Posts Tagged ‘Color Balance’


How To Color Balance Mixed Lighting Sources

Tuesday, March 5th, 2013

I just finished a series of four blog posts talking about the various advantages and disadvantages of different types of lighting for photo and video work (the first post, with links to the others, is here), and why I’ve decided, for the time being, to use a mixed kit of fluorescent and LED lights. This kit is great and should serve my needs very well, but there is one hurdle that needs to be overcome first: while these lights are all advertised as “full spectrum” and “daylight balanced” at 5600K, in reality they all have visibly different color temperatures, so they need to be balanced with each other in order to work well being used together to light the same scene. In this post, I’m going to give step-by-step instructions on how I took three different lights with radically different white balances and balanced them to work together beautifully.

Before: Unbalanced

Before: Yuck. When white balancing for the mini LED panel on the left, the fluorescent softbox in the center is very green and the LED 1x1 on the right is slightly magenta.

In the image above, which I designed intentionally to exaggerate the color balance differences of the three lights, I placed (from left) a miniature LED panel, a fluorescent softbox and a 1×1 LED panel next to each other and aimed them at a uniformly white ceiling. The difference is striking (and awful).

Before: Yuck

Alternately, the same image above, only this time white balanced for the fluorescent soft box in the center, the LEDs on the sides are both overly magenta and orange.

As is, it would be very difficult to use these lights in a scene together without them appearing different colors. So they need to be balanced together. How to do this? Gels! Pulling out my collection of gels, I got to work.

I keep an assortment of gels to color balance pretty much anything (from left to right): CTOs, CTBs, Plus Greens and Minus Greens each in 1/8th, 1/4 and 1/2 densities. With this assortment, no matter which way a light is off balance, I can balance it.

Gels Gels Gels!

Gels Gels Gels!

In order to balance the three different lights (from three different manufacturers!) I started with the one that is most difficult to gel: the fluorescent (this is one of the biggest weaknesses of fluorescent lights in my opinion… they’re a pain to gel). I used that as my basis and then adjusted the other lights to match it.

It is possible to simply judge the color of a light visually in comparison to others next to it, like in the photos above, and to experiment with different gel combinations to get the lights to the point where they visually look the same to the eye, but “eyeballing” it like that is extremely difficult to do accurately; I have a very good eye for color (I scored a 19 on the X-Rite Online Color Acuity test! Take the test yourself, it’s fun!), and even I can’t achieve the level of precision that I want by eyeballing it. So to measure the color balance precisely I decided to use a couple of precision instruments: my camera and computer.

To start with, I set up a simple 18% neutral gray card on a light stand (I use this one from Amazon… it’s cheap and does the job well), along with a color chart. I lit the gray card and color chart with the fluorescent light (placing the light at an angle so that the light is illuminating the card but not reflecting glare). I then blacked out the windows in my office and shut off all the other light sources (overhead lights, computer monitors, etc.) so there was no “contamination” and I knew all the light hitting the gray card was from the light in question, and I took a still photo of the gray card and color chart with my Canon 5D Mark II camera in RAW format.

Gray Card and Color Chart

Gray card and color chart on stand for determining exact white balance of a particular light

I downloaded the photo onto my computer, and opened the file in Photoshop (you could also use Lightroom or any other application that can work with RAW files, I just happened to choose Photoshop). Using the White Balance Picker / eyedropper tool in the Adobe Camera Raw conversion screen (the same tool is in the Develop tab in Lightroom in the White Balance box), I sampled the 18% neutral gray card to set the proper white balance for the image, the values of which are then displayed in the white balance section on the right (it is a good idea to click a bunch of times in a few places on the gray card as the individual measurements will vary slightly, then average the values).

Sampling White Balance in Adobe Camera Raw

The White Balance Picker tool is the eyedropper icon near the top left. I sampled a spot on the neutral gray card, which gave me the white balance values shown in the white balance box at the top right.

Sure enough, I now saw numerically what I had seen visually on the wall: that fluorescent light was very, very green (+28 tint!). Since that is the light that was most off balance, ideally I would have gelled it to match the other lights, but since this light is so difficult to gel and the other are so much easier, I instead went the other way around and gelled the others to match this one.

With the temperature and tint white balance values for the fluorescent light in hand, I then repeated the process (blacking out the room, lighting the neutral gray card with a single light source, and shooting a photo) for each of my other lights and then found the white balance values for them as well (I found that my miniature panel has white balance values of 5100K temperature and -3 tint, and my 1×1 LED panel is 5050K temperature and +5 tint).

