Colour Accuracy

The human eye sees an object as the same color irrespective of the angle and lighting you see it under. Your camera can’t do this because it can’t see; it blindly measures.

I talked about accurate colour and its dependency on selecting accurate camera profiles in a previous article.  I now look at the next logical step; creating my own camera profiles tailored to my camera/lens and the lighting conditions of each particular photo shoot.

The problem with camera profiles.

Using a camera profile for your camera is a step in the right direction, but as I have found since writing my previous blog on the subject, there are other issues to consider

  • Each lens has different effects on colour. My Tamron 90 macro records greater vibrancy than my Sony 18-250 super-zoom. A manufacturer’s (or Lightroom) camera profile ignores the effects of the attached lens.
  • The camera profile is created under controlled lighting conditions. You may get similar lighting in a studio, but not elsewhere.

The magic of perfect colour

Before looking at my solution, it would be instructive to show the results.

Take a look at the images below.

Mahonia Leaf
Mahonia Leaf
Red Robin leaf
Red Robin leaf
Elaeagnus leaf
Elaeagnus leaf

At the end of the photoshoot for this sequence of images, I took a photograph that would later allow me to calibrate for colour. This calibration ensures that

  • I can correct for colour inaccuracy of my camera body and the attached lens.
  • I can correct for ambient light colour.

The photos above were not tweaked for colour to make them pleasing to the eye; they were calibrated to be true to life.

On to the solution to my colour accuracy problems; I created my own camera profiles via X-Rite Color Checker Passport.

X-Rite ColorChecker Passport

All camera profiles are created with reference to a colour swatch. X-rite ColorChecker Passport (CCP) has such a swatch (albeit with a smaller range of colours than the one Adobe use). To complete the solution, CCP comes with Lightroom/Photoshop plug-ins that automatically create a camera profile from any photograph containing the swatch. Camera profile creation with CCP is easy and borders on trivial. In Lightroom you just right-click on a photo containing the CCP swatch and select Export > ColorChecker. The resulting camera profile will be saved to your Adobe profiles folder, meaning that it will now appear in Adobe Camera RAW and Lightroom.

Let’s stop to think what this allows.

Supposing I took a shot of an apple; I can create a camera profile for both my camera/lens combination and lighting conditions. I do this  by taking an additional photo that includes the CCP swatch.  This photo is used to create a bespoke camera profile. If I now apply this camera profile to my apple photograph, my apple is now colour calibrated to the CCP swatch.

Now supposing I then give you the apple, telling you to take the same shot.  Irrespective of your camera/lens or lighting conditions, once you apply your camera profile (which you create in exactly the same way I created mine), your apple will be the exact same colour as mine. The apple in your photo will be the same colour as mine, because both are calibrated to the same swatch.

It doesn’t matter if I used a Sony camera and you used a Nikon, Canon or Pentax. It doesn’t matter that I shot the photo outside on an overcast day and yours was sunny.

So what actually is the X-Rite ColorChecker Passport?

The CCP is a passport sized plastic case. Unlike the more traditional single sheet swatch, the CCP is both sturdy enough to be carried out in the field, and small enough to fit easily in a camera bag side pocket, or your own coat pocket. The CCP also comes with a lanyard (not shown), so you can even just hang it around your neck.

X-Rite Color Checker Passport
X-Rite Color Checker Passport

There are 4 plastic ‘pages’ inside the CCP. The first two pages look like this;

X-Rite Color Checker Passport: pages 1, 2.
X-Rite Color Checker Passport: pages 1, 2.

The first page (top) contains some custom white points, and clicking them allows you to change the warmth of portrait or landscape shots. Nice, but it’s the second page that contains the beef; this is the swatch that the plug-ins look for in your photos to create the all-important camera profiles.

The plug-in can always extract the swatch from the photo as long as the 5 registration points (the 4  corner brackets and the centre ‘+’) are in focus and visible. Even if they are not, all is not lost; you manually place the position of the 4 corner points. The image below shows how this works. The plug-in cannot create a profile from this photo because it is blurred and the registration points are indistinct. The swatch extraction is performed by manually dragging the four green circles, ensuring that the green squares are each on a colour brick.

Manual Swatch detection
Manual Swatch detection

Page 3 contains a grey card. This is essentially a larger version of the grey brick to the right of white on page 2.

X-Rite Color Checker; page 3.
X-Rite Color Checker; page 3.

The final page contains a quality guarantee. Worth noting is that the standards used to calibrate the colour swatches are traceable to agreed standards (via NIST; the National Institute of Standards and Technology), so in the apple example above, your apple colour will look like mine because both your CCP and mine will be calibrated to the same standards. It is worth noting that this is not the case for many other calibration systems (particularly electronic monitor calibration devices, more on this later).

X-Rite Color Checker; page 4.
X-Rite Color Checker; page 4.

