Forgot Password
Pentax Camera Forums Home
 

Reply
Show Printable Version Search this Thread
08-08-2016, 12:13 PM   #1
Moderator
Loyal Site Supporter
BigMackCam's Avatar

Join Date: Mar 2010
Location: North East of England
Posts: 5,544
AF accuracy and calibration - flat vs sloping targets

I've re-visited my AF fine adjustments recently for all of my lenses on both my K-3 and K-3II (time consuming, but satisfying! ).

In performing the tests, I've noticed a certain behaviour that seems more pronounced on two of my third party lenses - specifically, the Sigma 17-50 f/2.8 and Tamron 28-75 f/2.8...

If I set up the AF fine adjustment so that it focuses perfectly on a flat chart (in the same plane as the sensor) at close range, when I then try AF with a sloping target (eg. a ruler, brick wall, magazine print etc.), there is a tendency to front focus, and this is consistent.

Taking the Sigma 17-50 as an example, it focuses perfectly on a flat target with zero AF fine adjustment. But if I stand at an angle to the wall of my house (which has nice, gritty bricks that AF picks up on nicely), or point down at an angle to a page of print, it will consistently focus just ahead of intended the focus point. With AF adjusted to around -4 or -5, it consistently focus accurately on the sloping wall, but is then slightly out of focus on the flat chart target.

As I say, this seems more pronounced on my Sigma 17-50 and Tamron 28-75 - which isn't to say that it doesn't happen on my other lenses (mostly Pentax), it's just that I haven't noticed it.

Has anyone else noticed the behaviour I'm describing, and if so, do you know what's causing it - the size of the AF area, perhaps? Or something to do with the AF algorithm?

It's not a big issue... I'm just interested


Last edited by BigMackCam; 08-08-2016 at 12:26 PM.
08-08-2016, 12:39 PM   #2
Site Supporter
stevebrot's Avatar

Join Date: Mar 2007
Location: Vancouver (USA)
Photos: Gallery | Albums
Posts: 26,196
QuoteOriginally posted by BigMackCam Quote
If I set up the AF fine adjustment so that it focuses perfectly on a flat chart
Non-ambiguous target

QuoteOriginally posted by BigMackCam Quote
when I then try AF with a sloping target (eg. a ruler, brick wall, magazine print etc.)
Ambiguous target. The PDAF system is going to lock at the contrast peak (for lack of better words*) within the area of the AF point. Even with careful aim, this may or may not be what you have in mind. That being said, I would normally not expect consistency such as what you report. That the system would pick up the near edge of a receding field is unexpected.


Steve

* What the PDAF sensor "sees" is analogous to a split-image or superimposed rangefinder image only from a histogram perspective.
08-08-2016, 12:47 PM   #3
Loyal Site Supporter




Join Date: Aug 2011
Photos: Albums
Posts: 1,726
I agree with Steve. The focus point is actually bigger than it shows in the viewfinder and PDAF sensor just accidentally picked something closer in this area to focus. Can the camera and lens combo focus accurately in real-life situation, such as portrait? Is it closer to what you have seen when focusing to a flat chart?
08-08-2016, 01:14 PM   #4
Site Supporter
jatrax's Avatar

Join Date: May 2010
Location: Oregon
Photos: Gallery | Albums
Posts: 10,288
When shooting at a flat subject everything on the the target should be in the same plane. However when shooting an angled target that is not the case. Some portion of the target within the AF 'point' will be closer than the rest. Given the same contrast on the area within the AF point one might guess the camera would lock onto the closer portion. Thus giving you the results you are seeing.

As Steve says the sloped target is ambiguous. Now whether the camera is going to consistently choose the nearer portion of an ambiguous target, I do not know and like Steve I find that surprising. Yet your results would seem to bear out that theory.

Another thought is to question whether this was hand held or on a tripod. If hand held it is possible you are pulling slightly as you press the shutter button. If the angle is such that the nearer portion is to your right and you are pulling slightly to the right then you might be getting this result. Try the same experiment but reverse the angle and see if you are now back focusing.

08-08-2016, 04:32 PM - 1 Like   #5
Pentaxian




Join Date: Sep 2008
Location: Phoenix, AZ
Posts: 1,417
QuoteOriginally posted by BigMackCam Quote
I've re-visited my AF fine adjustments recently for all of my lenses on both my K-3 and K-3II (time consuming, but satisfying! ).

