[5teve]

Hi Guys

I'm trying to decide if its a camera body (k3) or lens issue here.

I have 2 lenses - Tamron 70-200 F2.8 purchased about 4 years ago and repaired (stiff focus) about 1 year ago. I also have a Tamron 28-75 f2.8 about the same age. Both were used previously with my K20 - of which I'm yet to do some testing on.. all testing was done using the lens on my new(ish) k3

Issue occured over the weekend using the 70-200 f2.8. focus seems a little off but live view focus was literally 10's of meters off at distance!

This of course led me to do some testing with both lenses. Chart was used (a4 sized taped to glass on frame). complete refocus was forced for every shot. centre spot focus used. 3 shots taken of each setup (6 on the case of LV 50mm on the 28-75 due to possible inconsistency) f2.8 used on all shots too. Title shows Live View (LV) or PDAF (PD), focal length and number in series (1, 2or 3) The results can be found:-

Here https://plus.google.com/photos/116282250397201644900/albums/6116283759041406...InQhdKAzsnn_wE for 70-200 - focal distance around 2.5-3m

at 70mm live view - consistent front focus
at 100mm live view - consistent large front focus
at 200mm live view - consistent large front focus

at 70mm PDAF - slight back focus
at 100mm PDAF - front focus
at 200mm PDAF - front focus
At f2.8 the lens looks pretty poor also. Not totally convinced that the repair job was a good one. Opinions on this would be welcome.

Here https://plus.google.com/photos/116282250397201644900/albums/6116275155362064...CKm2uv2X54D8WQ for 28-75 - focal distance around 1.5m

at 28mm live view - looks ok on average
at 50mm live view - front and back focus - not consistent
at 75mm live view - slight back focus on average.

at 28mm PDAF - Large front focus
at 50mm PDAF - variable front focus
at 75mm PDAF - med front focus
At f2.8 this lens looks poor also. Opinions on this would be welcome.

I cannot determine if the camera or lens are likely to be at fault here. The large front focus on the 70-200 explains what I was seeing at the weekend with the actual focus point being in a completely different place!

Your thoughts - direction to go from here etc are very welcome. I do have a 16-45 pentax left to test which i'll do when ii get time..

Steve

Also possibly worthy of note is that in live view with the 70-200 it tends to focus through the range and most of the while does it very slowly right until it reaches the end of the focus range then it snaps back much quicker.

---

Last edited by 5teve; 02-16-2015 at 02:21 AM.

[Adam]

Normally, in live view, the focus should be perfect if you have a contrasty target, so that would point to something wrong with the lens. Then again, the K20D's live view AF was a very early implementation so I'm not sure if it could also be to blame.

[5teve]

Normally, in live view, the focus should be perfect if you have a contrasty target, so that would point to something wrong with the lens. Then again, the K20D's live view AF was a very early implementation so I'm not sure if it could also be to blame.

Hi Adam

Sorry I have clarified the wording... all testing was done on my K3.. I just mentioned they were used previously on the k20 as they seemed ok on that.. but need to retest..

Steve

[Adam]

Hi Adam

Sorry I have clarified the wording... all testing was done on my K3.. I just mentioned they were used previously on the k20 as they seemed ok on that.. but need to retest..

Steve

Ah- in that case, I would guess that there's something wrong with the lens. FF/BF doesn't manifest itself in CDAF.

Adam
PentaxForums.com Webmaster (Site Usage Guide | Site Help | My Photography)



PentaxForums.com server and development costs are user-supported. You can help cover these costs by donating or purchasing one of our Pentax eBooks. Or, buy your photo gear from our affiliates, Adorama, B&H Photo, KEH, or Topaz Labs, and get FREE Marketplace access - click here to see how! Trusted Pentax retailers:

I have a similar problem on my Tamron 70-200 f/2.8. In LV it will not get the focus right. On the K-5 it was way off, on the K-3 better but not good. PDAF spot on. It is a bit soft @2.8.

---------- Post added 02-16-15 at 10:42 AM ----------

It has been at the workshop and they told me it was not fixable...

[biz-engineer]

I have K200, K-5 and K-3, Tamron 28-75 and Tamron 70-200. PDAF works fine with all three cameras. K-5 CDAF works for both lenses. K-3 CDAF definitely does NOT work with the 70-200. But K-5 PDAF does not lock with DA*300, while K-3 PDAF works fine with DA*300 also with 1.4 TC. K200 works fine with all lenses with no AF micro adjustments (K200 does not have AF adjust).

In fact, truth be told, the Tamron 70-200 focal length is NOT 70-200, and that's why the K-3 predictive CDAF does not work. It's not a problem with the K-3, it's a problem with the Tamron 70-200 which has false focal lengths. Just compare the field of view of the Tamron 70-200 with another lens from Pentax at the same focal length, you'll be surprised.

Found this; "The Tamron 70-200/2.8 Di VC USD does show an effect sometimes called "focus breathing". "Focus Breathing" is a phenomenon whereby the effective focal length of a lens (as shown by magnification at a given focus distance) changes with focus distance, pretty much always getting shorter as the lens is focused closer. From the measurements I made I'd estimate a focal length of around 196mm at the 200mm setting when focused at infinity (which is well within normal industry tolerance on focal length), 188mm at a focus distance of 10m and on focusing down to a distance of 2m the effective focal length reduced to around 140mm. Such behavior is the rule rather than the exception for zoom lenses. For example the Nikon AF-S 70-200mm F/2.8G ED VR II shows this effect quite strongly, possible to a greater extent than this Tamron lens does. It's not really a problem as such but can surprise some photographers when they compare images of close by subjects made with their 200mm prime to those made with their 200mm zoom. Normally just taking a step or two closer to the subject with the zoom is all that's needed to equalize subject magnification.