Examples and discussion

Below are two photographs I took as part of the work for this blog-post.

Image 1
Image 1
Image 2
Image 2

They were taken less than 2 minutes apart under full sun using the same camera, but the wooden table and the Sony Alpha symbol (on the lens cap) in each are different colours. This is because of the change of angle in the scene. Notice how easy it is to see variance in colour between photos even when taken a few seconds apart. In this case the variance is caused by a slight camera movement!

Before discovering colour calibration with CCP, I would have thought ‘oh, it’s caused by reflected light; the change in angle in phto 2 means more light is reflected and therefore there’s more light in the scene’, and the fix is simply to reduce exposure in post processing for image 2 to give us something like image 3;

Image 3
Image 3

Image 3 goes some way to fixing the Alpha symbol’s colour, but not the wood. The wood colour is more complex (the wood is not a solid tone and has a gloss surface) and so does not match Image 1 (the browns have become more orange) and nothing will fix it. You simply can’t make a series of objects look the same colour if they were taken under different conditions unless you explicitly calibrate all your photos to a common colour reference (such as the CCP swatch).

By calibrating using CCP once for the straight down view and again for the side on view, I get consistent colour despite the change in lighting conditions. I still have to change exposure for the second photo , but this time (image 4) both the Alpha symbol and the wooden table look the same colour as in image 1;

Image 4
Image 4

Again, think about what we have achieved here. The human eye sees an object as the same colour irrespective of the angle and lighting you see it under. Your camera can’t do this because it can’t see;  it blindly measures.  We have fixed this seemingly intractable and complex philosophical problem via a simple colour calibration!

Let’s get back from the high end philosophical and down to earth with a stupid question. Does it matter that the wood is wrong in image 3? Put another way image 3 doesn’t look wrong as a standalone photo. If the photo doesn’t look odd, does it matter that the colour might be a bit out? Here’s why it can really matter;

  • If you are shooting photographs with a main and backup camera, both will give different colour. You cannot be sure which (or even whether either) is right. By calibrating for colour separately from the camera hardware and optics, you get consistent colour irrespective of which camera or lens combination you use; all photos in your shoot will be consistent.
  • Colour calibration is vitally important when compositing in Photoshop. Taking a figure or sky from one photo and trying to composite it into another is easy to do in Photoshop, but hard to do realistically because two unrelated and uncalibrated photos will have different colour casts. By using CCP, you remove all colour casts, and the only issue you have to fix is relative exposure. Exposure matching is easy (you have only one slider to match); matching colour is much harder. It involves three sliders (and more likely 3 separate tone curves if you want to do it properly), and that’s assuming the simplest case where you are in RGB colour.
  • Camera manufacturors may tweak colour to make photographs look perceptually better, or more controversially, to hide high ISO noise. This makes the camera innacurate in non-linear ways. A camera designed to give perceptually better skin tones will be innacurate for background landscape tones in the same shot. Colour calibration corrects these marketing induced kinks.

Another stupid question; should you always use CCP? No, there are some exceptions;

  • Some colour casts are useful; you may want to capture the ambient light colour via its cast. A sunset is a perfect example of this; you want to catch all that gold tinted light, and what it does to the environment! Put simply, colour correction can correct out ambience and atmosphere. There’s always got to be some thought into whether colour correction is appropriate. Even so, I now always take the CCP out with me and always take a photo of the swatch if there is time. This gives me the ability to remove colour casts if I ever need to include the photos into a Photoshop composite image during post processing.
  • Some lenses give more vibrant rendition than the actual scene. You may not want to colour correct such a lens as the vibrancy may be desirable
  • The uncalibrated scene simply looks better. The ‘incorrect’ wood colour is actually warmer than the calibrated one, and the CCP shows up better in the brighter but ‘incorrect’ photo. Depending on your reasons for shooting the photo, the warmer, brighter version might be the keeper.

Taking a calibration for every set of shots is usually overkill though, and there is a third way to use the CCP; you create a separate camera profile for each camera/lens combination. It’s not as accurate as making a profile per photo sequence, but it way more accurate than using Adobe standard, and also better than using a single Adobe/manufacturer profile for the camera body only. See note 3 at the end of this post for more information.

