OK, I found my answer, and some animated examples here. <http://graphics.stanford.edu/courses/cs178/applets/autofocusPD.html>
The short story is that phase-detection measures the error and tells the lens which direction and how far to move to get into the correct position. It's faster, but depends on everything being calibrated. Sort of like saying "go three feet to the east and you'll be there." If you both have accurate rulers and compasses it will work fine. The contrast detection method used with live view is iterative and keeps sending correction messages until the image is focused. Slower, but more accurate. So it boils down to the fact that lens and body manufacturing tolerances are wide enough that, for phase-detect focus to be spot-on, each lens-body pair needs to be micro-calibrated. If you can live with the slower contrast-detection focusing of live view, it will probably be more accurate. Anyway, that sheds light on the old "my copy of this lens isn't focusing" statement. It's more like "this copy of the lens on this copy of the body aren't a good match." Bottom line, now I think I understand why the simple calibration method can work. gs George Sinos -------------------- gsi...@gmail.com www.georgesphotos.net plus.georgesinos.com On Wed, Apr 4, 2012 at 1:56 PM, Bryan Jacoby <bryan.jac...@gmail.com> wrote: > On Wed, Apr 4, 2012 at 2:32 PM, George Sinos <gsi...@gmail.com> wrote: > >> If that's true people should be saying this "camera body back-focuses >> with this lens" instead of the more commonly phrased "this lens has a >> back-focus problem." > > I can't say I understand exactly why this is, but phase detection AF > errors can apparently be caused by the body or the lens. > > I understand why a lens with spherical aberration could front/back > focus when used at an aperture setting that's not the same as what the > AF sensor is using (often f/5.6); I'm not sure if this would be > significant. But there seems to be more to it than that: different > copies of the same lens apparently will focus differently on the same > body (see > http://www.lensrentals.com/blog/2008/12/this-lens-is-soft-and-other-myths). > > Aside from the optics, at least in some systems, maybe all, AF lenses > tell the camera body what the focus distance setting is. This, > combined with the "how far out of focus is it" information from the AF > sensor, lets the camera body calculate how much to adjust the focus. > I would think that errors in this focus distance encoding would lead > to multiple iterations before locking on focus, but not errors in the > final locked focus point. > > Can anybody explain the origin of lens-related phase detection AF errors? > > -- > PDML Pentax-Discuss Mail List > PDML@pdml.net > http://pdml.net/mailman/listinfo/pdml_pdml.net > to UNSUBSCRIBE from the PDML, please visit the link directly above and follow > the directions. -- PDML Pentax-Discuss Mail List PDML@pdml.net http://pdml.net/mailman/listinfo/pdml_pdml.net to UNSUBSCRIBE from the PDML, please visit the link directly above and follow the directions.
