2

So I'm picking up a ColorChecker. Partly because I'd like to have more accurate colors and (let's be honest) partly out of curiosity and boredom. My question is: I understand that colors should frequently be calibrated for color-critical studio work, but how often should they be recalibrated for outdoors/available light work?

I don't plan on carrying the passport around with me everywhere I go, not least because I shoot Leica for the minimalism and I don't want to be encumbered by technical stuff. My plan was to make a bunch of profiles for the various light conditions I encounter: noon daylight, sunset daylight, indoors naturally lit, indoors incandescent light, indoors fluorescent light, outdoors sodium vapor streetlights, etc. I figure I can do this once and have my colors roughly where I want them for all situations.

My question, therefore, is this: is this even a worthwhile pursuit? I suspect color rendition is more than just white balance, but does it vary nearly as much with various light sources? I realize making a profile under one streetlight is unlikely to work precisely for another, but wouldn't it at least be within the ballpark of correct? Or at least more so than using the default profiles?

  • The importance of the Leica lifestyle vs. accurate color is something you have to weigh for yourself. If you want in-the-ballpark results, set the camera to automatic white balance and let it take its best swing at it. – Blrfl Mar 5 '17 at 21:46
2

As often as you need. If you do not need it do not use it.

Instead of trying to make profiles for all, it is better to carry the checker if you need it.

Cases where you need it/ do not need it.

  • You make product shoots and you need the colors of the product be as accurate as possible. / You are making a still life. The image or light just looks good.

  • You are using different cameras, a wedding photography where you have 2 or 3 photographers using different cameras, or a video sequence where again, you need to have consistent colors during the movie. / You are taking casual photos.


It has no sense of making a color profile of "sodium vapor" lights for example, because you will never use that light for a color product shoot. In those cases you want a mood, not to change that yellow light to "white".

In most situations, there is a chance that making a white balance is enough.

If you want to play, go ahead.


Just some notes regarding your comments.

As Michael Clark commented. Color calibration is made after the shoot is made. It is a post process.

1) You take a sample shot of a white target, or your Macbeth color chart but as a reference.

2) Based on that information you make a profile of modifications that will be applied to subsequent images in Lightroom or Photoshop.

What does improve your dynamic range, or better said, optimizes your existing camera's dynamic range is proper exposure.

A light meter gives you a reading of the light and recommends the proper exposition.

But there is a chance your camera has a slight variation on the middle gray for some reason, or has more sensitivity in the brighter zone.

Some high end exposimeters help you make a dynamic range calculation on your camera. The explanation is out of range of this answer, but basically after this series of measures and comparations with your camera images, it saves some adjustments on the light meter itself and the next time you take a reading, gives you a value optimized for your camera.

There are several methods to do that and depends on the light meter, some are a series of bracketed shots and others are using a grayscale target.

And after your exposimeter has this information, next time you take a shot using this values, the shoot itself, therefore the raw data captured, will be with the maximum dynamic range your camera can provide.

  • But doesn't proper calibration improve dynamic range? Consider the example of fluorescent lights: they have a particular color temperature, but they also have a spike in the green portion of the spectrum. Wouldn't a proper calibration compensate for oddities in various light sources' spectra? – Alex Mar 5 '17 at 23:12
  • @Alex Not really, because the WB isn't applied until demosaicing. It has no effect on the sensor's response to different wavelengths of light. – Michael C Mar 6 '17 at 1:05
  • But wouldn't a ColorChecker's super duper standardized color chips yield colors in different places on the gamut? For instance, such a green spike would cause most colors to look more green, but the green chips would be unchanged except for luminance. Wouldn't a profile allow me to adjust out green from all colors except green, which I would darken slightly? Or does the white balance's tint setting compensate for that already? – Alex Mar 6 '17 at 5:50
  • @Alex All of that goes on after the raw data is captured. Raw data is just a set of monochromatic luminance values. The dynamic range is defined by full well capacity of the pixels, not by later processing. For why there are no "true green" or "true blue" or "true red" pixels even with a Bayer masked sensor, please see RAW files store 3 colors per pixel, or only one? – Michael C Mar 6 '17 at 10:00
  • @Michael Clark : Depending on the converter, white balance can be applied before demosaicking, that helps to control maze artifacts. – Iliah Borg Mar 11 '17 at 0:59
2

Here's the thing you need to remember about color calibration: the goal is to make specific things look the same regardless of the type of light they are photographed under. This is very desireable for things such as product shots. It might be very undesirable for some kinds of outdoor photography.

