Analyzing the tint of LEDs by shining a light on your skin

First thing I do when I get a new flashlight is to shine the light on my skin. From my experience, shining the light on my skin gives me a pretty good idea of the quality of the light.

Now that I have somewhat usable measuring instrument for lights, I have become a little more confident of my testing method. I think it would be useful for community members, so I decided to share it with the community. I used Opple Light Master Pro (G3) to measure CRI and calculated DUV from The photos are taken with my phone Samsung Galaxy Note 20 Ultra. I used Pro mode in the camera app.

Testing method

  1. Shine your flashlight on your skin. I use low output levels.
  2. You can easily compare the skin tones of multiple flashlights by alternating between two or more flashlights.
  3. Low CRI or low R9 LEDs tend to show more of green and yellow tone. High CRI LEDs with high R9 show more red tone.

Why skin?
Humans have excellent sense of distinguishing skin tones. Also, red color of your blood shows how well your flashlight shows red.

First, here is the reference photo. The photo was taken indoors with overcast daylight. Camera WB showed 5000K and since the window was mostly covered with blinds there were relatively small amount of light. I think it represent my skin tone quite well. My ethnicity is east Asian.

Figure 1. Reference photo. Overcast daylight.

I took several photos of my hand using 9 flashlights at the same spot using the same camera setting. I used low output through out this article because high output tended to confuse the camera and sometimes ended up with random and weird representation of my skin. Which one do you like?

Figure 2.Which one do you like? (manual WB 5000K)

Here are the same photos with the names of the flashlights and LEDs.

Figure 3. Main result (manual WB 5000K)

The following are what I observed:
Known low R9 LED shows more green and yellow.
Low CRI LED shows a lot of green and yellow.

You may think fixing the white balance to 5000K is not fair despite the fact that most of lights used in this testing is in the range from 4500K to 5800K. To address this concern, I manually set WB to nominal CCT of the LEDs. For example, I set WB to 4500K when taking photo using E21A 4500K LED. This exaggerated reds in my sample.

Figure 4. Manual WB, set to nominal CCT of LEDs

I use auto white balance setting and took the photos again. This made all of the photos to show more pale skin and the difference between samples is a lot smaller. However, you can easily pick the best LEDs in terms of color representation.

Figure 5. Auto WB

Here are the Oppple Light Master Pro (G3) measurements.

Figure 6. Opple Light Master G3 measurements

Here are the list of lights used.

  1. Wurkkos TS10 Latticepower 5800K
  2. Sofirn SP10 Pro Samsung LH351D 5000K
  3. Convoy T3 Nichia 519A 5000K
  4. Lumintop FW3A CREE XPL Hi 5000K
  5. Emisar D4V2 Nichia 219C 5000K LCRI
  6. Noctigon KR4 Nichia E21A 5000K R9080
  7. Fireflies NOV-MU Nichia E21A 4500K R9080
  8. Nitecore MT06MD Nichia 219B 5000K HCRI
  9. Fireflies E07X Nichia 219B 4500K R9080

Figure 7. List of flashlights

2 Thanks

Thank you! I was tempting to order fw3a or edc18 with CREE XPL Hi 5000K or even 65000K for its lumens, but you got me out of it. :slight_smile:

Ahah, nice experience :wink:

Curiously, the ones I preferred were Wurkkos, NOctogon KR4 and Nitecore MT06MD.
Cool! :sunglasses:

The low R9 LEDs can be used in horror movies.

Nice work!

Great job!
the two my eyes went to first: sw45k, second E21a, your Opple DUV values shows my bias. Im trained to choose the lowest DUV… lol (Im a diehard sw45k fan)

for a more universal test, maybe use the palm of the hand, less likely to differ in melatonin
(I dont know what East Asian is, as Asia is to the West of Me, LOL)

definite yes to using a consistent daylight white balance, NOT the white balance to match the CCT of the LED

given what you have seen and learned, what are your two top LED choices?

Excellent work! Thanks for the lesson. :slight_smile:

I agree the palm can be more universal. However, the backside of my hand showed much more variation in colors across different LEDs. Therefore, it was much easier for me to detect the difference on the back side of the hand than my palm. I can still see the difference on the palm, but not as noticeable.

I prefer neutral tint that resembles overcast daylight. So, my top choice is E21A 5000K. I would rate it A+.

Latticepower 5800K, Nichia 519A 5000K, and Nichia 219B 5000K high CRI are my second best. These are my A0 lights.

E21A 4500K is also very nice to look at, but it is too warm for my taste. Also, it adds significant amount of red hue which is not a plus for me. I would rate it B+.

