CRI ratings - how useful are they?

[Karel Bata]

CRI - the Color Rendering Index. The higher the number (max 100) the better the color rendering. In theory.

 

http://en.wikipedia.org/wiki/Color_rendering_index

 

OK, it's quantifiable, and it's scientific. But how useful is it really in practice? When would a low CRI be acceptable? And just how low?

 

It wouldn't surprise me if there are lamps with a low CRI that actually work quite well with film or video. Some HIDs (aka HMIs) have a CRI of only 85, yet I've heard they work well.

 

Any thoughts?

[John Holland]

Sorry pressed wrong key !! I have never heard of CRI [as in this meaning ] so just wouldnt worry about it at all !!! . Use your eyes !!! you can tell what sort of quality and colour of light thats is hitting the subject you are shooting .!!!!!

[Karel Bata]

Can your eyes see the infamous green spike then...? :D

 

A CRI rating is a bit of a new thing, but I've been noticing them lately, particularly on these new fluorescent lights. And I'm shocked to discover that some lamps now sold as HMIs are not the same as those we've become familiar with.

 

I don't see how one number can be of much use though, other than as a warning. I think for most of us, if we just hire in from a decent company we can rest assured their lights meet the expected standards.

 

I'm still curious though. ;)

[John Holland]

No they cant see "the green spike " but i have always hated everything looking totally colour corrected , like a bit danger !! i dont know if thats the correct way to put it ? but everything now looks the same , just like a Kodak demo film !!! yuk .!

[Adrian Sierkowski]

Generally, you want as high a CRI as possible so you can then dictate what color to make the light via gels etc.

However, sometimes, you really want a nasty, ugly, gritty look. In which case, you may opt for the WORST CRI you can find. The Color Rendering Index, however, is really only dealt with in my experience in terms of floro fixtures. HMIs technically have one, in fact, all bulbs (sans Tungsten) ought to have a CRI, but for the most part, anything made for film, will be in the mid to high 90s, hence why a lot of people don't use CRI in terms of their film lighting packages. You just know that a Kino tube or an HMI isn't as kind to skin tones as a Big-Eye, for example.

But, as I mentioned (and sorry I'm a little scatters-shot, taking a break from editing a live event I shot for a friend... with children, which means the interviews are painful to scrub though), you generally want high CRI and then you modulate it, or you go for a low CRI for the fact that it looks "ugly."

[Karel Bata]

Thanks Adrian - that all sounds sensible. ;)

 

I'm just a bit surprised to see these domestic coiled fluorescent things with built in ballasts https://www.warehouse-lighting.com/category...cent-Lamps.aspx get a rating of 85 (and you know they must have spikes) while a (supposed) HMI provided by a film lighting place http://www.filmandvideolighting.com/msr57hmibuno.html gets a rating of only 80! And it says Philips on the side. But what's this? "Color Temp: 7200K" ? :huh: The Orsam HMIs we're all familiar with are rated at 5600 (though they come out of the factory at about 6000).

 

So is that Philips bulb rubbish? Is the CRI number the culprit here? Or am I missing something...?

 

 

P.s. John, I know exactly what you mean! ;)

[David Mullen ASC]

I wouldn't take the CRI ratings too seriously other than as a rough guide. I think Kino tubes are near 95, but depending on how hot the fixture gets, the green spike can get worse, etc. My notes say that a Cool White tube has a CRI rating of 62 and a Vitalite has a rating of 91... but obviously there are a lot of variations in Cool Whites out there, the cheapest ones have a huge green spike (horrible or wonderful depending on the look you want).

[Adrian Sierkowski]

Wish I could tell you better. Personally, I've never checked the CRIs of HMIs, I generally just get 'em form the rental house and trust in the equipment room gods.

As for the CFL bulbs, I have a big one which I got from Home Depot. It's a 64w fixture with 300W output, and I have no idea what the CRI of it is, but I know I used it on my XDCam first, and she was fine, visually, and have recently used her on S-16mm without much problem. with a lot of these newer bulbs it's a matter of testing, ya know? Generally, for film, you want a CRI of 90+, so it is possible that Philips bulb, while an HMI in construction, is designed for non-film useage. HMIs give more light per watt than Tungstens, so i'd not be surprised if it was used in some form of architectural or construction lighting, like those big light towers, as a lower CRI bulb should be cheaper to produce/purchase, and, as it's an HMI, give more light per watt making it cheaper to use to light things up. This is why, for example, Sodium Vapor lights are used, and as far as I can recall, work on a similar principle to HMIs. Florescents have the same benefits, and same drawbacks, hence why they're used.

