaly. The paper offers matrices for severities from 0 to 1 in steps of 0.1, and this page interpolates between them for the slider. At 0% you see typical vision, at 100% the dichromat. For pure red, deuteranomaly at 60% looks like #bb7d00 and deuteranopia like #a39000.
    How is the distinguishability warning calculated?
    Both colors are simulated, converted to CIELAB, and compared with the CIEDE2000 formula. A ΔE of about 1 is the smallest difference most people can see side by side. This page warns when a pair that differs by 10 or more with typical vision falls below 10 under a type, and says "nearly the same" under 5. The thresholds are rules of thumb: small shapes, thin lines and text need more difference than large swatches. For the sample palette, #2ca02c and #ff7f0e differ by ΔE 55.2 with typical vision but only 1.4 under protanopia.
    How is achromatopsia shown?
    Achromatopsia (rod monochromacy) is not part of the Machado et al. model. The page shows each color as a gray with the same relative luminance (0.2126 R + 0.7152 G + 0.0722 B in linear light). It is also a handy grayscale check for print and for people with low vision. The severity slider does not change it.
    How common is color blindness?
    The U.S. National Eye Institute says about 1 in 12 men have color vision deficiency, and NIH’s MedlinePlus Genetics puts red-green color blindness at about 1 in 12 males and 1 in 200 females among people of Northern European ancestry (roughly 8% and 0.5%); it is rarer in most other populations. Most of those are deuteranomaly, the most common type, and protanomaly. Blue-yellow deficiencies (tritan) and complete color blindness are much rarer.
    How do I make a palette safer for color-blind users?
    Do not rely on hue alone. Vary lightness between the colors, add labels, patterns, shapes or direct text on charts, and keep red and green out of the only cue (WCAG success criterion 1.4.1, Use of Color). Colors that differ in lightness stay separable under all four types, which the achromatopsia column shows. Re-run the simulator after every change.
    Is the image I choose uploaded anywhere?
    No. The file is decoded and drawn on canvases in your browser, and the simulation runs on those pixels on your device. Nothing is uploaded or stored. For speed, large images are shown at a reduced size, at most 480 pixels on the long side; the original file is not changed.
    Why do the tritanopia results look less certain?
    The physiological model behind the matrices handles protan and deutan deficiencies well. For tritan deficiencies the authors use a simpler assumption, a shift of the cone response, so the tritanopia and tritanomaly matrices are a rougher approximation than the other two. They are also very rare in people, so check them last.

    How it works

    Colors you enter are read into 8-bit sRGB with the same parser as the color converter on this site (hex, rgb(), hsl(), hwb(), CSS names); transparent colors are blended onto white and repeats are ignored. Each channel is turned into linear light with the sRGB transfer function, multiplied by the 3 x 3 matrix from Machado, Oliveira and Fernandes (2009) for the type and severity, clipped to 0 to 1 and encoded back to 8-bit sRGB, rounding to the nearest whole number. The paper tabulates matrices for severities 0.0 to 1.0 in steps of 0.1; in between, the matrix is interpolated linearly, as the authors describe. Achromatopsia is the Rec. 709 relative luminance of the linear color shown as a gray. To compare two colors, each is converted from sRGB to CIELAB with a D65 white point and the CIEDE2000 difference (Sharma, Wu and Dalal, 2005, with all weighting factors 1) is taken; a pair is flagged when it is at least 10 with typical vision and below 10 for a type, and called nearly the same below 5. Images are drawn at no more than 480 pixels on the long side and every pixel goes through the same function; the alpha channel is kept. All of this is a model of an average observer on an sRGB screen, so it approximates, and does not replace, testing with people. Nothing you enter or open leaves your browser.

    Sources

    More: Color and CSSAll developer toolsAll toolsHome & DIY calculators