I'm aware of some statements in forums and reviews that the focal length of the Tamron 70-200/2.8 VC is actually as short as 185mm. This simply isn't the case (and I actually measured two of them). What those testers were probably seeing was focus breathing. Most people test lenses in a fairly confined space, so the lens certainly isn't focused at (or even near) infinity. Under such conditions many lenses (especially internal focusing zoom lenses) will show a reduction in effective focal length. Note that the definition of focal length is the distance between the rear principle plane and the focal plane when the lens is focused at infinity."

[biz-engineer]

Thanks for this Interesting insight, so it's called "focus breathing". Pentax lenses perhaps have less focus breathing than Tamron counterparts, and CDAF algorithm being misled by this? K-3 CDAF works fine with Pentax 18-250, but fail with Tamron 70-200.

[maxxxx]

Normally, in live view, the focus should be perfect if you have a contrasty target, so that would point to something wrong with the lens. Then again, the K20D's live view AF was a very early implementation so I'm not sure if it could also be to blame.

I have had massive problems with a Sigma 10 2.8 in LiveView on a K-3. The lens would not focus at all. And sometimes it would make the shake reduction move the sensor so much that the whole camera body would shake. Never seen this with any other lens.

[stevebrot]

Thanks for this Interesting insight, so it's called "focus breathing". Pentax lenses perhaps have less focus breathing than Tamron counterparts, and CDAF algorithm being misled by this? K-3 CDAF works fine with Pentax 18-250, but fail with Tamron 70-200.

Focus breathing (strangely inaccurate term) should not effect CDAF performance. Actual vs. reported focal length is irrelevant. CDAF is based on contrast evaluation at the sensor.


Steve

[biz-engineer]

Yes Steve , I agree, CDAF should not be affected; I just thought that the CDAF of the K-3 might include some parameters such as focal length... it was only an assumption. CDAF should actually work... There is always something that does not work with these cameras. But regarding this CDAF issue with the Tamron 70-200, if Liveview is wanted, it is possible to choose PDAF also in liveview mode.

[5teve]

seems like its a problem for 3rd party lenses. There is a post about a sigma 70-200 that needed a software update for it to work in liveview on the k3. I'm not entirely sure that is the issue tho.

Debating whether to send off to CR Kennedy with my lenses and see what they say. As has been mentioned CDAF is contrast detected by the sensor.. so adjustment should never be required....

What else could cause the issue?

Steve

[biz-engineer]

In engineering, AF belongs to the field of servo loop control. If there was no motion, it would be perfectly correct than CDAF or PDAF are independant from focal length etc... In reality, the AF system is composed of sensor , actuator , comparator of actual sharpness versus targeted sharpness value and the feedback loop between then. Servo loop are characterized by what is called phase margin and gain margin, that are basically a measure of how stable is the servo loop in case of random errors introduced in the loop. Getting faster AF require some tradeoff in accuracy and/or phase/gain margins. If you tradeoff accuracy , you get stable AF but less accurate AF. If you tradeoff phase/gain margin, you get good accuracy and fast AF , but more risk that loop exit its stability zone. In CDAF algorithms in K-5 and K-3 are different: the K-5 employs basic maximum search including successive approximation, but in order to reduce the number of iterations (faster) the k-3 is doing several measurement in out of focus areas and then extrapolate to find calculated in focus zone. But if there is too much error between predicted focus zone and actual, the algorithm falls out of stability zone and can never reach the target. With the Tamron 70-200 lens , the K-3 read from the lens and can never reach the target because there is a difference between lens electronic read and actual which is larger that the error K-3 is able to cope with.

[awaldram]

Normally, in live view, the focus should be perfect if you have a contrasty target, so that would point to something wrong with the lens. Then again, the K20D's live view AF was a very early implementation so I'm not sure if it could also be to blame.

The 'reports' of CDaf infallibility are greatly exaggerated.

CD AF works on approximation it has no idea if its OOF to the far or near so multiple attempts focus left check contrast focus right check contrast are required to get close.

How accurate the end result depends on the lens ability to finely 'step' and latency in the drive chain.

Given the Tamrons known inconsistency when AF-c the latency for this lens will be high leading to substandard focus accuracy.

PD in AF-S on the other hand has no issue with latency as it knows exactly where and in which direction focus is, the only issue is whether the PD sensor aligns with cmos sensor but this can be 'calibrated' by the user in most modern cameras. It can eve to some degree overcome course stepping by utilizing the famous Pentax half step (move away from focus x step come back y steps)

Of cause with screw drive the ;'steps' are gear ratios not distinct steps as DC or SDM and represent the smallest movement the lens can achieve.

[5teve]

Thanks Guys...

I'm in contact with a tamron repairer over here to see if there is such a thing as firmware updates on the 2 tamrons I have as they are screw drive lens about 4 years old, not the later motor versions (as there are no pentax vesions of these)

CR Kennedy are willing to take the camera and perform a full calibration on it to check it out, and I can get the Lens's set up and re-aligned at the local repairer for a fee. I'll persue the lens option first i think after checking out the 16-45.. will let you know the results of that testing... I may also do a factory reset on the camera... see if that solves anything

Steve

Edit - Would anyone else out there like to test the live view on a tamron lens, especially the 70-200?