High CRI LEDs for Photography / Colour Calibration Underground

[jarvist]

I split this off from the Bisun review, as I didn't want to hijack the thread, interested in any comments!
http://ukcaving.com/board/index.php?topic=12767.0

The high CRI thing was from 2010.

Only one year off!

I'm not sure whether they'd be any better for cave use than the standard LEDs of appropriate colour, and there's a fair efficiency hit compared to simply using more neutral/warm regular LEDs.

For a photo lamp, I'd imagine you could run into CRI problems quite quickly. It's certainly a headache to have different types of LED sources in one shot. It's not just a matter of colour balance there, some LEDs seem to have chemically different phosphors. Certainly some white LEDs fluoresce (orange) under external UV light, and others don't (though this could be a result of a UV blocking barrier coating).

I'd imagine the low CRI of high efficiency LEDs is because they are essentially mixing the blue GaN directly with a semi-transparent Green (+ lower frequency) phosphor, effectively leaving a big dip in the spectrum between the blue peak and the green. So anything aquamarine will render weirdly and ill defined, and violets won't exist. ( The caving relevance I'm thinking of for this is formations like Daren's Blue Greenies )

Of course, more subtly, for photography, there's then also the question of how the rather 'peaky' spectrum of the LED interacts with the colour sensor in a digital camera (+ assumptions about the spectra of incident light used to regenerate a RGB value) or the dyes in a colour film, which may not be correctly described by the CRI.

Fundamentally at the end of the day though, nothing in caving has a 'true' colour as there's never any daylight down there to illuminate it. So you can get away with slightly red / purple tinted mud, as long as the rock isn't green it'll look OK!

Does anyone bother taking a colour chart caving with them? It'd be interesting to see how good the rendering was. The calibrated ones are pretty pricey, but I wonder how close you could get by simply getting a photo-print of the RGB values (i.e. find a Tiff image on the internet), as long as the printer itself was well calibrated...

[Bob Mehew]

Try http://www.qpcard.se/BizPart.aspx?tabId=28&tci=51 The 201 card is 14cm by 4cm.  Not tried it underground.

[footleg]

Shoot in RAW and you can adjust the white balance on the computer when you convert the RAW files to edit them. Better than letting the camera do it in the cave and risk it getting it wrong. You can tweak it until you are happy at your leisure above ground.

I have photographed with P7 LEDs and with XP-G LED home made lamps. In both cases I have not had problems with getting the colours to look right.

[jarvist]

In both cases I have not had problems with getting the colours to look right.

For sure. I do the same. In particular, editing software that has a 'colour temperature' and then a 'green' channel (like UFRAW) works well for tweaking LED lit scenes, I imagine as the Green peak is so pronounced (and variable between emitters) in LED light.

However, that doesn't really make it right, it just makes it not obviously wrong...

The problem is that the LED spectra is so 'peaky' + so mismatched to a continuous black body spectrum that there are colours that simply don't exist when illuminated by LED light (hence 'colour rendering' index), or end up shifting in hue. This is the same effect that makes people look sickly-green under low-quality florescent tube lighting.

Halogen, electronic (Xenon) flash and flash bulbs are all actually a pretty good approximation to a black body, as is sunlight. At that point, the only variable is the 'temperature' of the spectra.

Try http://www.qpcard.se/BizPart.aspx?tabId=28&tci=51 The 201 card is 14cm by 4cm.  Not tried it underground.

That's a much lower price than I've seen elsewhere for a calibrated target. I guess I'll add looking into this to my 'winter evening projects' ToDo list! It should be fairly easy to use a calibration target + a camera profile at home to generate some kind of correction curves for different LED lights, and compare to profiles for electronic flash / flash bulbs. From that you should be able to infer how much of an effect on photos it will have, and perhaps even generate colour profiles for LED emitters to use within your 'RAW' camera workflow, even if you're not taking a colour target caving with you every time!

[Fulk]

I find that compared with film, digital images of caves tend to be biased towards a 'warmer' rendition when using electronic flash, so that, for example, grey rock turns out well – grey on film, but reddish-brown on digital (I'm talking about bare limestone, not mud-covered stuff; the latter suffers even more).

