LED light color rendering

[Luminescent]

There have been a few recent threads relating to incandescent vs. LED performance in Fog, Snow, Rain, and for general color rendering.

Color rendering is a function of the overall spectral balance (color temperature) and is also influenced by the overall smoothness of the output spectrum (color rendering index).

A white LED source has a big spike in the spectrum at the upper end, due to blue light leaking through the phospher layer, and the output spectrum balance is influenced by the phosphor mix used and the consistency of the manufacturing process.

Even under ideal conditions, the spiky nature of the white LED's spectrum will put an upper limit the quality of the color rendering.

Unfortunately, most incandescent flashlights do NOT necessarily have a very good CRI either.

Yes, it's true that under LAB or studio lighting conditions, with a very high temperature halogen lamp running from a STABLE voltage source, you can get a very good Color Rendering Index, but in a flashlight, due to the variable conditions of battery voltage, the lamp is rarely running near the optimum voltage needed for a good CRI.

Incandescent flashlights shift spectrum as the battery voltage changes, and some LED lights start out with horribly skewed color balance right out of the gate due to manufacturing variability (the famous Luxeon Lottery).

To look at the practical color rendering quality that can be achieved by LED lights and incandescent lights, I thought some actual example images would be helpful.

Here are some examples from several LED lights, and a classic 2 D-Cell Maglight incandescent.

For each sample image, a reference image shot with a daylight balanced xenon flash is shown on the left and the light being tested is shown on the right. For consistency, in each case the white balance of the camera is locked to the "Daylight" setting unless otherwise noted.


Note: Due to the 'Luxeon Lottery' any light can get a good or bad bounce on color rendering, so I am listing the names of these lights only for reference, not to imply that these particular light models have any better color rendering than any other.


Dorcy 3 D-Cell (Luxeon) -

My Dorcy 3 D-Cell Luxeon is a few years old but has a virtually perfect white beam. The light has a very deep reflector, and is so tightly focused that it will out throw any of the Mag incandescent lights and most current generation of 200 lumen CREE lights.

Folks can talk all they want about how great incandescent lights are on color balance, but it's a fact of nature that your eyes need a certain minimum level of illumination to perceive any color at all.

The Dorcy 3 D-Cell light will show clear colors at distances where a incandescent Maglight's beam shows only indistinct dim gray outlines.



Jetbeam C-LE (CREE P4) -

The Jetbeam C-LE was quite inexpensive for a high quality CREE light, but has an orange peel reflector with a very clean white beam.


Fenix L0D-Q4 (CREE Q4) –

The L0D-Q4 is the smallest light I own, but also has one of the best color balances.


Maglight Incandescent 2 D-Cell (camera daylight balanced) -

This first Maglight incandescent image shows just how far into the RED the spectrum of a incandescent light appears when viewed using the same 'daylight' color balance on the camera that was used for the LED lights above.

Some claim that incandescent lights work better in fog, because the droplets don't backscatter this redshifted light as much as the purer higher color temperature white light generated by an LED. This may be true, but an LED light with an AMBER FILTER should do just as well.


Maglight Incandescent 2 D-Cell (camera tungsten balanced) -

This final image shows that with the proper corrections applied (as would be done by your eye's natural white balance), even the reddish lower color temperature light from an incandescent can give reasonably good color rendering. Though this looks much better with the proper corrections applied, it's still a little bit of a stretch to conclude that the incandescent is better overall.

In general it looks like the incandescent does a bit better in the oranges and reds, where the LED's give more pop to blues and fluorescent colors.

 

[Zenster]

Very interesting, and thank you for your effort.

Those first shots that show near perfect color alignment of LED lights sure accentuate how poor the incans look in the later pics.

Still, I'm looking for some clear amber filter material I can use to experiment with and see how a white LED (Q5 specifically) changes in appearance when filtered with the amber.

Personally, I have no problem at all with the color rendition of LED's at night, and in fact, I now prefer a nice bright Q5 such as my L2D-CE/Q5, Taiblo A9s, and DBS w/WCQ5, over any incan that I have.

I just want to play with an amber filter over LED for the fun of it.

 

[SilverFox]

Hello Luminescent,

Excellent study...

I find it very interesting that while the brown color on the chart seems to show up under the various lighting conditions, your background color seems to change.

Tom

 

[Handlobraesing]

There have been a few recent threads relating to incandescent vs. LED performance in Fog, Snow, Rain, and for general color rendering.

Color rendering is a function of the overall spectral balance (color temperature) and is also influenced by the overall smoothness of the output spectrum (color rendering index).

