Wednesday, June 4th, 2008

Digital Focus Hell

One of the biggest new issues with digital photography are problems with auto focus. It’s not exactly clear why, but digital cameras (DSLRs) seem to suffer from focusing issues far more prevalently than film cameras. It might be that we tend to look at our images more closely than film, with 100% pixel peeping. It might be the sensors are just more sensitive to focusing errors.

In any case it is prevalent, with some cameras like the Nikon D200 being notorious for it.

Focusing issues are generally grouped into two categories:

  • Front focusing.
  • Back focusing.

“front focusing” is where the focus is in front of the the object being focused on. “back focusing” is where the focus is behind the the object being focused on. So, for instance in the case of “back focusing” where you focused on the eyes, the ears might be in focus instead of the eyes. With front focus in the same example, the nose might be in focus instead of the eyes.

It doesn’t take much back or front focus to ruin a photo, particularly with portraits, macros, and images with low depth of field (DOF). Stopping down more can help, but DOF isn’t a panacea for accurate focus, and usually it’s evident when focus is off, even with sufficient DOF. It’s sharp, but not exactly the same.

Not all cameras suffer from the issue and often the effect is different depending on the lens being used. For instance on my Fuji S5 (based on the D200 body) my 85mm f1.8 was dead on, while my 60mm f2.8 was about .5 cm back focused. My recently purchased 35-70mm f2.8 was a full 2 cm back focussed!

I say “was” because after sending the camera in to Fuji for warranty repair and the lens to Nikon, it appears to have actually become slightly front focused, but much, much better. Similarly other lenses, like the 60mm look dead on, though I haven’t had time yet to do actual focus tests.

Speaking of focus tests, if you suspect you have front or back focusing issues, there are a number of sites with focusing charts, including:

Usually the tests involve shooting a chart at a 45 degree angle, however lining up a set of 5 AA batteries on a diagonal can be equally as effective, if perhaps not as accurate at telling you the exact distance of back/front focus.

There is also a lot of argument about of whether shooting a flat object for a test is accurate. In fact if you see this post:

http://photography-on-the.net/forum/showthread.php?t=10526

It shows an actual example from a Canon repair center where the focus point is perpendicular to the flat test chart, essentially creating a 3D object to focus on.

My personal experience with the tests seemed to show that the using a flat chart exaggerates the focusing errors. This isn’t a particularly bad thing because it makes the issue more obvious, but it also can convince you the issue is severe when it isn’t. In the end what I’ve done is placed batteries on a diagonal on the test chart lined up with the marks, effectively mixing the two, but primarily using the batteries as my indicator.

So, why do these focusing errors happen? There are a number of reasons including possible misalignment of the mirror box, but according to this post (it’s pretty technical but definitely worth a read):

http://forums.dpreview.com/forums/read.asp?forum=1030&message=24839477

most of it comes from an interaction of the focus sensor position and the design of the lens.

There is hope though, with adjustment things can be improved.

Now you might be tempted to use the D.I.Y. (Do It Yourself) technique as suggested by sites like these:

Personally I would avoid it because as pointed out in this post:

http://www.nikonians.org/forums/dcboard.php?az=show_topic&forum=147&topic_id=37618

at the manufacturer it isn’t as simple as a mirror box adjustment, which is what these are doing. Yes, for minor focusing adjustments it may help, but in many cases if you break it you bought it and it isn’t really the way to do it. The way to do it involves computers and special equipment.

This is consistent with my conversations with both Fuji and Nikon regarding the issue. Both technicians I talked to indicated that it isn’t simply moving the entire focus back or forth (which is what happens with the mirror box adjustments). In that case, say if you were fixing a back focus issue, you’d be moving the general focus to the front. Thus if you happen to have a lens that unlike your others is tack on, it’s going to become front focused with this adjustment. Not exactly optimal.

However with their computerized methodology, and both vendors indicated this, the lenses that were back focused come into alignment, while the lenses that are currently on, stay on! In fact Nikon even suggests sending your “standard” (most used) offending lens in with the body and they claim, when done, that the camera lens combination will be calibrated together, with the already “on” lenses staying on. Fuji on the other hand indicated that usually only the body is needed. Incidentally the term for calibrating the lens and body together is called “collimation“.

Because computers are involved, though I have no proof of this, what I suspect may be going on is (among other adjustment) is they’re actually programming the camera with lens information. On the D300 you can actually do this yourself through menus, which frankly is one of the more compelling features of that camera.

Sometimes yes, it is an issue with the lens itself, however mostly not. If the elements are misaligned it is possible it could throw focus off, but in most cases where it effects more than one lens, it’s the body.

Interestingly enough, according to Nikon technicians, AF-S lenses can have focus adjustments made to them. This I suspect is possible because they have some intelligence built into them, whereas D and pre-D lenses are more simplistic and cannot have the same firmware adjustments. However, again, if they have a real mechanical issue they can be repaired, which seems to be the case with my 35-70mm. However in general I wouldn’t count on it.

If you make these tests and do find an issue, a repair is likely to be around $120 in the examples I see. That to me seems a deal for the peace of mind knowing the work is being done by a professional who knows what they’re doing.

Well that’s it. Hope that helps.

Leave a Reply