TKP BLOG

The ability to fine tune your camera’s AF system for each specific lens is a great new feature on the D300. This function allows a user to dial-in a level of AF precision that was previously not available by correcting for minor front-focusing or back-focusing issues in a given body-lens combination.

From my very first attempts at bird photography with the new D300 I was somewhat stymied by the results. With my 300 f/4 AFS and 1.4x teleconverter attached I was having difficulties coming up with images that were as critically sharp as those I was accustomed to with my old D70.

Initially I presumed that the additional resolving power of the 12mp sensor was revealing inadequacies with my technique, or limitations of the teleconverter. Upon closer inspection, however, I began to notice that most of the images usually had at least some detail that was incredibly sharp, which wouldn’t be the case if lens vibration brought about by poor technique was the culprit. Upon even closer review, I began to notice that the field of focus seemed to be slightly in front of the intended subject. In particular, some of birds had great detail in the breast and wings, but a slight degree of softness in the eye (the intended point of focus). Since the wing and chest of a broadside songbird, for example, are actually somewhat closer to the lens than the eye, I suspected that my new camera might have a slight front-focus problem.

To test my new theory I set staggered three products with UPC codes on a table, with the first slightly in front and the third slightly behind. With the camera locked down on a tripod I fired a series of test shots focusing on the UPC codes of each subject. I was expecting to find that, when focused on the middle bottle, the foreground bottle would be detectably sharper (thus confirming my front-focus hypothesis), but this didn’t seem to be the case. I played around a bit with the AF Fine Tune menu, and confirmed that I was able to shift the focus around, if only at extreme values, but I didn’t find any consistent front-focus problem. I shrugged my shoulders and attributed my front-focusing thoughts to superstition. It turns out I probably shouldn’t have.

A low-res jpeg of one of the initial nail polish tests, showing the basic setup. The test was very crude, done in a dark cabin (ISO 1600) with to wide of spacing between foreground and background subjects.

After a few more wildlife sessions, including shots in very controlled circumstances (shot from a blind, subjects close enough not to present a challeng to AF, camera mounted on tripod, respectable shutter speeds etc.), I was still producing images that looked to be too soft and often could be interpreted as front focused. Granted, some were also coming up on my monitor looking tack sharp, but the percentage was less that I thought it should be (one can imagine how an AF ‘miss’ can lead to a ‘perfectly sharp’ image when shooting in AF-C and at 6.5 fps, right?).

Using continuous autofocus (AF-C) and 9-point dynamic mode (focus tracking) my hit rate (for perfect focus, as opposed to ‘not too bad’ focus) was less than 25%. Again, a number of perfectly broadside bird images (ducks and shorbirds) seemed to have tack-sharp focus on the side of the bird, with a touch of softness to the eye. Was the 9 point dynamic AF grabbing the part of the bird closest to the lens? The focus point was on the head/eye whenever possible, but it is certainly possible that the moving subject would move the active focus reticule to one of the eight surrounding points. The number of variables in this equations was starting to frustrate me. Again, I shrugged my shoulders in frustration and wondered what the heck was going on. I was beginning to suspect that I would have to send in my 300f/4 and TC for adjustment, but mostly I was grasping at straws and hoping for some moment of clairvoyance.

The other day, while the back of my brain was musing about the focus issue, I was browsing online and came across some discussions of how to properly dial in the AF Fine Tuning. In so doing, I began to realize that my initial test was much to crude to reveal any real front-focus or back-focus issues, especially when long telephotos with narrow DOF are concerned. Unfortunately, most of the test-charts that are widely available and discussed heavily online seem to have one fatal flaw (to my mind at least): they are too susceptible to setup error.

Most of the test charts I saw online were designed to detect back-focus or front-focus using a printed sheet of paper aligned at a 45 degree angle to the camera, so that one can discern how many lines above and below the intended point of focus are within the DOF. Small changes in the angle of this sheet (be it forward vs. backwards tilt, or side to side tilt, or even worse - both) can make it difficult to interpret the results. I came to realize one thing: these tests were designed to detect front or back focus on camera systems where the user could not adjustment the actual plane of focus him or herself. Logically then, in a system where one can directly manipulate the amount of front- or back-focusing it should be possible to devise a much simpler test to discern which settings produce the sharpest image possible. The logic here is that if your AF locks-on but the image is soft the actual plane of focus is somewhat different than where the AF system interprets it to be. By focusing on a flat subject positioned parallel to the lens element (rather than at an angle) and taking sample images along the entire continuum of AF Fine tune adjustments, you can find out whether your system is focusing in front of or behind the actual focus point. If the images produced in the entire series run the gamut from soft to crisp and back to soft again as you vary the AF Fine Tune ajdustments, then you should be able to find the exact AF Fine Tune ’sweet spot’ that produces the sharpest image. Confused? Hopefully an example will help.

I devised a test using nothing fancier than a 20 dollar bill, which I taped to a huge volume of hardcover books (The Complete Far Side, if you must know), and positioned on a table at tripod level. I mounted my 300f/4 and locked it down, attached my cable release, and set my D300 to Mirror Lock Up mode. I carefully focused on a part of the bill with minute detail (in the case, Queen Elisabeth’s right eye). Activating the AF Fine Tune function and manually setting the amount of fine tuning I shot a series of images, starting at -20 and moving up in increments of 5 (e.g., -20, -15, -10, -5, 0, +5, +10, +15, +20). After each photograph I manually moved the lens focus ring, so that I would force the AF system to reacquire. To ensure roughly equal exposure (variable cloud cover was changing the lighting in the room a fair bit) I used my flash in balanced fill flash mode, with the better-beamer attached.

Taking the images from the CF card and opening them in Capture NX, I was able to confirm that the camera had registered the appropriate AF Fine Tune setting using the metadat in the “Camera Settings” tab (CS3 does not have the ability to report on this feature, which is precisely why I used NX). Next I opened all of the images from a series of tests (or a subset that looked the sharpest on the camera monitor) and zoomed each in to 200% (this is serious pixel-peeping work). The differences in acuity through the range of AF Fine Tune settings was astounding.

The images below show some of the initial results. To make a long story short, my 300 f/4 needed to be pushed back by at least 10 increments (so yes indeed, it was front focusing before the adjustment!). When combined with the 1.4tc the results were even worse - this combo looked sharpest between +15 and +20.

Screen capture showing 300 f/4 initial results; AF Fine tune values are inserted in lower-right of each image.

Screen capture showing 300 f/4 + 1.4 TC E initial results. AF Fine Tune vales are inserted in lower-right of each image.

Now that the rough results were in (and supported my front-focus hypothesis) I went back to the tripod and ran another set of tests, using narrower increments of adjustment to further refine the sweet-spot. I wont show the detailed results but will say that the 300 f/4 alone took +11, and with the TC attached required +13. My most accurate lens on this body was my 105VR macro, which required only +3. My Tamron 90 took +7, and my 50 1.8 took +12.  So: it looks like my D300 consistently front-focuses.  I suppose I could send it in for an adjustment, but for now I’ll see where the fine-tuning gets me.  Next up: the zooms…

Cheers,

Todd

This entry was posted on Sunday, May 18th, 2008 at 2:39 pm and is filed under How To. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.