In performing the tests, I've noticed a certain behaviour that seems more pronounced on two of my third party lenses - specifically, the Sigma 17-50 f/2.8 and Tamron 28-75 f/2.8...

If I set up the AF fine adjustment so that it focuses perfectly on a flat chart (in the same plane as the sensor) at close range, when I then try AF with a sloping target (eg. a ruler, brick wall, magazine print etc.), there is a tendency to front focus, and this is consistent.

Taking the Sigma 17-50 as an example, it focuses perfectly on a flat target with zero AF fine adjustment. But if I stand at an angle to the wall of my house (which has nice, gritty bricks that AF picks up on nicely), or point down at an angle to a page of print, it will consistently focus just ahead of intended the focus point. With AF adjusted to around -4 or -5, it consistently focus accurately on the sloping wall, but is then slightly out of focus on the flat chart target.

As I say, this seems more pronounced on my Sigma 17-50 and Tamron 28-75 - which isn't to say that it doesn't happen on my other lenses (mostly Pentax), it's just that I haven't noticed it.

Has anyone else noticed the behaviour I'm describing, and if so, do you know what's causing it - the size of the AF area, perhaps? Or something to do with the AF algorithm?

It's not a big issue... I'm just interested
No one has addressed depth of field variance. Yes the sloped ruler may be ambiguous, but that is not what is in play here. The depth of field at 50mm and f2.8 is over 1 foot at ten feet (it's still 6 inches at 75mm) on an APS-C sensor. So yes, your flat target may be perfectly in focus, but is it in the front or back of the in focus zone? You have NO way of knowing. This is why one normally uses a sloped target because that reveals the location of the center focal plane. Your lenses are front focused. Adjust them. If you don;t fell 100% comfortable at -4, compromise with a -2.

As far as the wall goes, for some reason i think i recall something in my physics classes that angling your camera downward has a slightly different effect on how light reaches the sensor versus the sensor being perfectly vertical and shooting a target sloping away from you. the word parallax comes to mind, and i think it has to do with the angled sensor actually artificially INCREASING the DoF for the image, which would give you a false reading of front or back focus. I don't trust my recall. i;m sure someone will correct me...

I'd recommend using a sloped (45 deg) target in the future with the sensor perfectly vertical.

Last edited by nomadkng; 08-08-2016 at 04:42 PM.
08-08-2016, 05:17 PM   #6
Pentaxian




Join Date: Jun 2009
Location: GMT +10
Photos: Albums
Posts: 10,581
QuoteOriginally posted by nomadkng Quote
No one has addressed depth of field variance.
That may certainly be an issue, particularly since not all tests will be done wide open with a 50mm f1.4 on the K-1, or some other camera/lens combo with razor thin DOF.

Field curvature can cause problems with some lenses too, making it difficult for them to deliver perfect focus and/or sharp images across a flat image plane. It's common, except in macro lenses.

A flat test target is perfectly workable though, and in many ways is the simplest and best way to go about focus tuning. It can just be a little tedious to do - eg:

- set AF adjust to -10, shoot,
- set AF adjust to -5, shoot,
- set AF adjust to 0, shoot,
- set AF adjust to +5, shoot
- set AF adjust to +10, shoot
- evaluate results,
- finetune AF adjust incrementally around the best result above.

Sloping test targets are a useful short-cut for the kind of manual procedure above, but they are not any more accurate.

Last edited by rawr; 08-08-2016 at 05:29 PM.
08-09-2016, 12:05 AM   #7
Moderator
Loyal Site Supporter
BigMackCam's Avatar

Join Date: Mar 2010
Location: North East of England
Posts: 5,544
Original Poster
QuoteOriginally posted by nomadkng Quote
So yes, your flat target may be perfectly in focus, but is it in the front or back of the in focus zone? You have NO way of knowing. This is why one normally uses a sloped target because that reveals the location of the center focal plane. Your lenses are front focused. Adjust them. If you don;t fell 100% comfortable at -4, compromise with a -2.
Alas, I don't think that's the case in this instance. The flat target I'm referring to is part of Datacolor SpyderLensCal tool - so I'm focusing on the flat portion, then referring to the slanted ruler to the side to confirm the centre of the depth of field.

