We hear a lot about Color Temperature (CT) and CRI or Color Rendering Index in film and video production. You may have even heard of “Correlated Color Temperature” or CCT (and what is that all about?). Some are concerned about their ability to have colors rendered accurately and consistently each time they record an image of a product or person. They are also concerned with the color of the light they are using. These two concepts should not be confused but they often are. In the beginning I confused them just like many. In the past few months however, as I was searching for suitable bulbs to sell with my Cool Lights fixtures, I got a major education in more aspects of fluorescent tubes than I ever thought I would want or need! I never planned to have a “Cool Lights” brand bulb but the realities of what is commonly available in Asia drove this decision. I had to find an established manufacturer willing to produce a higher quality bulb, but still economically, or I would not be able to offer the bulb I wanted at the cost I wanted to. Before we go further, let’s define CT, CCT and CRI so we have a common basis of understanding.
Color Temperature and Correlated Color Temperature Defined
CT is measured in Kelvin or “K” for short. Kelvin is a measurement of temperature but what does that have to do with light though? The best way to think of this concept is to imagine a piece of metal (actually the proper term is “black body” which is a theoretical device–for our purposes we’ll refer to it as metal because its easier to relate to) being heated up and all the various stages it goes through while it is glowing hot (red, orange, yellow and finally white hot). A piece of tungsten goes through most of these stages too as it is being heated up to become a point or hard light source (what a coincidence!).
Figure 1: Color Temperature Spectrum Chart (wikipedia.org)
So, the comparison of the various colors from the visible spectrum of light in heating up the metal, and the corresponding temperature at that color, is how we get “color temperature” (see figure 1). We are just comparing the color of the light with the color of the metal at some point in its heating up process. For all tungsten lamps, this value is a true measure of the color of the light being emitted from the filament (and mostly falls into the range of around 2500K to 3200K). Common light sources and their color temperatures are shown in Figure 2.
Figure 2: Common Light Sources and Their Color Temperatures (compiled from various sources)
For fluorescent or high intensity discharge (HID) lamps this value is only approximate or a simulation (there is no piece of metal being heated up to produce the light in this type of lamp) so it is supposed to be called Correlated Color Temperature or CCT. So when we look at manufacturers specifications, we notice that they all express the color spec for fluorescent or HID bulbs as “CT” and not “CCT” in all their literature. For simplicities sake they just call it color temperature. We’ll just have to let it slide. It’s too big of a problem to tackle to get everyone to call it CCT and change all their packaging and brochures. From now on though, you know what this means when you see a fluorescent lamp specification.
Tungsten color light sources are not the only type available today: like the ever-popular “daylight” simulation type bulbs for instance. When you hear the term “daylight” this is simply referring to whatever color our fictional superheated metal block happens to be at a certain point to equal one of the typical colors of daylight (mostly from about 5000K to 6500K).
Color Rendering Index Defined
Once you have expressed what the color of a lamp is with CT or CCT, you also would want to know how accurate is that lamp compared to the ideal for that CT. CRI or Color Rendering Index is that measurement. This is where it starts to get easy to confuse the differences between CT and CRI. Many have the misconception that CRI is an absolute measure of all light of any color temperature all compared to each other. This is not the case however. Rather, it is more a relative measure (each individual light source CT is just compared with its ideal and not how it stacks up against all other CT’s) which will be seen in a moment. For our purposes, the color measurement and rendering ideal standards fall primarily into two categories: man-made like tungsten or natural like daylight. That’s where CRI comes in, as a scale of 1 to 100 to express how accurate a light source is compared to its standard or ideal—hence why I said it is a relative measure, not absolute. It’s worth repeating again: CRI is NOT a measure of how accurate a light is absolutely. It IS a measure of how accurate a light source is compared to the ideal for the standard category that light is in. Since tungsten really IS a metal block being heated up (as in our color temperature test to determine Kelvin value) the corresponding CRI value is 100% (or darn close) because tungsten is so close to our theoretical metal block in its properties.
With daylight type sources it’s a bit trickier. Artificial daylight is typically produced in HID and fluorescent type (not tungsten type) lamps and is expressed in a CCT value. Its CRI is then compared to the ideal of whichever color of daylight it happens to be closest to. If you thought of CRI as an absolute measure, it would be really easy to think that daylight renders colors better than tungsten light sources.
While it may be true that colors look better in the whiter light of daylight, it is not true that all daylight sources will always yield higher CRI values than tungsten ones. Remember, we compare apples to apples and oranges to oranges. Therefore all tungsten colored sources are compared to the tungsten ideal and daylight colored sources are compared to the daylight ideal (the CT of the sun in the sky in different conditions). As you may be aware, it’s perfectly possible to find a low CRI daylight bulb that is not as good as a higher CRI one—hence why we have the CRI value to tell us the quality level of the light each bulb produces.