I think my sample of 219B 4500K has too much red. It has even more red than my E21A 4500K. For this reason, 219B 4500K is not my preferred choice. However, it is quite nice to look at. I would rate it B0.

LH351D 5000K is high CRI LED, but it seems to have very low R9. Everything looks more green and yellow when I shine the light. I do not like this.

Of course, the above rating reflects my own preference and should not be taken objectively.

excellent comments, I agree the LH351D is too low in R9.

thank you for taking the time to share your tests, it also makes sense that your back of fist skin, offers more colors than the palm

enjoy your choices :wink:

Latticepower 5800K looks like a surprisingly good and neutral tint. :+1: The model name of Latticepower used in TS10 is CSP2323 (they don’t list 90 cri in their datasheet though :open_mouth: ) .

That Latticepower seems to be a thing of beauty for photography purposes.

Thanks for the info. I read in the Wurkkos TS10 thread that CSP2323 is often used as a general term, but I see that Latticepower is actually using it as their part number.

Yes, the tint is very neutral. The calculated DUV is actually the closest to zero among the flashlights I own.

However, neutral can also mean bland. It doesn’t make some colors pop and can be less exciting. A little kick in the reds can be desirable in some cases, because it can give strong impression.

Nevertheless, I agree Latticepower CSP2323 5800K in TS10 is a very good LED. I wish to see a detailed review on the LED that are similar to that of maukka. I was very impressed to see his/her review on KR4 with E21A 5000K. Review: Noctigon KR4 (18650, 4xE21A 5000K, CRI96) It was the best review and had the best quality I could possibly ask for.

The best 351D Led I got was at 2700k from Sofirn (when they were 1,5$ pcs)
with higher R9 than others CCT.
Domed is a just bit yellow and slight above BBL, but It turn nice after sliced

Very cool post, thank you for the time and effort. From your pictures, I like the 219b’s and E21a’s the most.

Thank you for the info. I do not have proper tools and skill sets to desolder the MCPCB from the lights at the moment, but when I do I will try slicing domes. However, without modding I am quite disappointed at LH351D 5000K as yellow and green are very pronounced. It seems at lower CCT the problem is less discussed in the forum, so the pronounced green and yellow may not be an issue in lower CCT.

I have hard time distinguishing light yellow from white (such as stains on white shirts) under lights of CCT below 4500K. This happens even under the single source of light. So, I usually purchase lights with higher CCT. In addition, 5000K matches my ambient lighting source. So, 5000K lights are the most natural lights for my use and environment.

Thanks for the kind words.

There are two versions of 219B and E21A in this sample, 4500K and 5000K. Which one did you prefer?

Usually yes, lower CCT LH351D as 3500k and 2700k have better tint. I got best result using flood pebbled Tir lens, you can get them at Convoy store. With different Tir lens or reflector the result can be worse.
My 5000k sample on SP10S light wasn’t so greenish but harsh with inconsintent tint (clearly lack in Reds value)

I like the 4500K the best, but the 5000K is nice too.

Interesting, especially the difference the color temp of the photos changes things.
Beam shots are so difficult to judge color accuracy.

On another note. That’s a heck of a set of knuckles you got there!
Tae Kwon Do?

All the Best,

Great tint comparison, one of the best here I think.

I’m surprised by the negative DUV of the 5000K E21A since I assumed it must be greenish due to the brighter D240 bin instead of the D220 one.

At which power level did you take this measurement?

Clemence once sold me the 4000K E21A in the D240 bin. It was as terrible green as the bad binned SST-20’s. The D220 on the other hand was pure neutral, very pleasant, much nicer than the overly rosy XPL-Hi 5D or the pink 219B SW40.

Thank you for the kind words.

I don’t recall exact numbers, but I tried to adjust the output so my lux meter (put on my hand) measures similar output. I recall the output in my reference light was about 100 lumens. Of course, I had to use higher lumen output for NOV-MU because the light was pure flood without any optics.

E21A 5000K is no longer available at, but it is by far my favorite 5000K led. I believe the E21A 5000K was sm503D240 bin. And of course there must be some sort of tint lottery and I may have won a very good bin in my lights. Even the spec sheets report that their leds are in a certain range in the color space (therefore CCT and DUV can vary even in the same bin). I liked the tint so much in my KR4 that I have four of them all E21A 5000K.

By the way, even the same led can give quite a different DUV depending on optics used. I recently swapped leds in Nitecore TINI2 to 519A taken from my TS25 which I believe had slightly positive DUV. However, when I measured DUV after I installed the same led in TINI2, the DUV was in the negative range of (–0.0030).

we enjoy your post
and reference it often
when evaluation time
comes for a new light
of any type.