Color temperature is not a factor in CRI.

CRI is a mathematically computed value derived from the SPD. All black body radiators such as an incadescent lamp and the sun has a CRI of 100.
http://www.eere.energy.gov/buildings/info/components/lighting/lamps/incandescentlamps.html

Chromaticity defines what the light looks like when you look at it, not how it will render iluminated objects and CCT or correlated color temperature is the best fit to blackbody radiator color temperature curve.

A traditional cool white fluorescent has a relatively smooth spectrum compared to a new tri-chromatic lamps, which as the name implies, have three very strong spikes.

A traditional cool white phosphor with a relatively smooth spectrum and a 4100K CCT has a CRI of ~65

Spikey spectrum from trichromatic 841 phosphor provides 4100K @ CRI ~85.

A white LED source has a big spike in the spectrum at the upper end, due to blue light leaking through the phospher layer, and the output spectrum balance is influenced by the phosphor mix used and the consistency of the manufacturing process.

Even under ideal conditions, the spiky nature of the white LED's spectrum will put an upper limit the quality of the color rendering.

Pesonally, I don't think CRI represent how color is represented to us. Candlelight and sunlight are both 100 CRI by book, but they show color the same way.

Unfortunately, most incandescent flashlights do NOT necessarily have a very good CRI either.

Yes, it's true that under LAB or studio lighting conditions, with a very high temperature halogen lamp running from a STABLE voltage source, you can get a very good Color Rendering Index, but in a flashlight, due to the variable conditions of battery voltage, the lamp is rarely running near the optimum voltage needed for a good CRI.

By definition, all blackbody radiator is given a CRI of 100. Don't ask me why, I didn't design the CRI system. For camera test, perhaps centering white balance to a sheet of white paper under each light source before taking the card photo would show how well the light renders color. Camera also don't respond the same to our eyes, so I don't think it's very good at simulating what we see anyways.

If you shoot film under fluorescent, you get a green cast that you don't see with your eyes, because film responds differently to the green spectra from mercury emission. CRI was designed to represent how a light source renders color to our eyes and don't represent the way CCD or film respond.

 

[EV_007]

The color rendering of LED vs. Incandescent seem to pop up more frequently as of late, especially when there is a new product release from a company which seems a bit coincidental.

Many have gone out of their way to bestow the virtues of LEDs within controlled white wall beamshots and color charts. Nothing wrong with that, HOWEVER, as I've stated before, the human eye is far more capable of rendering color, brightness and the various gradations in hue than ANY captured medium such as film and/or digital. The fact that the human eye can perceive detail in both highlight and shadow areas of a scene without blowing out highlights speaks volumes whereas the digital medium cannot even approach that of the naked eye

With that said, when demonstrating or comparing various light sources, it will NEVER look the way it does in person. Sure, we can represent some basic characteristics of the lights via beamshots, but unless we are all standing there in person to truly compare, images will never do it justice.

I'm an ncan guy, I admit it, HOWEVER, I do like the efficiencies of LEDs.and advancements are being made more steadily than incans. I've found that incans perform better outdoors through atmospheric haze/fog, rain, etc… I have tested many from both camps and although the LEDs impress me more and more, the simple fact remains that the full spectrum light of the incans are not out of the game yet.

That being said, cooler colors seem to pop out under LED illumination and warmer tones seem to render more accurately under incan lighting. Incan illumination seems to be a bit better in the contrast department. Shooting glasses are yellow or amber tinted for a reason, they are not blue or purple.

If you compare similar sized bulbs and reflectors of incans and LEDs powered by similar power sources, then you'll have to ask yourself, which performs best for YOU. No amount of copying and pasting from Internet sources on color spectrum and charts can convince others on what YOU feel is the right answer.



This is not directed at any person, just in general when I use YOU as an identifier.

 

[KeyGrip]

If the Mag is the only incandescent light you have, I suggest trying one that is properly driven. It is really shocking.

 

[MikeLip]

The comparison of the Mag (and incans in general) is misleading. The light does not look that orange in use, *unless* there is another light source to compare it to. Just as fluorescent does not seem green unless there is a truly white source around. But shoot a picture under it and ugh! You need a magenta filter to come close with film, and some photoshopping to fix it with digital, unless your camera can handle it.

That is because your brain adapts to the color temp and makes white things look white.