Measurement of CT, CCT and CRI
So how do we find all these values? In a complicated and expensive piece of test equipment called an Integrating Sphere (see figure 3 & 4) in combination with a spectroradiometric computer peripheral.
Figure 3: An Integrating Sphere (photo by Richard Andrewski)
Figure 4: Testing a 200 watt not-so-compact fluorescent (photo by Richard Andrewski)
When most manufacturers tell us these values, we have little choice but to believe them for the most part. Color can be measured with a color meter, but CRI and Spectral Energy Distribution (see next section) can really only be measured and intelligently printed out in the Integrating Sphere and spectroradiometer combination of test equipment. Before you run out to see about getting one, you should budget several thousand dollars. That is unless one of your neighbors happens to have one. Then maybe you can borrow theirs. For the most part, only light bulb manufacturers have these instruments–and most of those manufacturers are in China today.
When I was evaluating bulbs, I had to find a neutral, friendly manufacturer that would help me measure sample bulbs as they came in. Why did I do this? I was calling their “bluff” about their quality level and color temperature basically. Again, how many people really have access to an integrating sphere? The manufacturers know this simple fact and the more dishonest among them can exploit it. Many times a test report that came with a bulb from a manufacturer was shown to be wrong compared to the test I performed with my friend’s test equipment. It was so easy for a zealous sales clerk to go get a stock bulb off the shelf and just represent that it was a specially manufactured color temperature and CRI that I specified. For instance, many times I asked for a 5600K 85 to 90 CRI bulb and got a 6500K 80 CRI bulb.
There is so much room for liberal “manufacturing” of specifications along with the manufacturing of the bulb; so much so that you really are never sure of what you are getting unless you make some spot inspections from time to time!
Spectral Energy Distribution
The spectroradiometic peripheral prints a report at the end of its testing cycle that includes, among other things something called a Spectral Energy Distribution or SPD for short. Figure 5 shows an SPD comparison between two light sources: tungsten (2800K) and fluorescent daylight (5000K).
Figure 5: SPD’s comparing incandescent and fluorescent lights (wikipedia.org)
This shows the spikes or bands that light falls into with a discharge type light such as a fluorescent as opposed to the smooth curve across all colors that tungsten provides.
During our testing phase we tested another brand of bulb for comparison sake. The chart in figure 6 shows an SPD for a Cool Lights 3200K bulb followed by figure 7 which shows a comparison of an Osram Dulux L 3200K bulb.
Figure 6: Cool Lights 55 watt biax (3200K) SPD (Cool Lights USA Testing)
Notice the balanced spikes in the Cool Lights bulb which is CRI 88. In addition, the light output is 2900 lumens. The actual color temperature of the tested bulb was 3147K.
Figure 7: Osram Dulux L 55 watt biax (3200K) SPD (Cool Lights USA Testing)
Oddly, the Osram Dulux L 55/930 – 3200K bulb (actual tested value was 3007K) was also a CRI 88 (the spec only claims 85 but we tested 88) also but showed decreased lumen output (2500 lumens) over the Cool Lights bulb and a more apparent spike in the green range. Below, in figure 8, we show a Cool Lights 5600K biax bulb to compare how its SPD looks to that of the 3200K charts.
Figure 8: Cool Lights 55 watt biax (5600K) SPD (Cool Lights USA Testing)
The Cool Lights 5600K biax bulb has a bigger green spike but this is to be expected, there is more green and blue in 5600K! The CRI is 88. Lumen output was 2900 and actual color temperature of that particular tested bulb was 5430K. Unfortunately we didn’t have another 5600K bulb to compare to as there aren’t that many manufacturers that provide color temperatures in that range. Many fall in the 5000K and 6500K range—not close enough for comparison.
Since we are primarily talking about fluorescent lighting in this article, it might be a good point to talk a bit more about what is commonly available on the market and why it is such a big deal to find CT and CRI suitable bulbs for media production work.
Lack of choices in CT and CRI have been, in the past, one of the main reasons that fluorescent lighting has gotten a bad reputation for use in creating visual media. The perception developed that ALL fluorescent light flickers (because of older magnetic ballasts used in fluorescent fixtures that drove bulbs at 60 times a second or 60hz) and renders color poorly (thanks to the low CRI “cool white” or 4200K color T40 fluorescent tube commonly used in older style “magnetic ballast” fixtures). Another problem with this “classic” fluorescent tube is the big “green spike” in its spectral energy output you hear everyone talking about. They are referring to the spike in the color spectrum in the green range which is produced by the energized mercury vapor inside the lamp. Why mercury and green? Because of all the colors, green is best responded to by the human eye. So the light at the very least appears to be brighter thanks to the mercury. We’ll talk in more detail about this process elsewhere. All this was stated to say that the huge energy savings that could be realized by using fluorescent has in the past been secondary to the aesthetics involved for video and film producers.
