High CRI... not what I expected? (Happy ending in post 118)

Our ladies knew that very well in the times of incandescent bulbs and fluorescent tubes: If you want to know what that dress really looks like, see it outside the clothes shop.

I do think the brain is very good at interpreting incomplete visual data, and with a single light source, it will adjust its white balance. I dont think it adjusts the CRI of the LED though…

as to why an old Hippy into natural foods, and organic farming would be interested in High CRI… I think it is me trying to use the most Organic Light, given the option…

In the above photo, the wood of the table is not actually green. It is the lack of R9 in the Low CRI LED, that omits the red spectrum content of the wood.

a more subtle example:
I know the flower on the left is a Rose, but given the choice, I get more excited looking at the one on the right (it is a Higher Fidelity, fuller spectrum image):

so true!
which is why I think it is important to note ambient adaptation, when making an observation about the color of an LED output.

The reason I recommend High CRI, is because I like the superior color rendering, tint, and CCT.

Lets use a real world example. Someone asks, which AAA Tool should I get, the CW that is brighter, or the High CRI, that has less lumens.

My first answer is always, get the High CRI.
I dont necessarily go into the details that the High CRI being offered is a Neutral White Nichia instead of a Cool White Cree.

and I dont necessarily go into the details about the CW having green tint, while the NW has pink tint… all of those CCT and DUV factors are included automatically in the recommendation to choose the Nichia.

I do post photos showing the difference in tint, and Im honest about the difference in lumens…

some people prioritize lumens over CRI, Tint, and Color temperature…
it is their choice

if they ask which Emissar D4 LED to choose, then the answer is more complex, because the LED option is more than just Nichia vs Cree. I still say get the High CRI, but now the buyer must also decide the CCT and Tint they prefer.

the answer is more or less detailed depending on the actual real world choices available… for a given model of light.

Just chiming in to say I wish there was more bin (tint) transparency on emitters. Seems like CCT and high cri choices have opened up a lot, but the specific bin you’re getting remains a crapshoot to some extent.

The low CRI leds showing the wood as green looks more to be a green tint issue than lack of R9 issue.

I think both factors play a role

I understand we can use a Lee minus green (pink plastic) filter, to reduce the green in the Low CRI light, but that also reduces output… and it wont make a light with low R9 into a light with 90+R9, so I would prefer a High R9 CRI LED either way.

maukka has an excellent post about using Lee filters… here is one example:

I dont want to add a plastic filter to my lights, if I can buy a High CRI LED to begin with.

whatever the reasons
the observation remains:
Low CRI makes the wood color Less Realistic
High CRI makes the wood color More Realistic

another example,
left is NW High CRI, middle is NW Low CRI, right is CW Low CRI:

the High CRI LED produces the most realistic wood color

This is just kind of how the semiconductor industry works. You have to pay a premium to narrow down a bin for whatever device you are purchasing because it involves more detailed testing and logistics. Usually you’ll only see that done when purchasing in bulk. If you reach out to Luminus and want to purchase ten reels of SST20, they’ll probably let you specify bins for tint, voltage, etc

If you take the color brown and remove the red, you are left with green.

A really good experiment would be comparing WW, NW, and KW all in HCRI. Let the cam do the WB.

Edit: No, I have no idea what can be seen then.

That’s not how the visible spectrum works. That’s how subtractive color works (ie pigments/paint). In nature, different surfaces have a reflective property that reflects only specific wavelengths back to our eyes, otherwise known as their color. Brown as it refers to the light spectrum, is low saturated oranges/yellows/reds. Theoretically, a neutral tinted LED of little deep red rendering (note— little deep red. leds with low R9 still produce red) will still show brown, just not as vibrant as it usually is. An led with a green tint to it will show brown as yellower or greener than what it should be. An led of magenta tint to it, will show brown as a more saturated red color than what it should be. This is why some people really like the SW45K tint bin, as it makes browns really pop. An ideally neutral light source should not make browns really pop and look rich, but neither should it make browns look green.

I did see your request to post some more picture regrading tint differences vs CRI differences on wood surfaces Jon. I’ll try to get some later, but it won’t be an ideal comparison, because the lights are slightly differing CCTs. What I can compare is a bunch of 65 CRI 5700k Osram CSLPMs at low output, (really green at low output, certainly negative R9) vs my 5000k Optisolis (slightly green, but very neutral by comparison, warmer, and amazing color rendering). Then, I can put some minus green over the Osram emitters, which should neutralize its tint and also warm it up a bit, but minimally affect CRI.

Actually, no. I think that is a CRI issue.

I see the same thing even when I compare XPL HI with Lee Minus Green filters to reduce the DUV to below the black body line to high CRI SST-20 or SW45K.

