Is CRI really dead?

[Chris Ficek]

If your in an environment where you are only dealing with two mixed sources, usually sunlight and something else, you have to decide which light source will be the easiest to control (or alter), and that often will make the choice for you. If you can eliminate or control some of the light and get it down to the least troublesome light to use I always take the easy way and match to that. Closing curtains, putting garbage bags or cardboard over windows is a pain but it eliminates the ever changing sunlight (or streetlights) and lets you then control all the light sources. The fewer gels the better as you have mentioned but if you are using flos or leds you almost alway have to gel.

[Bill Davis]

Sure it makes sense to "gel the source" in a situation where you're facing mixed lighting.

But that's OFTEN not possible.

You going to gel the sun? Gel every square inch of all the exterior windows in an office suite. Gel 500 fluorescent tubes on the big factory ceiling?

Yes, people do this. For a big budget feature or for something of lasting value where the stakes are high, good gaffers do whatever it takes to control the lighting. I know a guy who for a twice a year shoot for a company that sells corporate jets, not only rents every Kino flo in this state, he imports more from California at well.

That's the point. Experience people light for the job. They don't buy a fixed group of lights then try to fit EVERY job to the capabilities of that kit. That's the road to mediocre results.

FWIW

[Tama Berkeljon]

CRI May be quite properly outdated...
some of the issues I see with CRI are the following- some of you have already commented on these, but I thought I would riff a littlle... Hope you dont mind!

CRI was based around colour rendering using Tungsten as the reference source, and does not really take in to account discontinuos spectrum sources such as HMI, Fluro, Or LED acuurately.

CRI Uses a small sample size (8 samples) to generate the averaged result .
These samples are typically all from low saturation colour samples and have an uneven spacing around the black body locus (the central line that runs through the planckian graph, along which is 'white' light)

Unfortunately, CRI can not tell you any info about which direction your light is incorrect, just that it is bad. (In the Film and Video, world a CRI > 90 good, CRI< 90 bad, which is not strictly the case. we need to look more at how our gear responds to actual colours rendered.)

I Agree with what Richard Andrewski said about green, unfortunately CRI has been the only thing that indicates if a light is deficient or dominant in some area for quite some time now (it was invented in the 60's)

I've Just written a bit of a blog on all things CRI, and also a bunch of links and images on the subject, if anyone is interested in having a look, the post is at http://tama.typepad.com/blog/2010/01...l-to-work.html

The good news is there may be a replacement for CRI in the Pipe, the Colour Quality Scale.
More samples from a broader range of saturated colours, giving more ballanced colour data..

Let me know if you have any thoughts or comments, would love to hear from you all !

Cheers,

Tama

[Bob Grant]

Tama,
The link to your blog is not working.

[Tama Berkeljon]

Thanks Bob, It was momentarily down,as I linked to an accidentally created post that was a duplicate, then deleted. My apologies !

Here it is again, Or just click on my signature.


CRI, LED's, CMOS and actually getting it all to WORK. - Tama Berkeljon's blog

Cheers,

Tama

[Chris Ficek]

Well Tama, I must add as the instigator of this thread that your reply was exactly what I was hoping for.
Of course I appreciate all the other replies and insights but you have given me some food for thought, however I must tell you I disagree with a few of your statements, particularly about the colour bias of CMOS and other semiconductor materials-----BUT---- I'm going to double check my facts before I blurt out a reply as you have also turned a few of my thoughts around. I have been looking at the CQS system as well as discussions on an LED only colour standard from some of the major LED manufacturers and will chime in on those topics as well.
I anxiously await your postings of meter readings (or should I say mis-readings as you seem displeased with the results) from your work with non-full spectrum sources.
Now we’re cookin’
Cheers Chris

[Tama Berkeljon]

Thanks for the Kind words, Chris ! Your instigative question was one of the motivating factors for me to finish my blog article and get it out - sometimes I need a bit of a push !
When It comes to silicon devices- my example with the diodes is slightly exagerrated to demonstrate the point. silicon purity and homogenity govern the electrical characteristics of any piece of silicon, and because silicon has a grain structure that is variable due to purity, and has a cut thickness that can vary even by small amounts, my understanding is that these factors can affect the way silicon interacts- even by small amounts.