[Karel Bata]

Cheers Guys. ;)

[John Sprung]

OK, it's quantifiable, and it's scientific. But how useful is it really in practice? ....It wouldn't surprise me if there are lamps with a low CRI that actually work quite well with film or video.

 

Well, it's sorta scientific. It's actually not perceptually linear, definitely not one of CIE's finest hours. It's all based on the 1931 two degree standard observer, so it's all about how people see colors. It has nothing at all to do with film or video.

 

Indeed you can put minusgreen sleeves on flourescents and get light so purple that it's outside the range for which CRI is even defined, and yet it works OK on film. You can pull the sleeves and have a high CRI, even though there's that green spike that shows up on film.

 

Bottom line, CRI does a poor job of measuring something that isn't even relevant to us. But it's in our faces all the time because, like shop vac horsepower, it's a number that manufacturers can print on boxes that get sold at places like Home Depot.

 

As so often happens, the only real way to know is to shoot a test. But there is one trick you can do. Use a scrap DVD or CD to diffract the light, and see if you get any spikes. Compare a flourescent with incandescent or daylight to learn how to interpret the results.

 

 

 

 

 

-- J.S.

[Richard Andrewski]

I'm just a bit surprised to see these domestic coiled fluorescent things with built in ballasts https://www.warehouse-lighting.com/category...cent-Lamps.aspx get a rating of 85 (and you know they must have spikes) while a (supposed) HMI provided by a film lighting place http://www.filmandvideolighting.com/msr57hmibuno.html gets a rating of only 80! And it says Philips on the side. But what's this? "Color Temp: 7200K" ? :huh: The Orsam HMIs we're all familiar with are rated at 5600 (though they come out of the factory at about 6000).

 

So is that Philips bulb rubbish? Is the CRI number the culprit here? Or am I missing something...?

 

There's all kinds and grades of "hmi" bulbs although the one you showed is not an HMI bulb. Only Osram's HMI (it's their brand name and doesn't mean anything else) is an HMI, everything else is an HMI compatible technically. The particular bulb you showed would be for use in follow spots, theatrical lighting, some architectural lighting and especially follow spots, hence the 7200K color to make the spot pop out all that much more from the other lighting. The CRI is more than high enough for those uses nor do they care about 5600K/6000K being a standard. And that's also why its not a hot restart kind which is why its not a true HMI compatible--hot restart is one of the major criteria for whether its a full compatible bulb or not. It will most assuredly operate off the same HMI ballasts but you wouldn't want to try and relight it before about 5 or 6 minutes after it was turned off. The horrible screeching sound tells you its not a good idea and its best to shut off the ballast right now.

 

As far as CRI goes, the lower CRIs will manifest themselves as an abundance of green when using discharge lighting like fluorescent or HMI and even many lesser expensive LEDs in an image taken that was lit with the offending light.

 

Actually CRI 80 is not all that bad especially in digital mediums where custom white balance can take care of any issues in many cases. It really starts to get bad below that and white balance can have a tougher time working on it. Especially when other kinds of light get mixed in. It really gets confused at that point and the results can be nasty.

 

I've posted an example color chart I took of a Xenon light I have thats about CRI 60 to give an idea of what that can look like. This is just camera white balance at 5600K, no custom white balance.

 

 

Needless to say, that one isn't much use for anything other than as a back or rim light.

 

For comparison sake, here's one taken with a Cool Lights CRI 87ish fluorescent tube

 

 

 

As for how CRI is really measured by manufacturers, we use something called an integrating sphere which is a kind of isolation chamber hooked up to a spectroradiometric computer:

 

 

And that gives you an accurate reading from 1 to 100 of the CRI index. You can get a report printed out from the test session which has all the stats about the tested bulb from color temperature, lumen output and CRI among other things.

 

Other than that, you could estimate CRI based on color charts like the ones above.

[Karel Bata]

Does that thing double up as an orgone acuumulator? :D

 

I've gotta say the more I read the less I'm impressed. To somehow quantify the quality of a light into one number is just dumbing down the whole science. I can understand why that happens, and how people make money out of it, but in the end, beyond choosing a light to illuminate an office shop or warehouse, it doesn't seem all that useful.

