4
\$\begingroup\$

I'm about to create custom profiles for my camera and in order to do so I need to shoot pictures of color reference card.

I don't really understand the impact of the lens I choose to create the reference pictures. I can imagine that if my lenses are from the same brand and generation I should have accurate results. But what if I take some reference pictures using, say, old lenses with yellow radioactive coatings and other ones with modern coatings?

Will a rigorous white balance adjustment (using a neutral grey reference card for instance) be enough to get perfectly accurate results, or should I create custom profiles for each lens + sensor combination because white balance adjustment won't be able to correct color transmission variations?

\$\endgroup\$

3 Answers 3

4
\$\begingroup\$

Short answer: Yes, you need to create separate profiles for each camera/lens combination. Unless the light in your test passes through the same lens, the system has no way of knowing what characteristics for which to correct. Applying a profile created using one lens and then expecting it to properly correct a significantly different lens would be like putting a colored gel in the optical path and expecting the color profile you created without the gel to to somehow magically remove the changes the gel creates!

Long answer: There's no such thing as "perfect" white balance. Color management can only go so far in reproducing on a display medium what exists in the real world. Different materials that look the same color under one type of light will appear to be slightly different colors under different lighting conditions. And even if there were such a thing as perfect color correction, where are the perfect eyes that could see it? Everyone perceives color in their own unique way!

For all practical purposes if your lenses are very similar in terms of color transmission you won't notice a difference when changing lenses. Other factors will likely introduce more variation than that between two of your lenses. But if you wish to get as close to "perfect" as possible, then you need to go a step further.

And, having said that, the best way to create custom profiles is to use the same lens with the camera that you intend to use when that profile is applied. It is also important to shoot under the same light (or as close as is practical) when you create your reference files as what you plan to shoot under when you shoot the photos to which your profiles will be applied. The more different the light is between your profile shots and the shots to which you will apply those profiles, the less accurate your results will be.

And if you are truly interested in color management to the degree your question indicates, shoot more than just a grey card. Use a colorchecker chart such as the Xrite ColorChecker Passport

\$\endgroup\$
3
  • \$\begingroup\$ The phenomenon of colors appearing different under different light is called "metamerism". We're going to see more of that now that incandescent lights are being phased out. \$\endgroup\$ Commented Nov 3, 2015 at 18:06
  • 1
    \$\begingroup\$ @Mark Ransom: Actually, the enemy is not metamerism per se, but metameric failure - big difference. \$\endgroup\$
    – Iliah Borg
    Commented Nov 3, 2015 at 23:17
  • \$\begingroup\$ Thank you for your response and for the advices. Someone is offering me to build a custom profile for the camera using a checker with a lot more colors that the Xrite's one (+ spectrometer, ...) so I'll try that. But I'll definitely try the Xrite one too as it can be very convenient to build custom profiles for different locations / lightning scenarios and see how the results can differ. \$\endgroup\$
    – thomas.g
    Commented Nov 5, 2015 at 11:12
3
\$\begingroup\$

Any through-the-lens methods for setting white balance will take the lens glass/coating/filters into effect, such as:

  • "Auto white balance" in-camera or in post.

  • Measuring with a grey card and setting custom white balance from it in-camera or in post.

  • Custom correction judging with your eye in post.

Any other methods for setting white balance will not, such as:

  • Using white balance presets in the camera or RAW processing software.

  • Measuring light colour with some other measuring device external to the camera.

Very few lenses themselves would impart a colour cast that ought to be a problem when viewed with the naked eye. However, when it comes to lens filters this becomes more relevant - for instance tinted or graduated filters definitely impost a colour cast, and even things like ND filters can do so a little.

Will a rigorous white balance adjustment (using a neutral grey reference card for instance) be enough to get perfectly accurate results, or should I create custom profiles for each lens + sensor combination because white balance adjustment won't be able to correct color transmission variations?

If you're going to try and come up with white balance presets in advance to cover all your intended uses, you'll need different profiles for every light source, time of day, setting, pose, weather condition, and so forth. All these factors will probably affect colour cast more than the lens. In the studio you can control the light somewhat but even things such as the backing will impose a colour cast to an extent.

\$\endgroup\$
1
  • \$\begingroup\$ Thank you for your answer. I'll try to create a profile under controlled light conditions so I can visualize the impacts of changing lenses. I suspect as you said that they won't be significant to the eye if I don't use esoteric lenses but I'll confirm the theory with practice :-) \$\endgroup\$
    – thomas.g
    Commented Nov 5, 2015 at 10:00
2
\$\begingroup\$

The first question you ask yourself is what accuracy do you want. Normally you want to get into 6 to 2 deltaE range of accuracy, which does not stress the system too much.

Splitting hairs, available white balance adjustments are linear by nature, while lens spectral transmission is not. However, any dust on a lens, or any so-called protection filter you are going to use will have significant effect on the profile accuracy if the profile is calculated without those (same amount of dust, imagine that). More, most probably you won't be able to take an accurate "studio-grade" shot of a colour target, simply because of flare and white balance irregularities across the target. Expect to see either crashed blacks or loss of contrast in shadows, and non-linear white balance in shadows - if you are going to inspect the results close enough, or, better to say, too close. Different amount of flare on a shot of a target and a shot of a landscape is a given, an is also a factor that has much more influence on the quality of the results than differences in lens transmission after white balance.

Yellow glass tend to have voids in spectral transmission, and that is a whole different story. With such glass you may occasionally get odd renditions of some colours, no matter how good is the profile.

The most important factor to deal with is flare when shooting targets for profiling. Alternatively, if you have the budget, you can profile using spectral response; you will need a monochromator and a light sphere to do this. Better if you use a spectrophotometer, too. In this case you can measure the components separately, and multiply the spectral data for the sensor, lens, and any filter you will be using, calculating ad hock profiles.

\$\endgroup\$
1
  • \$\begingroup\$ Thank you for your answer. Indeed I suspect non-linearity of the lens spectral transmission but I have no idea how much it will influence the image after white balance adjustment. As you said I suspect yellow glasses to have a stronger impact ! Thanks for the advice about the flare control. \$\endgroup\$
    – thomas.g
    Commented Nov 5, 2015 at 10:03

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.