If anyone can clarify this further I would be very interested, as these are only my own observations and also the result of conversations I have had with material scientists who work in the field of silicons.. Not popular things to publish if you are in the business of making money from silicon chips, but as a designer or an engineer you may come across this quite a bit..

Cheers !

T

[Marty Welk]

allintext: doped cmos etched - Google Search

i assume like a lot of the semiconductor things, they use the Doped and Etch type of stuff, to make the gate things. I have seen a video on the processes, and a person could say that the processes of making (at least some) semiconductor devices is what i call "organic"

we think of many of these things as locked rock stable made to microscope specs with robotics and clean rooms and that somehow every Atom in the process is exactally where it should be. instead the Microscopic surfaces of the semiconductors looks like rugged mountains of randomness, not perfect surfaces or aligned blocks.

They are all still made of highly purified elements of this earth, being eaten away by acids, coated with chemicals/elements , and other such "organic" processes, they dont physically or mechanically align every atom :-) Yet! The semiconductors i saw being made went through 6-10 processes cleaning, etching, doping, coating, cleaning, and back around again.

So i would ASSUME that is why a signal comming from each miniature photo diode gate thing (pixel), could vary one from the other, and especially when it is at its first "trigger point" the lowest light hitting it, where one item will react before another, and that being more noticable change from 0to1 vrses 50to52.
i assume, that is where color grain noise comes from as different locations trigger slightly different, add in that the filtered light waves trigger it, probably have randomness, and effect the triggering of the diode depending on where they land exactally.

all that can also be observed in many other semiconductor gates, like multi-emitter leds and transisters in processors and all.

So the whole array of collector items, would have variances among the "pixels" but each overall Cmos chip item, should be similar to the next one made in that batch. so the variation of each Pixel item would be pretty visable, the randomness there. but each of the whole chip items should be relativly the same.

like a Patch of organic DIRT :-) with different specs in it, but wide whole areas of dirt, that look similar to the rest of the dirt.
so each cmos of a batch should be overall similar to its brother in the bath.
then you have the chemicals , and baths and acids all changing over time, just like the batch of chemicals at 1Hr photo has to be changed up, cleaned, added to.

Why does Chip grain look like a small version of Silveroxide grain on Film , because they both have somewhat organic random distributed components reacting to light slightly differentally.

No i have no idea what i am talking about, but that is what i gathered from seeing many of the processes.

[Marty Welk]

Oh the thread topic :-)
we now have 8 different "colors" of lights in/out of this one house, most with rotten Spectrums, many with Rotten "CRI" pan five feet and the color is different .

What is it going to be like when there are leds of 20 "white" shades, florescent of 10 types, 3 forms of incadescent, plus the wonderfull sun.
I wonder if they shouldnt start doing somthing NOW, about the camera overreaction (compared to human eyes) to the different colors and spectrums and lacks of CRI in the efficency lighting. Single color lighting is Dead, that is for sure.
ATW is not capable of coping with the differences at all.

[Chris Ficek]

Ouch! My head hurts now:)

I'm still digging up some info here but I'm sure in the end I'll conclude (and share) that any real issues with the colour interpretation of a CMOS is more influenced by the camera circuitry that lies beyond the CMOS sensor in the image processor circuits. Any influence by the CMOS's random manufacturing variences will be infintismal (sp) when compared to the influence of the image manipulating actions within a camera.

More on this later.

[Bob Grant]

Thanks to Tama for bringing the proposed CQS standard to my attention. One concern I have is it relates to human perception of color which is not the same as how our cameras see color. Also the new standard uses more saturated colors and gives some weight to color differentiation as something desirable in human perception. Also of note it's been jigged so the existing fluro lamps get the same score to avoid confusing the public.

Certainly the new standard is better than the old, not certain though how much it will really help in our game and I sure lack the knowledge to translate how they are rating colors and the new color spaces being used, into how our cameras work.