A good high CRI led with high R9 really does make woody colors look much better. That seems to be its primary advantage.

thanks… it is not that necessary… I already tested a minus green filter on the NW XM-L2, and I can see that it helps reduce the green tint.

However, it does not make a low R9 LED into a High R9 LED,
which is why I posted the image from maukka:

thanks for your thoughts,

Organic things like wood, the palm of my hand, food, flowers, leaves… all have varying degrees of Red pigments that are not visible under Low CRI…

What Low CRI is really good for, is making leaves look greener :wink:
because Low CRI Fails at illuminating red pigments. Low CRI simply Omits Red Spectrum Information

whether or not Red spectrum is a High Priority to someone, or not, is a personal decision, based on other personal priorities, such as output.

this is the first comparison I did (a pair of Maratacs, about 6 years ago), that started me on the road to High CRI:

Ok here are two shots, comparing two different emitters. One is a Nichia Optisolis at 5000k, with 99 CRI and 90+ R9, with slightly green tint (barely). The other is an array of Osram CSLPM1.TG at very low output (Osram emitters get even greener at lower outputs). The Osrams are 65 CRI, unknown R9, and very green at low outputs. First shot is without tint correction applied to the Osram emitters, while the 2nd shot has tint correction applied to the Osram emitters.

Both pictures taken with a Sony A6300 and Sigma 35mm F1.4 using daylight white balance (5200k, +11 green), f/8, 2 second exposure, ISO 100.


As can be seen here, it’s very hard to discern the two emitters once tint correction is applied to the Osram emitters, at least when it comes to illuminating wood (low saturated oranges and reds). The amount of CRI or R9 here makes very little difference compared to the tint of lights.

I’m not trying to advocate for Lee filters, nor am I trying to say CRI isn’t justified. I’m merely trying to prove my original point in regards to the topic creator, in saying that CRI is a much more subtle quality of light compared to CCT and tint deviations. I love CRI lights myself. My favorite lights use 95cri E21A 4500k emitters and 95cri 4000k SST20 emitters with quarter minus green applied over them. However, I want to reinforce the idea that CRI is but a small portion of what makes a light pleasing to the eye, and can only be truly quantified with all other things being equal: that is equal CCT and tint deviations of two lights. Otherwise, those differences will overshadow CRI. Posting pictures comparing two lights of vastly different CCTs and tints doesn’t do much when making an argument for CRI.

1 Thank

great photos!
thanks for taking the time to show those examples
I appreciate your contributions to a very detailed an nouanced discussion

I see your point that once you add a lee filter to alter the tint of a low cri light, it seems to make the wood you used, look almost identical to the high cri light. I found the same effect when I put a minus green on a Low CRI LED… the wood looked very similar.

I agree Tint is a dominant variable in photo comparisons.

High CRI is very difficult to demonstrate in photos,
it is a more subtle component of light quality, than Color temperature and Tint.

btw, what is +11green setting doing for your photo? I dont have experience with manual cameras… I only learned the little I know about photos, from using my iPhone… You obviously have control of a lot more parameters.

generally speaking, High CRI LEDs tend to have less green tint, and warmer color temperature than Low CRI LEDs.

It is also true that we can alter the tint of a Low CRI LED, by adding a pink gel, so the two images look similar, even though they are not even close in CRI. I agree the wood in your photo is a poor target to demonstrate CRI.

Photos are a very complex medium, and it is much easier to show Tint and Color temperature, than CRI in a photo… White Balance is also an important variable.

I understand your wanting to equalize the variables, so we are only comparing one at a time… but that is now how LEDs arrive

LEDs arrive with a package of features that include CCT, Tint, and CRI… they are bundled together… So, when I suggest a High R9 priority, it also includes Tint and Color temperature considerations, that I dont necessarily mention. They all matter, to me.

Another variable I dont usually bring up, but I will now, is how I “feel” when I use a certain CCT, Tint, and CRI… I find CW Low CRI, and also NW Low CRI make me feel “washed out”, or “color deprived”… its difficult to find words for feelings… but I like the feeling I get from High CRI, and especially from high R9 CRI… as opposed to LEDs with low R9…

but even after all that, if the tint is green, I wont be happy with a high R9 CRI LED… the SST is a perfect example of that… it has great 90+ CRI R9, but with green tint. It makes sense to me that you would want to put a minus green filter on an SST…

I think we have similar taste in tint, as I would want a minus green on an SST also

I reach a similar goal,
by installing an sw45k, which also has 90+ CRI R9, and does not need a minus green filter,

I understand we can reach a similar tint, using an SST with a minus green filter… Personally I avoid Lee Filters, the sw45k works for me

I respect others preferences will differ…

question, given the lumen penalty from using a Lee Filter on an SST (which has higher output than 219b), does the SST with minus green filter added, still have a Lumen advantage over an sw45k? And if so, can you inform what Percentage brighter the SST w -green is, to an unfiltered 219b?

for reference
a 50% difference in lumens is only a small visual difference… it is the mode spacing of an HDS… each of the 24 steps on the dial is 50% brighter than the one before. 50% sounds like a lot, but it is not, when it comes to lumens.