Incans sometimes do better because they are a thermal source, and their spectrum is continuous. If you look at a white LED spectrograph, there is a chunk missing, generally in the green region of about 500nM. If the colors you want to differentiate are in or near that missing chunk, you're going to have difficulty with them. These chunks are not missing from incan light.

I really can't wait for someone to create a white LED that is truly continuous! Then I could dump my Welch Allyn exam light!

 

[easilyled]

From Wikipedia:-

Although an objective measure, the CRI has been criticised in recent years as it does not always correlate well with subjective color-rendering quality for real scenes, particularly for modern light sources with spikey emission spectra (e.g. fluorescent lights or white LEDs). It is understood that the CIE is looking at developing newer color-rendering performance metrics.

Personally I am pleased that these threads are questioning the previously
held view that incandescents have superior colour rendering period.

It is becoming quite apparent that both have strengths and weaknesses
and its not clear-cut at all as to which is better.

It boils down to type of application, personal preference and visual
interpretation.

 

[Luminescent]

Hello Luminescent,

Excellent study...

I find it very interesting that while the brown color on the chart seems to show up under the various lighting conditions, your background color seems to change.

Tom

This is a very good point, which cuts right to the issue of what the CRI is all about.

The phosphor mixes of the Luxeon and CREE's are different, and give a slightly different shape to the lower end of the spectrum, but all the LED sources are weaker in the deep red than an incandescent source and showed the background fabric more coldly than it would be rendered by a broad spectrum source like daylight.

Also, I suspect that something specific in the organic dyes in the background fabric reacts strongly near the center wavelength of the blue driver LED wavelength and is being tweaked by the big blue spike in the white LED's spectrum.

Broad spectrum illumination (incandescent or daylight) spreads out the upper end of the spectrum and tweaks this specific wavelength LESS, so the fabric reflects less blue when illuminated by a broadband source, and this along with the greater red level mentioned above gives the noticeably 'warmer' tone.

These things together account for the fairly noticeable shift, because with slightly less red and slightly more blue you get enough of a shift to be quite noticeable.

I think this is very much a 'worst case' matchup, but it does give a good example of why the CRI is important, and why the LED manufactures have been working to improve it.

As to why the browns and reds in the color targets look closer to the mark, I think these color swatches were created with dyes that are more broadband and not so twitchy, so they will give a better overall idea of the color rendering of a source.

 

[Dances with Flashlight]

What a great educational thread. Thank you all.

 

[Hitthespot]

More good reading gentlemen. I will have to go over it a couple more times but I still feel from looking at my own comparisons that LEDs have a ways to go before they render color the way incand do. The charts Bernie posted showed the blue spike and lack of reds in the LEDs. This not only effects color but IMO depth perception also. Having said that LED is the only lights I have used in the last 6 months which says a lot about how far they have come.

Thanks Luminecent and EV_007 I really enjoyed your posts.

Bill

 

[EV_007]

More good reading gentlemen. I will have to go over it a couple more times but I still feel from looking at my own comparisons that LEDs have a ways to go before they render color the way incand do. The charts Bernie posted showed the blue spike and lack of reds in the LEDs. This not only effects color but IMO depth perception also. Having said that LED is the only lights I have used in the last 6 months which says a lot about how far they have come.

Thanks Luminecent and EV_007 I really enjoyed your posts.

Bill

Glad to help "illuminate" the subject.

And I do appreciate anyone that takes the time to go out of their way to present subjects which brings out the flashaholics in all of us. Healthy debate spurs the addiction.

 

[KROMATICS]

Folks can talk all they want about how great incandescent lights are on color balance, but it's a fact of nature that your eyes need a certain minimum level of illumination to perceive any color at all.

The Dorcy 3 D-Cell light will show clear colors at distances where a incandescent Maglight's beam shows only indistinct dim gray outlines.

While I appreciate the effort I find this is rather odd. It's hardly a fair comparison if the lights you are comparing put out drastically different amounts of light. I can see colors better in daylight than in moonlight. This is supposed to be a test of color rendering not lumen output. Judging from the comment above you seem to be under the impression that a 2D Maglite is the best one can do for an incandescent flashlight. Yikes!

A comparison between two similar lights (a Surefire A2 and a Surefire 6PL for example) would be legitimate as they both put out around 75 lumens.

 

[Luminescent]

While I appreciate the effort I find this is rather odd. It's hardly a fair comparison if the lights you are comparing put out drastically different amounts of light. I can see colors better in daylight than in moonlight. This is supposed to be a test of color rendering not lumen output. Judging from the comment above you seem to be under the impression that a 2D Maglite is the best one can do for an incandescent flashlight. Yikes!