Oddly enough, anything in the shot the colour of which is known – oversuits, tackle bags, wellies, a 'model's' face – seems to accurately portrayed.

Sometimes, before going off to take some snaps underground, I'll check that flashguns and slave-units are working by firing off a couple of shots in the garage – and the colours there seem spot on.

This I simply don't understand. Why does electronic flash render 'known' colours OK, but cave walls differently?

[paul]

This I simply don't understand. Why does electronic flash render 'known' colours OK, but cave walls differently?

Could it be your perception of the colour of the cave walls? After all, you normally see them using some sort of artificial lighting which will have some sort of colour cast. Since everything else in your photos seem to be portrayed fairly accurately colourwise, I would tend to think that the cave walls are as well.

I was surprised at the difference underground when first using LED lighting after years of yellowish halogen or carbide lighting.

[Fulk]

Could it be your perception of the colour of the cave walls?

Well, yes, but – if I perceive that everything else looks 'OK', and that cave wall looks grey – then why is it that the only thing that doesn't fit with my perception is the cave wall?

Actually, I've noticed this phenomenon in other people's pictures, not just my own – whether it be in the pages of Descent, in these UK Caving threads, or whereever.

[footleg]

I find that compared with film, digital images of caves tend to be biased towards a 'warmer' rendition when using electronic flash, so that, for example, grey rock turns out well – grey on film, but reddish-brown on digital

So with film, the film was colour balanced for daylight (unless you used specialist film). So with flash which is designed to mimic daylight the colours were balanced as expected. But digital cameras have the ability to adjust the colour balance (or white balance as it is usually called in the menus). So they make a best guess at the setting to use based on the scene (unless you set the white balance to a specific setting). Different models of camera chose different settings and you have no way to be sure you get it right. Hence my suggestion to use RAW, because then you actually set the white balance after taking the photo and can pick the setting that works best for your photo back in the comfort of your home.

[footleg]

LED or Flash? Can you tell?

[Fulk]

LED or Flash? Can you tell?

Dunno – but brilliant pictures!

[Roger W]

1:  LED?
2:  Daylight?
3:  Flash?

As Fulk says - brill pictures!

[jarvist]

So with film, the film was colour balanced for daylight (unless you used specialist film) (.....)

It is actually a lot more subtle than that. Colour (print) films are still 'corrected for' at printing (i.e. a 'white balance' is applied, curves are rebalanced to stop shadows ending up blue etc.). These days, most printing is done digitally, so the film is actually scanned at development rather than the direct use of colour filters in an enlarger. Of course, the user never got to see this, as it was done for them unless they were in the darkroom / print shop themselves.

Different films have different spectral responses (not just balances). Daylight vs. Tungsten film not only had varying sensitivities of the Red vs. Blue dyes, but actually choose dyes with different spectral responses, I don't know why but I recall they pushed the blue higher and extended the green response down into yellow / orange. Different film types (even amongst the consumer versions) also had differing chemistries, and therefore differing rendition of colour. And then you have Fuji & their '4th colour layer' technology (basically a weak Cyan absorber) which was meant to help with rendition in odd light, and coping with the 'difficult' colours between green and blue.

No one has ever pretended that any colour photography is perfectly accurate. You can't possibly build a system with the same colour response as the eye, let alone include all the perceptual effects.

With digital all the colour filtration is done on the Bayer filter over the CCD sensor, you have issues with cross-talk of light between pixels on the CCD, and then the spectral response of the CCD itself to contend with. So essentially your digital camera comes with one 'film type' for its entire lifetime.
Perceptive colour is then resynthesised when the RAW file is demosaiced, which is also when a computer model of the colour response of the Bayer filter & CCD is applied. People often don't realise this - your 8MP camera actually has 4MP of green pixels, 2MP of red and blue each, which is also why you often see 'colour noise' in the blacks of a digital caving photography.

You may have noticed that when shooting with 'warmer' light than the digital camera system is optimised for, the reds blow out to fully exposed early. This is because the CCD is essentially 'daylight' balanced at the factory, and so the red channel is over sensitive when you're not at the 5500K 'blackbody' that it was designed for.

