[planedriver]

If anyone wants to compute the effect of "focus and recompose", I developed a formula in another thread about image sharpness. Plug in values for "d = distance to subject" and "alpha = angle from frame centre to point of focus" into

x = d (1/(cos alpha) - 1)

and "x" will amount to the additional distance introduced to the point of focus due to the rotation of the camera. A DOF calculator will then tell you whether the point of focus is still within the DOF. BTW, the effect is often quite small but big enough for Hasseblad to built in rotation sensors into some of their models so that they can readjust the focus accordingly. I didn't try to assess the respective values for the shot provided, but I very much doubt that the FF can be explained by a "focus and recompose" error.

..and in my case I used the selectable AF points..

Andras

[Class A]

..and in my case I used the selectable AF points..

...which is a very good idea. I also avoid "focus & recompose" by using manually selected AF points.

I was just providing the formula for those who were wondering about the magnitude of the effect.

[planedriver]

...which is a very good idea. I also avoid "focus & recompose" by using manually selected AF points.

this is getting slightly "off topic"..still useful..on my K-x the lack of focus confirmation points forced me to use the "focus and recompose" way..now with shallow DOF I found it very unreliable.

Andras

[Jonson PL]

Here's my K-5 results I posted in an old thread ...

f/1.6 ISO 6400 100% crop. no PP. subject distance about 25 feet. dark street, yellow license plate lit by car halogon headlights. I think it's about EV2

Live view AF


Regular AF

These pictures reminds me of Lloyd Chambers work with the Nikon D3 and 24/1.4 :

diglloyd.com blog: May 2010

[jstevewhite]

The whole thread reminds me of my first "real" camera. It was a Minolta SRT101 with a matte focusing screen and a 50mm f1.4 lens. Focus was, you might say, erratic - even with my 18-year-old eyes.

My K-5 still "misses" sometimes with the 50mm 1.4 @ f2. With firmware 1.02, it was just 'off', and really badly off with shorter lenses, to the point where my 10-20 was unusable (in AF). With 1.03, now, it's on *most* of the time. Still misses sometimes with the longer, faster lenses, but generally "on".

I wasn't presenting the focus shift as you leave the axis as an explanation; I was just saying that yes, focus *does* shift as you move away from the axis, as Class A further clarifies. Sorry if I wasn't clear.

[Emacs]

I was shooting a dancing performance yesterday in quite a low light conditions (was force to use the only 31Ltd, since there were too many people and I couldn't reach the table where I put my lenses at). The f1.8 dissapointed me much, it was always missing with FF, but the AF improved when I closed the aperture down to 2.8. The FA* 85Ltd acted the same: quite a large FF at f1.4, but great improvement at f2
BTW My 31Ltd copy seems to be broken once dropped, I've taken it to the repair service (as well as 77Ltd, which had slight FF and was dropped too). Unlike 31 and 77, my 43 has no noticeable impact of FF issue since 1.03 arrived.
85 f2.0:



31 f2.8

[falconeye]

I updated my low light focus study to reflect the firmware 1.03 changes. The photo above is a reminder to myself that all of this is meant to deliver photos in the end.

The corresponding blog article is here:

-> Falk Lumo: LumoLabs: Pentax K-5 low light focus with firmware upgrade 1.03

As I have already mentioned in this thread, I have difficulties to see what Pentax did in release 1.03. It doesn't seem to be visible in a straightforward testing situation.

I therefore invite further tests around the dark background and/or low contrast/dim AF feature situation. One would need a release 1.02 K-5 camera to study if the differences are in this domain. Something I don't currently have anymore.

My best guess now is that the changes are marginal except if the background is dark indeed and AF spot and overall color do match.

[Class A]

The photo above is a reminder to myself that all of this is meant to deliver photos in the end.

Focused manually?

Hope you didn't use the automatic horizon correction in which case the camera needed calibration...

Love the colours! And the image really does intrigue a beholder's curiosity. Interesting structures in front of the sun. Seem to be just clouds, but that doesn't become clear at first glance.

One would need a release 1.02 K-5 camera to study if the differences are in this domain.

What really matters, AFAIC, is how the K-5 performs with 1.03 in absolute terms. Comparisons to 1.02 might be interesting in order to find out what Pentax did, but isn't it much more relevant to determine in what situations 1.03 still fails?

It seems that the K-5 can create FF when focusing on skin and a slowish lens in EV 5-6. That doesn't sound great to my ears, but maybe my expectations are exaggerated? Is it correct to say that the K-5's AF module is more sensitive than previous modules but also is plagued by higher dispersion and hence struggles more to correct for this effect?

I'd like to learn how the K-5 compares to other cameras like the K20D or K-7 in terms of front-focusing in low light. The question for me is: Is the K-5's AF worse than that of previous Pentax cameras in some situations or is it not?