 

I don't actually see that first clor chart (the xenon) as being all that bad. It's a bit green, but the colors look pretty vibrant, while the second looks a tad washed out. A bit of copy, color adjustment (nothing complex that can't easily be done in post or with gels) and we get:

But maybe I'm behind the times here. In these days of video post and DI, is a low contrast image actually more desirable..? I did a course in Shake a while back and someone from Cinesite said to keep the contrast of images down, because you can always turn that up later, but you can't recover shadow detail lost in a contrasty image. This was a shock as it went contrary to all my experience of lighting and getting everything right in camera!

 

All right then. with that in mind I'll tweak the contrast and brightness of the second (basically to get that gray scale right):

OK, I'm ready to have my thinking or methodology criticized (and I didn't spend a lot of time on this either) but to me, after a bit of correction, the xenon with a CRI of 60 looks better than the flluorescent rated at 87. But really, with these two there's not a lot in it...

[Richard Andrewski]

Sure you can tweak it because it was the only light shining on the color chart in both cases so its easily controllable just like white balance could have taken out the green too. Thats not the point.

 

The point is that the CRI 87 light is easily mixed with real daylight and there won't be any grainy / gritty issues in the pictures taken with it. The CRI 60 on the other hand, if mixed with daylight or another kind of light will create a mishmash that may not be so easily be sorted out.

 

The chart may not be vibrant because of the angles I was shining the light on it. If I had bounced the light off a white bounce card, it might have been more uniform. It may also be a bit overexposed too. Take on different days with different cameras.

[Karel Bata]

Fair enough. ;)

 

Though it seems to me the xenon is perfectly usable if gelled correctly - with any luck a minus green would correct it. After all xenon lamps are used in many types of projectors.

[Marc Roessler]

Richard, are you really sure this was real xenon and not some knock-off as used in car "xenon" headlights or similar?

 

In my experience xenon is quite sun-like and has excellent color rendition... Osram specifies a CRI of > 95 for their XBO line (commonly used in film projectors)... Those are also used as color matching lights.

 

Greetings,

Marc

[Richard Andrewski]

Richard, are you really sure this was real xenon and not some knock-off as used in car "xenon" headlights or similar?

 

In my experience xenon is quite sun-like and has excellent color rendition... Osram specifies a CRI of > 95 for their XBO line (commonly used in film projectors)... Those are also used as color matching lights.

 

Greetings,

Marc

 

Yes indeed, a real bonified Xenon car 35w 6000K headlight. Back in 2007 I was experimenting with some 35w hid bulbs of various kinds for use in a small pepper fresnel (equivalent in size and output to something like an Arri 150 tungsten but only using 35w power) to complete our "CDM" line of fresnels. They all had the common problem of low CRI. As Karel points out you can color correct it out if thats all you're using. Most of the time though, you don't have the luxury of using only one light for everything. Real daylight streaming in the window, etc. That's where you get in trouble with these low CRI types. They would be acceptable however with a bit of minus green but I was holding out for some other low watt / high power output technology where the CRI was a bit higher. In reality, it would be just fine for most things you use a pepper fresnel for like back / rim / hair lighting but I don't want to fight customer perceptions so we'll wait for a better solution before we do the lowest wattage model in the CDM line.

[John Sprung]

I've gotta say the more I read the less I'm impressed. To somehow quantify the quality of a light into one number is just dumbing down the whole science. .... beyond choosing a light to illuminate an office shop or warehouse, it doesn't seem all that useful.

 

Yup, that's it exactly.

 

 

 

 

 

-- J.S.

[Marc Galerne]

CRI is just a value. As all values, they are a landmark.

CRI of a 100 is basically the color rendering of colors exposed to natural daylight.

You want to be as close as this number and then you do the effects you want to see.

If you have lamps with a bad CRI, what your eyes see is not totally accurate as you can check in the chart enclosed. So you might, from this wrong information, start to over expose a bit or change you camera setting to adjust this to YOUR own eye.

We all know that, depending on your film stock and your video camera technology, colors will be different from what you saw on the set. They will probably be altered in post production, anyway.

The thing is we judge from the start with our own vision. CRI and color Temperature are guidelines to help the cameramen create the picture with a greater accuracy of his own vision.

