I'm shooting theater plays and starting to shoot some alternative fashion shows.

Many colored lights these days are LEDs, not white + gel as they used to be. Let's simplify things a bit and consider that an inexpensive red (or blue etc, but just one pure R or B channel) led light used in a theater will have almost 100% of its power output concentrated on a spread of red frequencies. Do correct me if my assumption is wrong.

Bayer matrix based cameras will have 4 photosites but 3 of those (2G + 1B) will filter out red light. Same for blue. However, those 3 other photosites will still generate noise.

Would I be right in saying that under such conditions I'm essentially dividing any latitude my sensor has by 4? In other words, to get to the same perceived amount of subject brightness I'd be compensating with +2 EV (or +4 EV ??) on aperture, speed or ISO?


2 Answers 2


While it is true that if you're shooting only under red light your camera will collect less light than if you had full spectrum lighting you must also keep in mind that all of the light you are collecting is mostly going into only a quarter of your sensor's pixel wells. The lack of green or blue light in the scene doesn't increase the amount of red you can capture before the red is totally saturated and you begin to lose details in the highlights. You still must expose so that none of the color channels are allowed to blow out.

Another question here perfectly illustrates what happens if you try to increase exposure when shooting under very limited spectrum lighting: Blown out blue/red light making photos look out of focus

When shooting under only a single color of light it is probably more crucial than ever to check the RGB histogram to be sure you're not completely blowing out one of the color channels. Due to the lack of green or blue light, looking at a combined histogram will make it seem like you're underexposing even when red is completely blown out! If your camera, like most, has a monochromatic light meter it too will probably recommend exposure levels that will blow out the only channel of light you are actually capturing. Cameras with RGB+IR light meters are becoming more prevalent and tend to do a much better job of metering in such light, but setting exposure manually is, it seems to me, still the best option.

Also keep in mind that the Bayer color filters are not absolute. The green filtered pixels do not reject 100% of red and blue light, they just let a higher percentage of the green light through than the red and blue light that strikes the filter over a "green" pixel well. Some red and blue light makes it through the green filter. A good bit of red light (over 10% in the example below) will be registered by the green filtered pixels. The blue filters let a significant bit of green light through. Even a small amount of the red light that lands on a blue filtered pixel will be included in the photon count for the "blue" pixels. Each pixel well only registers a single luminance value and can't tell the difference between red, green, or blue light that makes it through each filter.

enter image description here

The reason LED lights don't give us accurate color isn't necessarily because they are dimmer than more full spectrum lights, it is because they emit a narrower spectrum at whatever brightness they emit. When we combine a red, green and blue set of LED lights, they still don't emit light at a lot of wavelengths that some objects in the photo would reflect if light at those missing wavelengths were present.

  • \$\begingroup\$ Under strongly coloured light I start by underexposing 1 stop and adjust from there (making use of the RGB histogram). I often end up at 2 stops to get the saturation. And of course raw is the only way to go. This also allows you to lift the exposure a little in post to get faster shutter speeds and freeze movement on stage. Most meters have a bias towards green, the question is how much of a bias (although a simple silicon photodiode is most sensitive in the orange/red) \$\endgroup\$
    – Chris H
    Commented Sep 14, 2016 at 14:20

Getting a bit more technical than Michael's excellent answer... Here are the spectral curves for a red LED and a Roscoe red filter. red led

Roscoe Red Fire

You can see the dramatic difference in spectral range. This may not matter much if there's no other illumination in the scene (i.e. standard light meter might see both as 'pure red'), but if you look up similar curves for other color LEDs vs. gels, there's likely to be a lot of 'spill' across the 3 or 4 filters in your camera.

  • \$\begingroup\$ You've picked a slightly orangey red example there -- there are filters which don't turn on until 630-650nm and won't excite a green subpixel at all. \$\endgroup\$
    – Chris H
    Commented Sep 14, 2016 at 14:17
  • 1
    \$\begingroup\$ @ChrisH True, but that's not the point of my post. Anyway, until you know what gels are being used, you won't know where the cutoff is. \$\endgroup\$ Commented Sep 14, 2016 at 14:22

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.