11
\$\begingroup\$

I've noticed that the subject color seems to affect auto-focus, and am curious if I am just imagining this, or if there is a reason for it.

For instance, I shoot a lot of figure skaters, and synchronized skating teams. The background is constant (a rink, and boards, sometimes the stands). My auto-focus seems to work differently depending on what color dresses the teams are wearing.

Blue, purple, and red seem to give rock solid auto-focus. White or grey (and even sometimes black, which surprises me given the background is white/grey) seem to not always focus correctly.

However, even more surprising, I will sometimes find a team wearing a colour that causes my auto-focus to hunt continuously (I can hear/feel the motor thunking back and forth). This has happened with pastels (light blue in particular), grey, or even black.

I'm using a Nikon D90, and a Sigma 70-200mm F2.8 (though I'm not sure that's relevant).

If there is a color/auto-focus correlation, I'd love to know the reasons why too.

\$\endgroup\$
4
  • 1
    \$\begingroup\$ The problem with black, specifically, is that there is often very little in the way of contrast within the black for the camera to work with. You should be able to get snappy focus on the edge of the costume/uniform against the boards or the ice, but if the AF sensor is directly over the black, the whole thing can seem like a hole in the light. The myriad tiny highlights in spandex don't help at all -- it's a perfect example of "the subject contains many fine details" AF failure problem the manual taks about. \$\endgroup\$
    – user2719
    Jan 2, 2012 at 14:53
  • \$\begingroup\$ Seems like sometimes using the hood can prevent the auto-focus hunting, but hasn't happened often enough yet to confirm. \$\endgroup\$
    – seanmc
    Jan 16, 2012 at 4:13
  • \$\begingroup\$ AF-S or AF-C? Nikon's 3D Tracking actually uses color to follow subjects. \$\endgroup\$
    – Itai
    Jan 16, 2012 at 4:27
  • \$\begingroup\$ In the rink, I typically use AF-C (and 1/320 to 1/500, and try to get some DOF, but often I need to go to F2.8 to F4, so that ISO can remain under 1600 - rinks aren't always very well lit!). \$\endgroup\$
    – seanmc
    Jan 16, 2012 at 4:31

3 Answers 3

5
+50
\$\begingroup\$

Autofocus sensors are essentially complex arrays of light-sensitive strips, and pairs of strips can detect when focus is "out of phase". Some sensors only detect phase shift in one direction, some in two, some in two diagonal directions, each with successively higher sensitivity to contrast differentials and more phase sensitive strip pairs in various arrangements. To my knowledge, AF sensors are simply tonal in nature, and do not factor in color. This is in contrast to metering sensors, which these days on prosumer and professional grade cameras, usually DO factor in color and possibly a variety of other factors. AF points also usually only "see" a fairly small area of the overall image frame, and as they are designed to sense phase shift, targeting an AF point on a surface of relatively uniform color (regardless of what that color may be) is unlikely to produce enough phase shift to lock onto and adjust focus, or if at most start AF hunting, but lack the necessary contrast to lock focus at a certain point.

While AF sensors effectively see in monochrome, and see an entirely different world than image sensors or the human eye sees, they likely have similar color-sensitivity ranges. (If there is any kind of filtration, either deliberate or as a result of the materials used in the AF sensors themselves, they may have unconventional color sensitivities as well.) Most photo-sensitive devices are less sensitive to blue light, and to some degree red light, than to "middle" colors, such as yellow, green, and very light blues. That could introduce less sensitivity to contrast within those color ranges. I wouldn't suspect that such differential in sensitivity would affect AF nearly as much as a lack of contrast at the selected AF point, but it may have an impact in some cases.

The key to AF is focusing when the AF point is very near areas of high contrast. The edge of a dark figure skater against the bright ice will likely lock every time, where as the center of the dark figure skater is likely to result in a bit of AF hunting until it locks onto something, and that something may not actually be what you intended. Using multiple AF points can assist in locking AF somewhere useful, however it broadens the area factored into AF, and its entirely possible you will still lock focus on something you did not intend.