[Karel Bata]

But Marc, even if you can quantify a human eye's response like that (and I have serious doubts about that!) any other visual recording medium will react differently. You would need a different number for each medium.

 

Xenon car headlamps with a CRI of 60 would be totally unsuitable in (say) a carpet warehouse where the customer needs to 'see' a color accurately. So the system works well for that application. But it turns out that with a bit of correction a color chart is rendered rather well, so the low rating is unjustified as a universal measure. The CRI system is a good heads up for potential problems, but doesn't seem to be of much real help when trying to make choices between lamps for use in film/video.

[John Sprung]

Yes indeed, a real bonified Xenon car 35w 6000K headlight.

 

Very interesting -- What kind of ballast/support circuitry does it take to run a Xenon headlight? How long does it take to strike and come up to full output? None of my cars are new enough to use anything beyond halogens.

 

 

 

 

 

-- J.S.

[John Sprung]

The CRI system is a good heads up for potential problems, but doesn't seem to be of much real help when trying to make choices between lamps for use in film/video.

 

Right, that's because CRI is based on the CIE 1931 tristimulus curves for human vision. You could do the math again substituting the spectral sensitivity curves for the three layers of a color film stock to get a number that's specific to that stock, ditto for electronic cameras, too. Then we'd have to keep track of them as "Human vision CRI", "Kodak 5219 CRI", "Panasonic Varicam CRI", etc.

 

 

 

 

-- J.S.

[Karel Bata]

Err... not sure that would work. I think you'd have to start from scratch with the sensors calibrated differently.

 

"What kind of ballast/support circuitry does it take to run a Xenon headlight? How long does it take to strike and come up to full output?"

I've never heard of arc car headlamps! :D I think it must be something like this:

"TPO Bulbs are made with the latest filament technology and have high pressure Xenon gas fill. This gives you maximum efficiency and purity of color. TPO bulbs give you the brightest, cleanest, light you can get without exceeding OEM spec. In fact, we guarantee it! If you find another bulb that meets OEM specifications and has brighter, cleaner light--EVER--we'll give you a full refund on ours!"

So that's just a 12V supply. (Well more like 13V actually) Are you thinking what I'm thinking? :huh: (But it only has a CRI of 60! :D ) I wonder what all those bits inside do? :o

 

This must be exasperating to Richard and Marc... :angry:

Edited May 30, 2009 by Karel Bata

[Marc Roessler]

John, the "xenon" car lamps are not tungsten lamps with xenon gas but real arc discharge lamps.

The "xenon"is only marketing wish-wash. It's actually HID discharge lamps with a tiny bit of xenon added so the lamp will light up right away when starting the car. However, it takes some time until the HID really kicks in and reaches full brightness (just as a HMI does) - you can notice this when starting the lamp.

 

Greetings,

Marc

[Karel Bata]

Is that true of all xenon lamps? Including the ones pictured above (where the seller claims they are "made with the latest filament technology")? Or has their marketing dept got it all wrong?

[Richard Andrewski]

I think I mentioned somewhere recently in a post on here about HMI bulbs that they're used in follow spots a lot these days. Well the Xenon bulbs are getting quite a bit of use for that too. They have a much higher lumen per watt rating than HMI bulbs so more efficient. When you see the huge moving beams cutting into the night at some place in your city (like a night club or whatever trying to attract people there), its generally a Xenon these days. So powerful, they're used in a lot of military lights too. Like this surplus tank light that some hobbyists snap up:

 

http://www.candlepowerforums.com/vb/showthread.php?t=92396

 

 

They're super efficient and create an incredible beam. We've looked at a lot of the 35w type bulbs which are also used in HID flashlights. The ballasts are super simple and can be very small. Nothing great just yet but you never know what you'll stumble across in research.

 

We've also tried some high wattage LEDs from 30w to 100w (there are even 300w ones now) but the issue with those is they need super large and heavy heat sinks so they aren't very suitable for small and portable lighting. CRI is an issue in most of those also and they are mostly used for street lights these days.

 

Luxim is making an interesting "plasma" type bulb with a high CRI but so far the unit comes sealed in a very large package that includes a solid state bulb and ballast with a fan so thats not a very usable unit yet either for film/tv lighting. We're always on the lookout though.

Edited May 31, 2009 by Richard Andrewski