So with flash which is designed to mimic daylight the colours were balanced as expected.

Well, different flash tubes produce quite different temperatures, and it also varies depending how hard you drive the tubes.
It's not so much the way it was designed, as the physics of it. It really is a little bundle of gas heated to thousands of degrees Kelvin! I think natively it's often a little too 'hot', rather more blue than daylight. Certainly some old flashes have gold reflectors, which was probably to downshift the colour temperature.

[littletitan]

LED or Flash? Can you tell?

Flash
LED
Flash??

[footleg]

LED or Flash? Can you tell?

<snip>

Flash
LED
Flash??

100% correct Robbie. Anyone would think you do this sort of thing for a living 

I'm not sure how much the colours give away. Generally it is the sharpness of the shadows that gives the clue as LEDs (especially when hand held) give a large area light source and so softer shadows.

[pwhole]

Remember also that 'daylight' itself changes its color constantly, from dawn to dusk, but also as differing weather conditions prevail - so there's little 'standard' in terms of color perception or depiction anyway. Our eyes have the best white-balancing available, and so we tend to ascribe 'true' colors in caves that may not actually be so in 'reality'. Looking outside right now, there's a distinct green-yellow hue (caused by storm clouds) compared with my indoor white light.

All reflected light in a cave will contain the colors of the surface it last hit, so light bouncing off a reddish wall will produce a reddish color cast in the opposite direction (assuming it's more or less flat). But a blue section in another part will reflect blue more, and so the likely color result of a cave photograph with artifical light is extremly difficult to predict. When you have reflective minerals involved, as in many mines, and highly irregular surfaces (pretty much anything underground) it gets even more complex.

The only way to truly create a 'standard' view of a cave would be to illuminate it totally and evenly - ie from some fixed centre-point with a spherical light and no shadows. That wouldn't make for much of an image, but as a 'catalogue index'-type photo, it would probably be handy to geologists and speleologists, as it would produce the most useful color information - ie the least corrupted by color-casts.

[sluka]

Any high power LEDs has its color temperature yo may find it in technical parameters. It is enough to switch color temperature of white balance of camera according to color temperature of light from LED. Don't forget "lightpainting" method when photographer use strong focused light and "paint" the scene by light. See: http://www.korba.cz/preview.php?ID=112

[WolfgangZ]

Hi,
can anyone tell me what a good CRI is? Ok, 100 ist best but what about a value of 80? Unusable, bad, mediocre or already good?
Many thanks
Wolfgang

[footleg]

I've just done some research into exactly what CRI is, and then looked up the data sheet for the Cree XP-G LEDs I built my latest photography lamps with. I am using Cool White R5 XP-G LEDs which according to the Cree datasheet have a CRI of 75. I have been very pleased with these as photo lights. So I would say 75 is good enough. The data sheet also lists LEDs with a CRI up to 90. But the brightness of these is around 62% that of the cool white version for the same input current. So you are trading off brightness at the same battery life. They are also a warmer white.

So I'm not sure that CRI is all that important for photography where you can adjust for a blue colour cast. But it is important for lighting used as general illumination where you want something less cold looking (like in your living room).

I don't have the data to hand, but my previously posted photo was lit by SSC P7 LEDs which are even bluer than my XP-G lamps. They still took great photos and there is no sense of a blue colour cast once you process to RAW files correctly. Here are a couple more photos taken with those P7 lights:


Kickapoo Cavern 3 by footleg, on Flickr


Kickapoo Cavern 4 by footleg, on Flickr

[Mr Mike]

Really nice pics those, what sort of exposure time did you use, and it looks like 2 sets of P7 lights? Was each one just 1 x P7 or did the lights have more than one?

[sluka]

They are also a warmer white.

The warmer white are better for photography just they have more red and yellow spectrum.

[footleg]

Really nice pics those, what sort of exposure time did you use, and it looks like 2 sets of P7 lights? Was each one just 1 x P7 or did the lights have more than one?

I had 3 x lights. each one had one P7 driven at full power (2.8A current). I think the top photo was only using two of the lights. The bigger chamber shot below it had the 3rd light providing general area illumination from back near the camera position.