Workflow

Taking all the above into consideration, I now use the following workflow;

  1. I always shoot RAW, and always explicitly select a camera profile when converting RAW to TIF or JPEG. I no longer use Adobe standard because I find it particularly inaccurate for my camera make (Sony).
  2. If I have time, or colour is very important I take a shot of the CCP per photo sequence and use the generated profile in post processing. For me, colour is generally only ‘very important’ if the photos may be used later in a Photoshop composite image. Doing this allows me, for example, to composite a sky and terrain, both taken on two different days, but both calibrated to the same swatch.
  3. If I don’t have much time (e.g. I am on ‘walkabout’), I take two shots of the CCP (one facing the sun, one facing away) for my main lens of the day, and use the 2 generated profiles for all photos taken on that day’s walkabout.
  4. I have created a profile for each combination of my Sony a500 camera and 4 most used lenses. For each combination I have three profiles; inside with incandescent lighting, outside sunny, and outside cloudy.
  5. If I haven’t taken a photo of the CCP for a shoot via (2)  or (3) above, I use one of the profiles from (4).
  6. Colour is never important for certain photos. The main candidate here is a true HDR; colour is rarely conserved in HDR, so I don’t bother calibrating for it. NB – there are ways to minimise colour inaccuracy when using HDR, see this previous post.

Using the ColourChecker the way it is intended (to give calibrated colour output from your camera) is only part of the work flow. One of the coolest things about using a calibration swatch is how easy it is to either calibrate other parts of the work flow, or measure their inconsistencies.

The Camera LCD.

Most people consider three colour accuracy areas in the digital photography work flow; camera image file accuracy, monitor accuracy and printer rendition accuracy. We forget that the camera LCD plays an important role, at least for the enthusiast or non-studio based photographer. Of course, you can’t calibrate a camera LCD, but you get a very good impression of your LCD’s strengths and weaknesses simply by pointing your camera at the swatch and comparing the image on your camera screen with the actual swatch. Since using my X-Rite ColorChecker Passport in this way, I know that my Sony A500 LCD screen:

  • Brightens blues slightly
  • Desaturates skin tones a little
  • Is otherwise extremely accurate, and certainly more accurate than an uncalibrated computer monitor.

The monitor screen and printer.

There are several hardware monitor calibration devices out there, and Both Mac and Windows 7 users can use colour profile accessories provided by their operating systems. I don’t trust hardware monitor calibration because (a) certain combinations of calibration hardware and monitor just don’t work, and (b) most are not calibrated to any standard; a Spyder and Huey will give different results on the same monitor!

A colour calibrated swatch such as the CCP is better because it is calibrated to a standard. To calibrate my monitor, I set up the monitor with the manufacturers drivers (and never use the OS default plug and play monitor). That gets me pretty close to calibrated, and for final tweaks, I visually eyeball the physical CCP swatch in front of me with an image of the same thing on the monitor, and set RGB values so all the swatch bricks match. All that remains is to change the monitor colour temperature (9300k, 7500k, 6500k) depending on ambient lighting.

Finally, for the printer, I use a print bureau who calibrate their printers to a standard; I don’t think consumer level printer devices are up to it, and leave print calibration to the experts.

Conclusion

The X-Rite ColorChecker (CCP) is now my one-stop solution for colour accuracy. It allows me to;

  • be sure of colour accuracy (or on the inverse, have an idea of how inaccurate my methods will be) irrespective of my camera/computer equipment and lighting conditions
  • to remove colour casts when using multiple photos in a Photoshop composite.

A CCP is quick and easy to use. All I have to do is place the CCP in the scene at the end of each set of shots, then export the same photo to the ColourChecker plug-in to generate a bespoke camera calibration for that set of shots.

Although the CCP is not intended to calibrate them, I find it invaluable in knowing how accurate (or otherwise) a camera LCD is, and I also use it to calibrate my monitor’s colour rendition by eye.

Notes

  1. For more information and videos on the X-Rite ColorChecker, take a look at the X-Rite ColorChecker Passport Blog. The X-Rite product page is here.
  2. The leaf photos in the opening section of this post were corrected (in Lightroom) for colour via a camera profile and had sharpening and slight contrast enhancement applied. The background (a piece of white paper) was then overexposed via the adjustment brush.
  3. The X-Rite ColorChecker Passport is a 24 colour swatch of the type Adobe recommends for use in creating a general camera profile. See in particular tutorial 5 on this link.
Advertisements

5 thoughts on “Colour Accuracy”

  1. Hi,

    I’ve read this article with great interest since i also have a Sony A500 and am very dissatified with the color rendering in Adobe Lightroom.
    For the moment I don’t have the cash for a colorchecker passport, so I would like to ask you if you would be willing to share the profiles you created with me. I understand they would not be 100% accurate for me because of the slight discrepancies between each sensor and also the fact that we probably don’t own the same lenses, but they would probably be an improvement over my current situation. FYI, my current lenses are the Tamron 90 macro, kit 18-55 and the Tamron 55-200.

    Thanks in advance,
    Christian

  2. Hi,

    Yeah, stuff like that always happens on reinstalls and upgrades, good backups are the only way to go in my experience. Good luck with the transfer and I’m looking forward to when you can get them online.
    I have tried the ones at nibdata but they tend to get way to unnatural for my taste (landscape for instance has a purplish tint in the sky and green is way off).

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s