A Little Color Reference for Photography




Introduction: A Little Color Reference for Photography

About: I'm an Electrical and Computer Engineering Professor at the University of Kentucky. I'm probably best known for things I've done involving Linux PC cluster supercomputing; I built the world's first back in Fe…

Get better color accuracy in every photo using this little device.

The trick to accurate color capture is to always have and use a color reference each time you start shooting in different lighting... but carrying a color reference card big enough to focus on is a pain. As I disclosed back in 2003 (more information at http://aggregate.org/DIT/LENSCAP/ ), the color reference need not be in focus to work -- so you can use a tiny, easy-to-carry, reference. For example, got lenscap?

It should be noted that technically accurate color imaging is very difficult. Photons can have any of a wide range of wavelengths, but most sensors just measure approximately how many photons hit each pixel after passing through the corresponding Red, Green, or Blue filter. Further, human vision employs a mechanism of color consistency which makes the relationship between spectral properties and apparent color quite complex. In short, some very nonlinear things can happen in these transformations, the discussion of which is well beyond this Instructable.

You also should keep in mind that color accuracy doesn't necessarily make good art. People like sunsets, and many other things, in less accurate but more intense "Kodachrome" colors.

Step 1: Making Your Reference

Serious photographers often carry a gray card or a color checker chart with them. Lots of people argue about things like what reflectance the gray card should have (anywhere from 12% to 18%), but it doesn't matter much unless you're using it for exposure metering, which we aren't. All you need is one or more known colors that you can use for calibrating color correction, and it is easiest if one of them is neutral. Gray works better than white because digital cameras can clip color channels when overexposed, and gray works better than black because sensor noise is less significant. So, make yourself something gray that you'll always carry with your camera.

Although cutting-up a color reference card works great, those cards are expensive. Many laser printers produce very neutral blacks which, when printed on a neutral white paper, produce an acceptable reference. For example, just scale the circle shown below to fit within your lenscap, laser print it, trim it, and tape or glue it inside your lenscap.

You can fit at least 4 color reference patches inside a typical lenscap before the area of each color becomes too small to be effective. The colors can be anything, but avoid colors that, as seen by your unaided eyes, appear to change dramatically when you move the reference to different lighting (i.e., try to avoid metamerism). Colors matching your printer's inks are probably not a bad choice. More reference colors allow fancier non-linear corrections of color, whereas a single gray reference really only allows correction of the average color temperature.

No space on the lenscap? Ok. How about a little paint on something? Or perhaps a piece of cloth or duct tape on your camera bag or strap? I used to have a neutral gray pair of sneakers....

Step 2: Using Your Reference

There are two distinctly different ways to use the color reference. The first method works only for digital cameras that allow you to set a manual color balance by sampling a scene:

As shown in the figure, simply hold the neutral gray reference in the lighting that is hitting your subjects and sample it to set the "manual white balance." That's it. Jpeg images straight from the camera should be reasonably well balanced until/unless the lighting changes. Of course, don't forget to change the white balance when you use the camera in different lighting (some cameras retain manual white settings across being turned off and on again). Incidentally, you can also use this method to digitally apply any color filter you wish -- simply use a gray card that is tinted the opposite color (e.g., blue tint reference to produce orange filter effect).

The second technique works with any type of camera (even film) and is potentially much more precise, but is more complex and requires post-processing of the images:

If you are using a digital camera, set the white balance to "daylight" (or whichever fixed setting seems closest to correct in the current lighting -- but not "auto") or shoot in "raw" mode. Raw images usually are not affected by the white balance settings in the camera and can yield higher image quality. For a film camera, stick to the same film type, chemical processing, and scanning process.

1. Take a separate photograph of the reference -- possibly out of focus, but filling most of the frame.
2. Shoot as many photos as you like under the same lighting conditions used for the reference.
3. Once you've downloaded the images to your computer, use your favorite image editing program (e.g., photoshop, gimp, cinepaint) to tweak the color balance so that the reference image colors closely match the color values you know they should have. Keep in mind that the RGB values don't have to match, because the reference colors can be slightly under or over exposed. In a color space like LAB or YUV, the L (lightness) or Y (luma, brightness) component might not match.
4. That same set of adjustments is then applied to correct all the photos you shot under the same lighting conditions.

If you made your own color reference, you might not know precisely what the digital values of your reference colors should be for step 3. If so, you can get a usable approximation by taking a photo of them under nice, standard, "daylight" conditions and matching to that. Most cameras and films deliver pretty accurate (or at least pleasing) colors under standard daylight conditions.

Be the First to Share


    • Tinkercad to Fusion 360 Challenge

      Tinkercad to Fusion 360 Challenge
    • Colors of the Rainbow Contest

      Colors of the Rainbow Contest
    • Remote Control Contest

      Remote Control Contest



    7 years ago on Step 2

    Geetings! I'm well aware of metamerism, and that is the primary issue -- avoiding it is discussed in step 1. The paper white-point isn't mentioned and should be. Back when I first did this, there were still papers (I usually used Bristol board) that were quite good whites, but since then the world has gone nuts with brighteners....

    To make your own good reference (as opposed to using part of a commercial one), there is some trial and error involved. I haven't had much difficulty getting pretty good gray references within a few tries; in fact, I've even found that well-chosen gray PLA plastic can be close enough to be useful. As long as metamerism is minimal, you can correct for (known) minor color biases in postprocessing. That said, to make your own accurate reference for lots of colors is very hard, arguably not worth even trying; I use a mini ColorChecker for that.

    In any case, the main point of this Instructable (and my 2003 disclosure) was that you don't need a color reference that is big enough to focus on -- a tiny reference photographed out-of-focus works fine, so there's no excuse for not having one with you. Using a little piece of a commercial gray card should ensure accuracy if available printers and papers don't allow you to make a good enough one from scratch, and you cut, mount, and use either reference card the same way.


    7 years ago on Step 2

    Greetings Professor! Thanks for putting this ible up and contributing to the community. While the idea behind the workflow is sound, the actual practice of making the target is shaky. There are two things working against us in this process; metamerisms and white-point.

    Metamerism occurs when an object does not reflect wavelengths of light equally. This is always an issue with inkjet printing. Inks and pigments for inkjet/giclee are manufactured to appear neutral under 5000k light-sources, but will unevenly drift out of balance as color temperatures deviate away from this "daylight" norm. This is to say that the target may perform properly under 5000k illumination, but will fail under other sources such as halogen, incandescent, open shade, mercury vapor, etc. The end result is a reflective target that exhibits non-linear variations in spectral power. In layman's terms; the user's color balance will be inaccurate with differing levels of error depending on the light-source used.

    Now let's address the white-point concern. Printer papers deviate wildly from blue cast to warm casts such as yellow, and often contain optical brighteners that fluoresce, and/or additional bluing agents. These additives also contribute to metamerism. It is extremely rare that "white" printer paper is actually a neutral white. Since this non-neutral/non-linear paper is a significant contributor to the reflectance of your target, additional color temperature errors are introduced.

    Good targets, and I would like to put emphasis on "good", are manufactured to be as close to free from metamerism as technological and economical conditions will allow. I realize that this goes against the spirit of DIY here, however, I feel it important to address this potential issue for anyone who may be depending on, or merely hopeful for, critical white balance.