[Chris Ficek]

Marty--I disagree! Single colour light is NOT dead, single colour fixtures maybe yes, but we have to correct existing fixtures to generate light that works for our cameras. We dont have to accept poor light unless we are lazy and dont want to get out the gel kit. Until they come out with a camera that can have multiple colour correction capabilities and a sensor that can activate the various options automatically as we change the light falling on the sensor our only option is to correct all the light sources until we are happy with their uniformity. In the very near future I think we will be seeing some fixtures that have very high effiency as well as very high CRI and full spectrums. Keep our fingers crossed until then.

[Giroud Francois]

CRI is not about light temperature. You can get an yellowish light with excellent CRI (but a yellow cast) and a white light with bad CRI.
fluorescent tubes or LEDs are known to give good white light with bad CRI, because the technology of exciting a gaz (giving mainly UV light) and transforming it into visible light by a photonic reaction (the chemical layer inside the tube or over silicon) can only give bad result (CRI wise). sodium vapour light is the worst example.
Mainly because this give burst of energy in particular wavelength (like laser are) .
So it can happens that almost all color can render correctly and only one color cannot render at all (giving a grey instead purple for example).
there are tons of color chart you can find showing the problem.
It is particularly difficult to correct because you cannot use regular color correction applied to all the picture. You have to set the pixels that are bad and replace them with good color.
and you cannot correct that in the camera either.

[Marty Welk]

Even when they have LED with Multi Phosphors, which they are working on making as we speak. single Color Leds that gain thier efficency with phosphors just like the white ones do with phospors. meaning they can put groupings of RGB phosphor based leds in for lighting, in the same House , someone will have florescent there, and Some incan in the bathroom, or down some hall, or some halogen thrown in there.

The phosphors are what is making the light efficient in ccrt led and florescent, even when they pepper these things with different color outputting phosphors or mix groupings of different colors, there is spikes in the color areas, not a full spectrum. When they have a full spectrum, they arent "efficent" at all. a person could even say that efficency is just tricking the human eye. it just doesnt fool the camera :-)

using RGBYC leds on (say) the top of a video camera without phosphor based leds, takes almost as much power as a halogen :-(

Phosphors being energized outputting the photon things do not live forever, different phosphors die sooner/later than other, more or less of one type or the other are needed, and therefore, the one type is depleted before the other , , etc etc. so phosphor based (efficient) lighting with multicolor spiking phosphors will also change color over the life of the output item. So some stuff will look great on a chart today, and a year from now , mabey not so.
i think most people can observe that in thier phosphor based florescent lighting of different types and (original) CRI indexes. something a little different about that older bulb that has 7000 hours on it ?

Energy to light efficency isnt magic, the way they are getting energy to light AND being efficent about it has major ramifications to the spectrum, and untill they are all made by ONE huge corporation the ramifications are all different.

dead , it is dead untill they make the Fusion light and replicate the sun everywhere :-)
they have the Microwave Plasma light out, and its spectrum again different.

Sure WE can get lights with good "CRI" or just good RGB and send that RGB into 3 RGB chip things, but the stuff they are working on for efficent house and commertial lighting is all over the spetrum. Different things in different fixtures IS efficent by design of the fixtures themselves.

Over in the light forum, i dont see some standard Efficient lighting comming on the horizon, i see 50 different ones, each with thier advantages and disadvantages, and little with any concideration for video. i luv that efficient lighting, just dont like it to try and video with.

if your saying we should be Buying light for doing all the lighting of the entire set , with a "CRI" index number things, we should indeed. If we want an expanded CRI in EFFICIENT lighting it will be little spikes in the colors the chart is supposed to show correct :-)

If we want the full total spectum, then we still need the energy to output that full total spectrum from something?

[Chris Ficek]

When I started this thread I thought it would go in a different direction but the comments have been very interesting. I'm thinking it may be time to throw out a statement that may stir up some reactions.

Really now, aren't all these claims that CRI is outdated and not important for consideration in selecting an LED device for video lighting just the skewed ramblings of LED light manufacturers that can't get their products to perform very well and still make a bundle selling them. Surely it must be possible to make a good LED video light without all the colour issues at a fair price. I have seen LED lights offered for $75 and for $7500....there must be a way to bring the price and quality paradigms closer together, afterall a high power top-of-the-line 80+ CRI LED like a Lumiled Rebel only costs $6 in its raw form. MAybe it's time for some of the big LED boys to fall off 5mm LED and look at the new breed of high power LEDs.