How often do I need to recalibrate using a ColorChecker?

As often as the light changes and you need to insure critical color accuracy of the things you are photographing.

Assuming you are only shooting with one camera the only reason to do calibration profiles for noon light and late afternoon light is if you want your subject shot under both to look identical. Matching the late afternoon light to look like the noon light means you are intentionally processing out the look you get from shooting at the "golden hour."

Calibration profiles can come in handy when shooting under limited spectrum lighting that often requires much more WB adjustment than a simple color temperature shift.

But doesn't proper calibration improve dynamic range?

It can, sort of but not really, because the WB isn't applied until demosaicing. It has no effect on the sensor's response to different wavelengths of light. What it can help with is providing an in-camera preview image whose histogram more accurately reflects the raw data collected by the sensor. As such, it can help avoid blowing out a single channel under wierd, limited spectrum light that might fool the overall "brightness" histogram. But setting the camera to show separate channel RGB histograms can do much the same thing if the WB is set approximately the same as will be used when the raw data is converted. It need not be exactly calibrated.

Some folks claim that a concept called UniWB (Universal WB) will create histograms that match the raw data. What UniWB really does is create histograms where the R, G, and B values are equalized by using a multiplier of 1.0 for all three color channels. But UniWB may still allow all three to be equally inaccurate compared to the actual raw data because gamma correction response curves are still being applied to all three channels to produce the ugly green jpeg preview. When multipliers for the red and blue channels are applied to the raw data later in post processing, particularly for data collected under limited spectrum lighting where at least one multiplier used is much higher than is typical, values that were not blown out in the raw data may, in fact, wind up clipping in the converted image.

  • UniWB is unitary WB, all 'ones' in all 4 channels.I would love to see an example of standard in-camera processing that results in UniWB in-camera histogram indicating no clipping while raw is clipped. – Iliah Borg Mar 11 '17 at 0:51
  • @IliahBorg You won't, but you can see the obverse very easily: The in-camera histogram showing clipping that doesn't exist in the raw data. – Michael C Mar 11 '17 at 1:29
  • The point is not to indicate false clipping. – Iliah Borg Mar 11 '17 at 1:54
  • The point of most people who tout UniWB as a way to get a histogram of the raw data is to fully saturate the highlights without clipping. That is, to expose to the right as far as possible without exceeding full well capacity. – Michael C Mar 11 '17 at 1:58
  • FWC is not used in 90+ percent of modern cameras even at base ISO setting. I suggested UniWB to help with false clipping of red and blue channels (like it happens with roses). Yes, that helps preventing unnecessary underexposure because of histogram of the red channel showing clipping in red. UniWB has no effect on the green channel, except that on some very rare cameras where green WB coeff. goes below 1 for incandescent light. To prevent clipping of the raw in green channels one needs to know how much space is left in highlights after the histogram indicates green clipping. – Iliah Borg Mar 11 '17 at 2:05
0

Something to remember about a color checker: You are calibrating your camera/sensor/lens system to 24 color pigments. As the answers above mention to "calibrate for critical color" and "as the light changes". The hole in those comments is the fact that your images may not contain any of the 24 reflected colors in a color checker. You may in fact be shooting products that have nothing to do with pigments at all. For example: Clothing on figure. or Landscapes. Remember this that your sensor/lens combination does not satisfy the The Luther Criteria, thus cameras spectral responses to light wavelengths are not the same as the human eye. So calibrating on reflected colors provides calibration for those reflected colors. Metamerisem is still rampant and other color issues are uncalibrated. Understanding that may ease your felt need to "calibrate" with a color checker, depending on the product your providing.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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