[quote name='Frank' timestamp='1342775556' post='19482']
Thank you for the answer.
However, I am still confused. According the article [url="http://mansurovs.com/how-phase-detection-autofocus-works"]http://mansurovs.com...autofocus-works[/url], the phase-detection AF focus system works in the following ways (cited from that article):
- The light that passes through the extreme sides of the lens is evaluated by two image sensors
- Based on how the light reaches the image sensors, the AF system can determine if an object is front or back focused and by how much
- The AF system then instructs the lens to adjust its focus
- The above is repeated as many times as needed until perfect focus is achieved. If focus cannot be achieved, the lens resets and starts reacquiring focus, resulting in focus “hunting”
- Once perfect focus is achieved, the AF system sends a confirmation that the object is in focus (a green dot inside the viewfinder, a beep, etc)
"Once perfect focus is achieved, the AF system sends a confirmation that the object is in focus (a green dot inside the viewfinder, a beep, etc)". So I assume that when the AF sensors "sees" a state in the Figure 2 in that article the camera "thinks" that the subject is in foucus, then the camera tells the motor stop moving and let the lens elements stay in their positions, then images are taken. So, according to your answer, it can happen that for some problematic lenses, although the motor puts the lens elements in wrong positions, the AF sensors in the camera can still see a state as in the Figure 2 and then a out-of-focus image is taken. Is it right?
I am not Serkan, of course, but I'll give this a try regardless <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='
With PDAF, the camera doesn't actually try to focus at all, that's just the theory. It tries to calculate a best match for the amount of lens extension or focus element movements (stepper motor increments) required for the lens to focus correctly based on the starting position of the lens. Furthermore, there is no active feedback loop with PDAF, although focus hunting in difficult cases comes fairly close.
A lens really only reports back that is has executed the command provided by the body successfully (or not), i.e., it has activated the stepper motor the required number of increments in the direction as indicated by the camera.
Essentially, the camera AF system has a special profile for a specific lens, or a default profile. The lens electronics need to know how to react to stepper motor increment increase requests. If a lens does not occur in the camera AF lens database, the lens manufacturer has to either adapt (and reverse engineer) the stepper increment requests for the lens mapped to an existing lens, or to the default. Do note that the signal given by the lens to the camera really only indicates that it could successfully execute the requested command as generated by the camera AF system, e.g., move the lens x
steps forward <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='
This means that there are three potential problems possible here, where there still appears to be an in-focus signal, as the lens executes the command requested by the camera:
1) The camera vs lens tolerance for increases is not within tolerance, f.e., the body really is at -4 and the lens at -4 (generally causes by mount tolerance variations), for a total of -8, while tolerance is, e.g., -5, resulting in OOF images all of the time (even if at smaller apertures they may still appear in focus)
2) The lens tolerance does not conform to the camera AF system lens profile, whether default or specific, which may result in variation in achieving focus. It may be out all the time, or only at certain specific requests. Generally speaking, one would expect it will be out all of the time in this case, although images again may appear in focus due to DoF.
3) The lens profile as implemented in the lens doesn't conform to the lens profile of the camera AF system. Focus will likely be all over the place, although in repeatable fashion, with different focusing points at different distances or even with different incremets asked for by the AF system. The exact variations may be repeatable, but they may also be not, depending on the movement the lens has to make.
Of course, the lens (or camera AF) electronics may be faulty too, but I have excluded that from this discussion.
A problem like #1 is typically an OEM problem, and generally can be fixed by the manufacturer. #2 may be an OEM problem, but also a 3rd party lens manufacturer problem. #3 typically is a 3rd party lens manufacturer problem.
Besides the above, there still are lens AF problems possible due to optical aberrations, specifically spherical aberrations, which may cause focus shift varying with distance and aperture (less with focus further away and less with smaller apertures), and things like astigmatism etc., which may throw out the PDAF system especially at larger apertures. WA and UWA lenses also may give problems, because for PDAF the beams required for focusing should not cover a too large area at a fairly large AoV, because it will have problems matching these. The latter is inherent to the PDAF system. This is also why the AF point coverage with PDAF systems never covers the entire image in the VF. Do note that the "extreme sides of the lens" for interpreting AF is incorrect; this should be extreme sides of the (specific) AF sensor, or rather, the two elements (or more, but always in pairs) capturing the split image part used for AF calculations. Another optical problem potentially causing focus problems with PDAF is field curvature, but there are a few more anyway. When one thinks about it, it really is amazing how well it generally works, if you ask me <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='
BTW, I have experienced problems in all the 3 categories mentioned.
#1 generally is the easiest to fix, by the manufacturer, or by oneself with MFA, provided, in the latter case, a body actually has MFA. I have had this happen with a bunch of large aperture lenses, especially.
#2 can generally be fixed by the manufacturer, 3rd party or not, but may require some time, especially if the lens electronics do not correspond properly to the camera AF system lens profile. I've had problems with Sigma lenses specifically in this regard.
#3 is the hardest problem to fix, and can really only be done by the OEM or 3rd party manufacturer. I reckon an example of this is the Canon 50L, which requires a special profile IMO to work correctly at all distances and apertures. The problems with this lens, if not manufactured and assembled by the smallest of tolerances, were very repeatable on an individua lens basis, however. Canon improved the lesser PDAF systems from the 50D and 5D II onwards, so there were certainly less problems with the newer bodies from thereon, but I reckon they also improved the lens profile and lens manufacturing and assembly to get better results.
Kind regards, Wim
Gear: Canon EOS R with 3 primes and 2 zooms, 4 EF-R adapters, Canon EOS 5 (analog), 9 Canon EF primes, a lone Canon EF zoom, 2 extenders, 2 converters, tubes; Olympus OM-D 1 Mk II & Pen F with 12 primes, 6 zooms, and 3 Metabones EF-MFT adapters ....