With that information, I then knew precisely how off balance my lights were from each other. I then added a gel to one of the lights, repeated the process of measuring the white balance values, and noted the numerical effect of a particular gel (bear in mind that as much as the gel manufacturers try to keep the color of their gels as pure as possible, a Plus Green gel will never be purely plus green…for example, my Rosco 1/4 Plus Green gels turned out to add +28 points of green tint, but also knocked off 300 degrees of temperature. But after measuring the color balance values of each light and the color effect of each gel, it was very straightforward to figure out which gels to add to each light to balance them together.

In the end, to balance my LED panels to my fluorescent lights, I needed to add 1/4 CTB and 3/8 Plus Green (one 1/4 and one 1/8) to my mini LED, and 1/8 CTB and 1/4 Plus Green to my 1×1 LED, which, while not numerically perfect, got my lights as closely balanced as possible with 1/8th-increment gels. Now I can comfortably use all of my lights in the same scene together and be confident that their colors will be balanced and visually indistinguishable.

After: I've Brought Balance to the Force

After: I've Brought Balance to the Force. While I can still see differences on this uniform white wall, in practical use these lights will never appear unbalanced.


Tags: , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , ,
Posted in Equipment, Gear, Techniques | 6 Comments »

Tungsten vs. Fluorescent vs. LED lights: Light Quality (CRI) and Conclusions (Post #4 of 4)

Monday, March 4th, 2013

This post is the last of a series of four comparing the advantages and disadvantages of different types of continuous lights for photo and video work. Here are the previous ones:

Post #1: Choosing Lights: Tungsten vs. Fluorescent vs. LED
Post #2: Efficiency (i.e., power usage) and Heat Generation
Post #3: Portability and Speed of Setup & Ease of Use

Light Quality (CRI)

When choosing a light source, the final (and perhaps most important) issue is the quality of the light that that light source emits; in other words, its spectrum, or “CRI.” As most people know, light is made up of a whole spectrum of wavelengths, which results in different colors (in terms of visible light, red light is at one end of the visible spectrum with long wavelengths and blue is on the other end, with short wavelengths). Different types of light sources (LED, fluorescent, tungsten incandescent, HMI, the sun) emit different mixtures of wavelengths of light, and the best ones, like the sun, emit a nice, broad, even spectrum of wavelengths (without large spikes or dips at any given wavelength), which allows objects of different colors to appear as vibrant as possible in a photograph or video.

Color Rendering Index (or “CRI”) is a measure of the mix of spectrum that a light emits. This is the biggest advantage of the traditional xenon and tungsten lights: they emit the broadest spectrum of light of any of the types of photo and video lights here (this is because they closely mimic the behavior of a concept known to physicists as “black body radiation.” If physics is your thing or you really want to understand this subject in depth, read about black body radiation and Planck’s law). As a result, tungsten and xenon lights have the highest CRI (100, or close to it). LED and fluorescent lights, on the other hand, emit light that contains more of certain wavelengths and less of others (uneven spikes and dips), and therefore have lower CRIs (from as high as 95+ for good quality lights to as low as 60-70 for low-quality lights). The effect of lower CRIs is that some colors, including skin tones, can appear muted, washed out or unnatural in photos and videos. For this reason it is essential to choose lights with high CRI values.

Both my LED panels and my fluorescent lights have CRI values above 90. That is, they emit a quite broad spectrum of light that will illuminate objects of all colors well (that is not to say that they are accurately color balanced or white balanced for any particular target: CRI and white balance are two completely separate issues… more on that in tomorrow’s post). To illustrate the broad spectrum of light from each of these sources, take a look at the images below.

Emission Spectra ©2013 Chris Conti Photography

The emission spectra of my 1x1 LED panel (left) and fluorescent lights (right). Both show good, broad, consistent output. ©2013 Chris Conti Photography

To make the images above, I projected near-parallel beams of light from my LED (left) and fluorescent (right) light heads into an optical prism, which refracted the light into its constituent wavelengths, and photographed the results. Note that all of the colors of the visible light spectrum are well-represented. This is an indication of the high CRI value of these lights.



After experimenting with and testing various lights both in theoretical tests like the emission spectra above as well as practical ones like lighting a white seamless with them, using them for portraits, etc., I’ve made a few conclusions. In no particular order, here goes:

- The concern that LED and fluorescent lights emit poor-quality light as compared to tungsten lights is unfounded. With a tiny bit of adjustment via gels (more on that tomorrow) both from a subjective standpoint (how they look) and a technical standpoint (technical measures of their light emission), these lights look great.

- Both LED and fluorescent lights consume a tiny fraction of the amount of power that tungsten lights do (which makes them more usable in the field), and don’t generate the searing heat of tungstens (which is always inconvenient and can be destructive and painful, and uncomfortable for subjects).