It may also help to understand how the AF system works, as there is a world of variety in AF modules, and they all offer different capabilities. I recently researched the AF capabilities of a lot of cameras while debating whether or not to get a 7D. The D90 has a decently advanced AF system that has the ability to predict and track subject motion while focusing. If I remember correctly, its motion detection and tracking work differently depending on whether you are in a single-shot or continuous-servo AF mode. In single-shot mode, I do not believe the D90 will actually lock focus on a moving subject, and will wait until the subject has stopped moving to actually lock focus. In servo mode, the D90 will track the subject that it believes is in focus, but as far as I can remember, there is no actual focus lock since the camera is continually adjusting focus so long as the shutter button is at least half-way down. If you are using single-shot AF mode, you might not be getting AF lock because your subjects are moving. You might want to try switching to continuous-servo mode and see if you have more success with getting in-focus shots. I would offer that this is much more likely to be your issue than the color of your subjects. The D90's AF system is quite configurable, and I believe there are ways to "tune" it (for lack of a better word). You might want to hit the D90 manual or do some research on how to control the D90 AF, and see if you can get it working better for the way you use it.

For reference, here is the article I read on D90 AF (the article states D200, however the AF module is the exact same one):

Understanding the Nikon Multi Cam 1000 AF Module

\$\endgroup\$
3
  • \$\begingroup\$ Very interesting article. Doesn't exactly map to the D90 because the menu options are quite different. \$\endgroup\$
    – seanmc
    Jan 20, 2012 at 15:26
  • \$\begingroup\$ I did find this: "Lens movement, especially with long lenses, can be interpreted by the camera as subject movement. Predictive Focus Tracking in this case is tracking your camera movement while simultaneously trying to track your subject. Attempting to handhold a long lens will drive your camera NUTS (as it will you) when you later view the shaky pictures. Use a vibration reduction (VR) lens or a tripod for best results." - so I wonder if that is related to my "hunting" problem? Unfortunately, my 70-200 is non-VR. \$\endgroup\$
    – seanmc
    Jan 20, 2012 at 15:28
  • \$\begingroup\$ @seanmc: It might indeed be. My only telephoto lenses are IS lenses, and I always use IS, so I've never encountered any issues with too much camera shake. I highly, highly recommend IS/VR if you can afford it...its one of the most amazing features of lenses these days. \$\endgroup\$
    – jrista
    Feb 12, 2012 at 1:14
5
\$\begingroup\$

Some color combinations will have greater contrast than others. The auto-focus is generally hunting for the best contrast. Black against white should have the most. There would be less between colors that differ only in one primary (example: cyan vs. green). Some cameras may auto-focus only in a part of the spectrum or be affected differently at different colors (for example blue being sensed weaker, have more flare, or not sharing the same focal plane if additional lensing is involved in auto-focus). I'd expect the D90 to be better than most, so it might be just a matter of subject color. I'd suggest a visit to an art supply store and pick up a box of construction paper in mixed colors. Try different sheet color pairs to see which focus faster or slower.

\$\endgroup\$
0
\$\begingroup\$

a canon rep once told me a 5d mk ii focusses in the red channel only, while the 7d adds the green channel to it.

not sure if that is accurate though. But any sensor can only capture a certain wavelength range - no matter if af or actual "photo sensor", hence the beyer array on the latter.

\$\endgroup\$
2
  • \$\begingroup\$ Thats interesting, about the 5D II only using red light. As far as I know, the 5D uses the same 9-point low-end AF as my 450D, and the 450D had a TERRIBLE time focusing in dimmer light (i.e. just after sunset, where the only red light is in the sky.) I wonder if its because of the limited AF bandwidth... \$\endgroup\$
    – jrista
    Jan 16, 2012 at 18:36
  • 2
    \$\begingroup\$ The "any sensor" part is misleading - the range of wavelengths an image sensor can capture is limited, but it's wider than visible range; both IR cutting filter and Bayer layer are added on top of sensels to remove the frequencies that could be captured but are considered undesirable for that sensel (e.g. you have to throw away other colors in order to measure how much green there is). There are companies (like MaxMax) that remove those filters to turn cameras monochrome and/or IR-capable. \$\endgroup\$
    – Imre
    Jan 18, 2012 at 8:51

Your Answer

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

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