A comparison between two similar lights (a Surefire A2 and a Surefire 6PL for example) would be legitimate as they both put out around 75 lumens.

Actually my comparison was for two similar lights, a 3 D-Cell Maglight and a 3 D-Cell Dorcy Luxeon. Two almost identical sized lights, running on exactly the same batteries, and with one you see colors and with the other you see MUSH at the same distance. If the Dorcy puts out more light and has much better throw (and does it with more than 10 times the runtime) that's just the way the cookie crumbles. Don't ask me to cripple one light to accommodate the weakness of the other.

Your right though, it's not a fair comparison, because the Luxeon emitter in the Dorcy light is already YEARS out of date and it's only about TWICE as bright as the Mag incandescent, so I should have used something like a 3 D-Cell Maglight incandescent against a 3 D-Cell Malkoff modded Mag LED drop-in, which would have maybe 5 times the output. This would have pushed the threshold of useful color vision out to an even greater distance.

Because of the way our eyes work, brightness is a HUGE factor in getting good color rendering when you are near the limits of the lights throw, so I think this is a perfectly valid point.

I used a 2 D-Cell light for the test picture only because I had a good bright new un-darkened krypton bulb and two very fresh D-Cells for that light. That's why the last image example with the camera set for tungsten balance looks as good as it does, otherwise even the final tungsten balanced image would look way too red.

I am well aware that Surefire incandescent lights will outperform a Maglight, but I dumped my only Surefire incan after it let me down one cold drizzly night when I was changing a tire on a very dark and lonely stretch of country road. If I hadn't had a backup inexpensive Dorcy department store LED light under the seat, I probably would be dead now (clipped by the first car to come around that blind curve).

As far as the Surefire Incan vs. LED comparison goes, I have no doubt that the Surefire incandescent will look better than the Surefire LED lights. The question is will they look good compared to a top quality LED light?

Personally I have only seen a few Surefire LED lights (a G2L and a couple 6PL's). None had what I would call a particularly nice even white color (but then again that Dorcy 3 D-Cell light in the pics above did cost me pretty near $10 on clearance at Target, so I guess it's not really fair to expect that Surefire would be able to match that tint quality for only $75 to $150 dollars).

In the end it looks like the Luxeon Lottery (or it's CREE equivalent) applies to all lights regardless of price (you just feel worse when you get a stinker on a $100 plus light).

 

[easilyled]

More good reading gentlemen. I will have to go over it a couple more times but I still feel from looking at my own comparisons that LEDs have a ways to go before they render color the way incand do. The charts Bernie posted showed the blue spike and lack of reds in the LEDs. This not only effects color but IMO depth perception also. Having said that LED is the only lights I have used in the last 6 months which says a lot about how far they have come.

Thanks Luminecent and EV_007 I really enjoyed your posts.

Bill

What you don't mention is that the incandescent spectrum, although not
as spikey is very deficient at the lower end of the spectrum.

It may be a straight line roughly, but its not a level straight line but a steep
gradient starting at nearly zero at the blue end and going to very high
at the red end.

They call this continuous emission because its not spikey, but in fact its
inherent weaknesses and strong bias towards reds creates the problems
in color rendering that so many of the pictures in this thread illustrate.

Just look at post #12 by Newbie in these spectrographic charts:-

http://www.candlepowerforums.com/vb/showthread.php?t=110795

You can see spectra for both LEDs and incandescents here.

Its very clear just how little blue and green are present in the incandescent spectrum.

In fact there is a more equal balance of blue to red in the LED spectrum than
for the incandescent spectrum

I have found it interesting that in daylight on a sunny day at 10.00am I have
shone both incandescent and LED flashlights on my white ceiling next to a
patch of sunlight and the beam of light from the LED appears to be much
closer to the beam of sunlight than the beam from the incandescent which
always is much redder.

 

[Gaffle]

This past summer I did my own "led vs incan" shootout. I just shined my Mag up into my tree. At the time there were threads about incans being able to penetrate foliage better (somethine like that...it was awhile ago). Well, I was able to see just as far up the tree with my Propoly as I was with my Mag. For me, color rendition is not why I use my light.

Last night I decided to shine my C-LE up into my tree. There are a lot of squirrel nests up there. I found a nest just because I knew where it was. The LED did not bring out the brown dead leaves to life. Since I search for squirrel nests once in a blue moon, I suppose most of my color rendition will be fine.

Good post Luminescent.

BTW, I am exceptionally happy with my C-LE. Great frickin light!