- On the other hand, fluorescents and LEDs don’t generally generate the quantity of light that most tungsten heads do, so it may be necessary to use more of them for certain applications (flooding a white seamless, much less a full cyc wall, requires a huge sheer quantity of light), so these lights might not be terribly well-suited for these applications.

- Fluorescent lights are much less portable and more time-consuming to set up than tungsten lights, but LEDs are easier and faster.

- Light modifiers and accessories like softboxes don’t really exist for LED panels yet (although I did just make a softbox for my 1×1 LED… perhaps that’ll be a future blog post…), but fluorescent heads can usually easily accommodate anything mounted on a standard speed ring.

So what does it all mean? These lights are tools (just like all of our other kinds of gear), and they each have advantages and disadvantages, and are better suited for some tasks and worse for others, and the right tool for the job will depend on the particular job: lighting a large stage with a two-wall cyc wall is still best done with high-power, high-output tungsten or HMI lights. A quick location interview is probably best done with a couple of LED panels. For a small- to medium-sized studio shoot, fluorescents are probably the best bet.

Tomorrow I’m going to be going to a location and shooting in the same room all day. What will I bring? Fluorescents and a couple of LEDs.

I’ve learned a lot experimenting with all these different types of lights. If you’ve read this blog post and the few that came before it, hopefully I was able to share some of that with you. If you have any comments, different opinions or questions, let me know!


P.S.- Since I’m going to be using a mixed bag of different light sources that result in a mixed bag of  color temperatures, in order to work well together the lights need to be balanced to each other. Balancing my mixed bag of lights will be the subject of tomorrow’s post…

Tags: , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , ,
Posted in Equipment, Gear | 6 Comments »

An Unusual Request

Thursday, March 25th, 2010

Take a look at this photo – does anything seem wrong here?


After shooting a portrait for a client the other day, the subject emailed me with an unusual request. Now, in editing portraits in Photoshop, I often trim a few inches off the waists of subjects, remove skin blemishes, etc. This client’s request though, was a new one for me: he wanted me to give him a thicker beard.

Apparently, he had given up shaving a week or so before the portrait was shot, and therefore had several days’ growth of stubble, but wanted a fuller beard for his portrait. In fact, the subject’s beard was pretty stubbly. Here is the original image:


I emailed the subject back that this kind of digital Rogaine would be tricky to do without it looking fake, but nonetheless he asked me to do what I could. I generally don’t like manipulating photographs to such a degree that it crosses the line in my mind from simple retouching (correcting skin blemishes, for example) into outright fakery (which this clearly did), but at the end of the day, my job is to make the client happy. So I proceeded. After my standard, basic Photoshop tweaks (slight Curves, Hue & Saturation and Color Balance adjustments in the form of Adjustment Layers), the image looked like this:


To give the impression of a fuller, heavier beard though, I was going to have to get creative… I decided that in order to achieve a realistic look I would need the texture of actual fibers and strands of hair. So, I identified the thickest part of the subject’s facial hair (which looked to be an area under his nose, highlighted in yellow below), selected it and Identified it as a pattern for the Pattern Stamp tool. Essentially, I was going to use the subject’s own hair to clone in additional hair.

3 patterns

I then used the Pattern Stamp tool to “paint on” the additional hair into a separate layer, varying the opacity from 90% where the appearance would be heaviest around the subject’s mouth, to 10% (in 10% increments) where it would be the lightest, on the sides of his face. The “mask” created by this process is represented below as a Quick Mask, showing the varying degrees of opacity. This process got me close, but didn’t give me quite the look I wanted, so I repeated the procedure using another pattern, this time taken from the subject’s head (highlighted above in green), rotating the selection such that the strands of hair flowed the correct directions. Again, I used the graduated-opacity mask shown below.

4 mask

The end result came out pretty well – I was able to significantly thicken the subject’s beard, but don’t think a typical viewer would look at the photo and think anything had been altered. Placed side by side with the original, however, the difference is substantial:


As I said above, I don’t like altering photographs to an extent that constitutes fakery (the distinction of which of course is a completely subjective judgement, but which in my mind means going beyond simply erasing an unattractive pimple here and there)… I find it distasteful. But as a corporate headshot, the purpose of this photo is to make the subject look good (this isn’t a documentary or photojournalistic shot), and the job of the photographer is always to make the client happy. With this little bit of Photoshop creativity, I think I succeeded.

(By the way, this portrait was shot on my new Canon 100mm f/2.8L Macro prime lens using my Canon 5D Mark II body. This was the first time I used the lens for client work, and it is fantastic! The lens is so good I think I’m going to have to write another blog post about it when I get the time! For now though, great lens, highly recommended!)

Tags: , , , , , , , , , , , , , ,
Posted in Equipment, Field Notes, Gear, Projects, Techniques | No Comments »