QuoteOriginally posted by nomadkng Quote
As far as the wall goes, for some reason i think i recall something in my physics classes that angling your camera downward has a slightly different effect on how light reaches the sensor versus the sensor being perfectly vertical and shooting a target sloping away from you. the word parallax comes to mind, and i think it has to do with the angled sensor actually artificially INCREASING the DoF for the image, which would give you a false reading of front or back focus. I don't trust my recall. i;m sure someone will correct me...
Thanks, that's interesting.

I think I need to do some more formal testing. @jatrax mentioned the possibility that I may be pulling the camera as I shoot, and that's a possibility. The flat plane focus adjustment I did was tripod based, but my shots of the angled wall or page of text have been hand-held, and it's conceivable that I could be the problem rather than the AF system I'll try some more tests and report back...
08-09-2016, 01:00 PM   #8
Pentaxian




Join Date: Dec 2009
Location: Dallas / Yucatan
Posts: 988
QuoteOriginally posted by BigMackCam Quote
I'll try some more tests and report back...
I appreciate the conversation here, because I'm learning from all the educated responses.
Thanks for asking the question and continuing to update us on what you find.

08-09-2016, 02:24 PM   #9
Site Supporter
stevebrot's Avatar

Join Date: Mar 2007
Location: Vancouver (USA)
Photos: Gallery | Albums
Posts: 26,196
QuoteOriginally posted by nomadkng Quote
So yes, your flat target may be perfectly in focus, but is it in the front or back of the in focus zone? You have NO way of knowing.
If so, then it is not perfectly focused. It is easy to assess, however by testing to see if the focus can be improved manually (magnified live view). Your comment regarding distance is very valid, though, and not just from the perspective of DOF. There are multiple reasons to do both focus and resolution testing (sort of the same thing actually) at the conventional distances of 10x or 20x the focal length, not the least of which is the greater ease of actually working the focus mechanism at close-moderate distance.

As for parallax, that is definitely a concern whenever the target is not strictly orthogonal to the lens axis. Software such as FocusTune uses a flat target and applies correction for parallax when assessing the ramp sharpness.



Steve
08-10-2016, 07:38 PM   #10
MSL
Pentaxian
MSL's Avatar

Join Date: Feb 2012
Location: Greater Toronto Area
Photos: Albums
Posts: 3,359
QuoteOriginally posted by BigMackCam Quote
In performing the tests, I've noticed a certain behaviour that seems more pronounced on two of my third party lenses - specifically, the Sigma 17-50 f/2.8 and Tamron 28-75 f/2.8...
Mike,
When you do the calibration, which focal length did you choose? And do you test that it is still valid at other focal lengths?
Reply

Bookmarks
  • Submit Thread to Facebook Facebook
  • Submit Thread to Twitter Twitter
  • Submit Thread to Digg Digg
Tags - Make this thread easier to find by adding keywords to it!
accuracy and calibration, af, af accuracy, calibration flat vs, camera, chart, dslr, focus, lenses, photography, sigma, target, wall
Thread Tools Search this Thread
Search this Thread:

Advanced Search


Similar Threads
Thread Thread Starter Forum Replies Last Post
Select AF point(s) vs. single (spot) point ... any difference in accuracy? jpzk Pentax K-3 35 05-06-2016 05:31 AM
Using AF for moving targets? Ilovemypentax Pentax K-3 7 05-28-2015 09:02 AM
AF vs MF logic and accuracy alvarossorio Pentax SLR Lens Discussion 11 07-27-2011 10:14 AM
AF and focus accuracy using older firmware vs. new (1.03) on K5 plus DA*300 jpzk Pentax K-5 65 05-13-2011 12:35 PM
AF Speed vs Accuracy Workingdog Pentax News and Rumors 9 12-25-2007 08:19 AM



All times are GMT -7. The time now is 04:43 PM. | See also: NikonForums.com, part of our network of photo forums!
  • Red (Default)
  • Green
  • Gray
  • Dark
  • Dark Yellow
  • Dark Blue
  • Old Red
  • Old Green
  • Old Gray
  • Dial-Up Style
Hello! It's great to see you back on the forum! Have you considered joining the community?
register
Creating a FREE ACCOUNT takes under a minute, removes ads, and lets you post! [Dismiss]
Top