That’s my bad… I meant to say +11 magenta. White balancing is just setting a true neutral white point, where warm to cool color temperature is completely balanced, and green to magenta tint is completely balanced. Since sunlight is slightly green tinted, my camera’s sunlight white balance factors in a little bit of magenta when set to sunlight white balance, because it wants to make whatever the sun lights up void of any tint. Since the Optisolis is designed to mimic the sun, it too also has that tiny bit of green tint to it, so this is the best possible white balance for the Optisolis. Other cameras’ sunlight White Balance vary from 5000 to 5700 kelvin, but they’ll all have a slight magenta boost to compensate for the sun’s tint.

I agree with that.

Depends on the tint bin of SST20 being offered. Fireflies has been good as of late for providing pink-tinted SST20s. The problem with the SST20s however, is they turn greener at lower outputs, whereas the Nichia emitters mitigate that. Having 12 SST20s in my ROT66, each individual emitter is really low output for most of my usage, so the 804 brings them back down to a neutral duv (to my eyes anyways). In turbo modes, they look just as pink as SW45k.

great info! thank you
I agree that Sunlight has a higher DUV than Incandescent, and that the Optisolis also has green tint…

I think the Optisolis makes a great daytime light, as its green tint is hardly noticeable, when sunlight adapted…

I sold my 6500k Optisolis for two reasons
the color temperature was “too” blue for my after sunset use
and the green tint was not to my liking in the evening, when I was adapted to incandescent

otoh, 6500k is a good option for a mechanic working in the sun… the sw45k is pathetic for that…

1 Thank

Thanks. Figured it might be something like that, along with niche flashlight producers generally being very small potatoes to the big manufacturers.

I think Tint is affecting peoples judgement here .whenever you put two lights side by side you'll almost always choose one over the other and even your favorite high color rendering light suddenly looks green when side by side compared to a "Better" emitter. So everyone knows what subjectively they think is a better light . John is just comparing his own lights against his own lights and noting what he subjectivly thinks he likes best .

The somewhat green hi cri sst 20's and the high cri 219C's made everyone rethink what they thought was true of hi cri . The rosyiness of the 219B 45k made lots of people equate red or rosy as being high cri .

Being able to fix a tint with a Lee filter ..and doing it to a green tint that almost everyone considers as just plain nasty is a pretty easy and quick n dirty fix .

the issues of the uv coated lenses that Olight and 4 7's used for so many years that created foul green tints hasn't even begun to be addressed . Or the fact that the newer emitters in general seem slanted towards green too.

My simple thought is that high cri is paramount ....output is relative and pretty limited in most cases and tint can be fixed or modified . but low cri is only fixed by changing the emitter .

I used to love to bring up the inverse square law to illustrate the point that your brain can't see moderate increases in power .Visually the differences between 300lumens and 600 lumens is minuscule ,it looks like someone made a bad mode spacing choice on a driver .... but you can't miss the difference between low and high cri ..My point was.. I want to pay for what I can see ..Not what I can't .

AWB.

SST-20, 4000K

LH351D, 2700K

XM-L2, 3000K (don’t remember, between Ra 80 and 85, Opple is undecided)

XP-L2, 5D

I think the camera does with AWB what our mind does, so this is a fair comparison. Data says, the color rendering quality should degrade top to bottom. Eyes say…?

bottom two inferior as expected, 2nd is compensating its crap r9 by being low cct. also, cri is only really perceivable on natural high contrast items, like skin. wood works, but it’s lower contrast than skin a lot of the time. plus, the cameras people use are not great - and most are in sRGB space, which is drastically smaller than what we percieve. not to mention many monitors are crap.

Either way, 4000-5000k is ideal, and cri doesn’t matter a whole lot except on high contrast natural items. the biggest issue with most LEDs is blue bump, cyan dip. the red dip is not very perceptual to our eyes, because we hardly see the cutoff anyway

In direct comparison with the bottles the SST-20 wins hands down. The XM-L2 is not bad, but a little weak on cyan. The Samsung is terrible. Outside, the 4000K SST-20 feels a little to warm, but still performs incredibly.

Rendering comparisons on a single piece of wood without other other colors around is a fairly useless comparison other than simply showing how that particular object looks under the light. Where CRI/Ra really shines is giving more comparative visual information. A rosy tint will make the wood look good, but it also makes everything ELSE look very rosy, which may be pleasant, but doesn’t offer the same benefit to distinguishing/identifying colors.