EDIT: Reading the updated version of your report leads me to the conclusion that the K-5 still has an FF problem. Too bad. As much as I would like to believe in a possible fix to a hardware limitation (or less likely a firmware bug), the long time it took Pentax to release a fix doesn't make me optimistic that they will be possible to address the problem in the future. Could still be fixable but I won't hold my breath.

Even if a new firmware came out that allowed user input to alleviate the problem, this would still be disappointing AFAIC because the K-7 didn't need such a crutch.

[Ash]

An interesting finding Falk.
In real terms there is little noticeable improvement from your tests.
I'm going to be trying out my own K-5 v1.02 before and after firmware to see the difference myself.

[Class A]

Before it gets lost in yet another edit to my previous post: Thanks a lot for the further work you've done, Falk!

I hope Pentax will deliver in the future, one way or the other. I understand the K-5 is a great camera the FF problem notwithstanding but I personally take issue with unfinished work like this and it doesn't seem to be very typical of Japanese engineering either.

[bjan]

Thanks a lot Falk for all the work you have done.
I have made my mind up finally after the firmware update and decided to NOT buy the K5 for the second time after having parted from it in January.
I have been praying and hoping for real improvement till the firmware came out but with a broken heart I have to confirm my decision.
The FF issue under artificial light is a common problem with fast lenses in the DSLR world and it seems to get more obvious the "better" the sensors get in the digital era. It is not only PENTAX suffering but also the CANIKON world. (see posting in NIKON D7000 and CANON 7D forums to stay with comparible cams)
I do enjoy my CANON S95 a lot in many occasions but still will be on the watch for a proper focussing FAST camera in the future that will fill in my needs for a lot of situations as well. One thing is sure. The technical development of Auto Focus technology will master the issue one day. With mirrorless or Translucent or conventional technology or whatever will be further developed. Contrast AF might have a future here and we will see fast development in this for sure.
Thanks all for the input on the K5 FF issue and hope to meet soon again

[bjan]

By the way Falk . On your nice pic of the sunset the horizon is not level. Something wrong with the electronic level on your K5 ?

[ogl]

I updated my low light focus study to reflect the firmware 1.03 changes. The photo above is a reminder to myself that all of this is meant to deliver photos in the end.

The corresponding blog article is here:

-> Falk Lumo: LumoLabs: Pentax K-5 low light focus with firmware upgrade 1.03

I have 1.02 firmware.
http://www.easy-share.com/1914409908/k5_e102w.EXE

[Michaelina2]

Before it gets lost in yet another edit to my previous post: Thanks a lot for the further work you've done, Falk!

I hope Pentax will deliver in the future, one way or the other. I understand the K-5 is a great camera the FF problem notwithstanding but I personally take issue with unfinished work like this and it doesn't seem to be very typical of Japanese engineering either.

May I ask, what about all the positive reports from users regarding improvements attributed to v.1.03 during field use? Are they to be casually dismissed?

While Falk's test read-outs may be correct for the subject camera, is it possible that the camera used in his experiment is NOT a typical K-5? Given all of the subtle adjustment options in the K-5, is it possible he missed something and somehow set the camera up wrong?

Questioning un-replicated N=1 test findings generalized to a larger population (findings that are contradicted by field users) before postulating an un-testable 'corporate conspiracy theory' seems only fair to Pentax and Japanese engineering, would you not agree?

Just curious...

[falconeye]

May I ask, what about all the positive reports from users regarding improvements attributed to v.1.03 during field use? Are they to be casually dismissed?

That's the interesting question now!
I don't dismiss them as being casual and look forward to find out under which circumstances the progress is real. In my updated report, I explained circumstances when it is not.

While Falk's test read-outs may be correct for the subject camera, is it possible that the camera used in his experiment is NOT a typical K-5? Given all of the subtle adjustment options in the K-5, is it possible he missed something and somehow set the camera up wrong?

I consider this to be an unlikely possibilty. Possible but unlikely so. There are others not seeing an improvement too.

Also, if you look at the absolute EV values where the problems emerge, I consider it to be a more likely possibility that Pentax addressed the problem where the problem emerged under even higher EV values.

This really needs a test with less favourable testing conditions than the typical lab test constitutes.

Questioning un-replicated N=1 test findings generalized to a larger population (findings that are contradicted by field users) before postulating an un-testable 'corporate conspiracy theory' seems only fair to Pentax and Japanese engineering, would you not agree?
Just curious...

Your "Just curious" posts are not necessarily your best. I agree, asking critical questions is a good thing to do. The scientific way would be to additionally provide own testing results contradicting mine.

As a side note, there are currently a number of "I don't see the progress/little progress only" reports and a number of the "problem is fixed now" reports. I am really curious to find out what makes the difference between both populations. And it isn't an N=1 anomaly.