None of this has to be the case any more though! Today, with electronic ballasts that drive bulbs at frequencies upwards of 20,000 times a second (20Khz) and all of the CCT and CRI options available to simulate various kinds of light–a choice exists now, not only to save energy but also to have lighting that can be very flattering when used properly. Is it a complete substitute for use in all stage/studio lighting? The answer has to be “no” because there are some cases where a point (also known as hard) light source is absolutely necessary to dramatize or produce a certain effect; and fluorescent (a broad or soft light source) is incapable of being made into a hard light source due to its very nature (but I’ll cover more about this and the energy saving alternatives in hard or point light sources in another article). In addition, film seems to show the effects of the green spike much more readily than digital medium (DV and Digital Photography) which are quite a bit more forgiving thanks to their white balance mechanisms.
Fluorescent Tube Manufacturing Process
All fluorescent tubes are coated on the inside with a phosphor powder which glows when the mercury gas inside is charged, producing the light when the bulb’s electrodes are charged (see the article on our website “Green Spike or Why Do Fluorescent Lamps Have Mercury in Them” for more details on this. The quality of the phosphor used inside the bulb determines the quality or CRI of the light that the bulb produces. The particular mix of different colors of phosphor produces the CCT of the bulb’s light. Normal, commonly available phosphors are very inexpensive and yield anything from about 70 to 82 CRI. To get above 82 CRI takes a lot more work (and correspondingly, more cost too). Part of those costs are the rare earth, tri-phosphor powders that are necessary to produce a higher CRI lamp; but few manufacturers use these powders or even fully understand the process of making a tube with them.
Most of the fluorescent tubes in the market today are now produced in China using the normal, inexpensive phosphor powders and two commonly available CCT’s: 2700K and 6500K (the world lighting industry “standards” for simulation of incandescent “tungsten” light and daylight respectively). The CRI of these common lamps is typically 80. Other CCT’s and CRI’s are available on a custom manufacturing basis for a large quantity order. It should also be noted that another barrier to high CRI lamps is that as CRI goes up, lumen value drops as the “green” or mercury spike is lessened, so is the amount of light or apparent amount of light coming from the tube making the bulb less energy efficient. A balance has to be struck between energy efficiency and aesthetics. Why choose 90 or higher if 85 to 90 is good enough for your application? That sounds controversial! Maybe it’s another article for another day.
CT’s for Video Production
Of course, in video and film production, we are mostly concerned with 3200K and 5600K (the white balancing standards for tungsten and daylight) but we are only a small minority of the total users of fluorescent lamps in the world. This is why 3200K and 5600K bulbs are still a bit more expensive than the others (in addition to the rare earth powder options which produce higher CRI values too). So the 2700K (a bit too red or warm) and the 6500K (a bit too blue or cool) are not ideal for our use. Grocery stores, warehouses, office buildings, and other common users of fluorescent lamps do not care that much about color or CRI. They care about bulb cost and operating life in that order. In fact, the bigger the green spike, the better they like it. That means more apparent light.
Figure 9: Cool Lights 55w Biax Bulb CRI >87 (photo by Richard Andrewski)
All is not lost for us though! That’s where Cool Lights and selected other photography/film/video specialty companies come in. We order the large quantities necessary to be able to produce the media standards for CCT and CRI. So now you understand that if you see a company offering fixtures for film/video/photography use, but they only offer 2700K or 6500K bulbs, it should be evident that they did nothing special in ordering their bulbs from China (in fact, they took the cheapest available stock bulbs) and you can also correspondingly expect that the CRI of these bulbs will also be low (80 average but possibly lower). These decisions are made primarily to economize on the bulbs that are bundled with fixtures. The perception being created is that the company is doing the client a big favor by bundling the first set of bulbs with the fixture. Perhaps for some photography uses, this default CCT and CRI may be just fine; but for many other uses it would be totally unsatisfactory.
We covered the basics of fluorescent tubes and their suitability for media production use. The commonly used terms of color temperature, correlated color temperature and color rendering index were all covered to explain what they mean to media professionals and so we have a common basis of understanding when looking at the specifications of fluorescent lighting tools for video, film and photographic use. We also covered the common color temperatures used in productions and why they are not-so-common in the general market place. In future articles we will cover other aspects of energy saving stage and studio lighting.
Wikipedia: “Color Temperature”
NEMA Website: “Why Do Lamps Need Mercury”
Photos: Richard Andrewski / Cool Lights USA
(c) 2007 Cool Lights USA. All Rights Reserved. May not be used without the permission of the author.