I have some no-name LEDs and I would like to estimate how good their CRI is. What I need is only a rough estimate, since I would use the best ones for studio lighting or product lighting, while the worst ones would go to garage/workshop lighting.

How could I rank their CRI?

I thought about finding a set of objects with colors that are usually difficult to correctly light with low-CRI LEDs, like cyans and bright reds, but which ones would be good match?


I bought a ColorChecker from some Chinese seller and I kept it well lit with different light sources to try to judge the light. The shots were taken with a dSLR and later in Lightroom white balance was adjusted using the whitest patch as reference. I attach the three results from a halogen light, a IKEA led bulb, and from a Philips LED spotlight (however, for this last one there may be reflections from a wall with a painting and the light may not be the pure LED light).

As additional information: when I switched from a halogen spotlight to the Philips LED spotlight, the colors of the painting on the wall were CLEARLY different. Not so much in this test, though.

I couldn't equalize the brightness, I don't know how to do it in Lightroom and I don't have Photoshop. The frame is differently lit because of the different angles between light and patches, see shadows. The patches are however quite matte.


IKEA led

Philips LED spotlight

  • 1
    \$\begingroup\$ You might shine the light through a prism to see the spectrum directly, I don't know how effective that would be. \$\endgroup\$ Apr 21, 2017 at 3:52
  • 1
    \$\begingroup\$ You could try some color-blindness test pictures to see if they help. \$\endgroup\$ Apr 21, 2017 at 3:53
  • \$\begingroup\$ How much effort can you put into it? \$\endgroup\$ Nov 2, 2017 at 16:37
  • \$\begingroup\$ @EuriPinhollow well my original question was about finding objects that can easily show the weak colors. I just found that now some cheap ColorChecker patterns are sold for little money, the question is a bit outdated. For 15 Euro I can buy one of those and see the 18 test colors used for the (extended) CRI. No need to look elsewhere. \$\endgroup\$
    – FarO
    Nov 2, 2017 at 17:03
  • 1
    \$\begingroup\$ You're roughly "ranking" them from best to worst according to your criteria. If you want a means of "critically comparing" different ones for ranking, however, carefully re-read the answer after you have also reviewed the definition of CRI. Note that without a high CRI, you will not achieve a neutral. That's the reason neutrals are used for very critical comparison of colour accuracy. \$\endgroup\$
    – Stan
    Nov 3, 2017 at 3:41

2 Answers 2


I recommend you take a look at this answer: Do I always get the same colours when I set the white balance correctly?

A color checker is not meant to do white balance only. It is to make a custom light-camera profile.

As you can read in that post, the produced adjustments are not simple curves but can be different shaped ones.

I do not know the quality of the color patches of your target, but you need to complement a color profile using a software that:

  • recognizes that pattern
  • make adjustments
  • and prepare a color profile to be applied to the sets of photos.
  • \$\begingroup\$ I don't want to correct the images, I want to use the pattern to estimate the quality of the light source... and you can see that several patches in the second image are duller than in the first one. \$\endgroup\$
    – FarO
    Dec 25, 2017 at 20:15

Find or make a neutral grey scale or neutral patch to use as the subject.

A Color-Checker™ has such a scale along one side of the card. The dark tones along with the white patch will give you some idea of the linearity of the source. Sometimes there can be a noticeable colour shift from shadows to peak white. Alternatively, you can choose to use a single pure white patch and varying the exposure time and intensity to compensate in a known way to get a scale of tones if you prefer.

One of the purest neutral white pigments to photograph is magnesium oxide whose spectral response is quite linear over most of the visible spectrum. You can buy tubes of it as a paste from a pharmacy near you and use that as your standard white.

Photograph the "subject" after carefully masking ambient and extraneous illumination that is not from the source you want to test.

Do not use or activate automatic white balance in your camera settings or software.

Make as many photographs as you wish. Keep detailed notes.

You'll evaluate your results comparing them with the neutral standard white patch you made to photograph. Compare only similarly exposed patch lightness to better compare colour shift apart from other factors. You'll want some density so avoid blown-out exposures. Peak white (not highlights) typically falls about 2-1/2 stops over mid grey.

Strive for a neutral environment with no dominant colour to affect your evaluation. To best evaluate your results, set yourself up in an area where the correlated colour temperature of the lighting is 5000K.

ONLY examine your results under continuous source incandescent lighting with the highest colour rendering index you can get (not fluorescent). Quartz halogen will give you a good source if the operating voltage is carefully controlled. Solux makes a 12 volt calibrated 5000K source

The amount of illumination (light level) should be 500 lux at the display plane where you will make your comparisons. At this illuminance and above correlated colour temperature, the eye is most efficient and accurate. Be well rested as your ability to judge colour accurately diminishes with eye fatigue.

The colour of the light can change our perception of colour up to 14 ∆E units.

Just ONE ∆E unit is a perceptible difference in colour!

Always judge colour in the same environment!

Sort your patches according to your colour acceptance. Your notes will help you keep things organized and relevant.

In theatrical lighting, many multi-hued (gelatine filtered) sources are mixed to produce desired effects. In the same manner, you can mix multi-hued LEDs to achieve your ends.

Edit: If you wish to make a statement about the colour rendering index of any specific patch you produce, you'll need a calibrated spectrophotometer as an aid. You can construct one and calibrate it; but, that's another project for another weekend.

  • 4
    \$\begingroup\$ This is not sufficient. The issue isn't color temperature, or a general off-neutral cast. The problem is that white LEDs are actually fluorescent rather than broad-spectrum lights. You need to analyze colors, not just a neutral gray scale. \$\endgroup\$
    – mattdm
    Mar 20, 2017 at 4:17
  • \$\begingroup\$ That is why i asked for objects to specifically test colours that are weak in LEDs: testing those will already give me a strong idea about the color quality. \$\endgroup\$
    – FarO
    Mar 20, 2017 at 7:01
  • \$\begingroup\$ Aside from the "using only neutrals" aspect, I am not sure why this got downvoted. The object is the ColorChecker. this will work, especially for the loose requirements of the question. Even with a "loosely calibrated" setup, the best CRI LEDS can easily be estimated by direct swatch comparison using eyeballs. CIE 1999 recommends ColorChecker as a render target. \$\endgroup\$
    – Yorik
    Nov 2, 2017 at 15:48
  • \$\begingroup\$ @Yorik The answer is indeed about Color Checker, but specifically about its grey patterns. While the ColorChecker would be a good answer, focusing about the neutral patches isn't. I didn't downvote but I understand the people downvoting. \$\endgroup\$
    – FarO
    Nov 2, 2017 at 17:05
  • \$\begingroup\$ @Stan if I disable white balance, the grey patches will never be neutral, even with incandescent light. I don't get this part. Also, if I enable WB, some patches will still look duller under poor CRI lighting. Is WB-off really needed? \$\endgroup\$
    – FarO
    Nov 3, 2017 at 9:52

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.