[MrPetkus]

I can confirm this and Christine Tham's results. AF fine adjustment wiped, marked improvement in focus.

No...

Do a global AF reset for your lenses...

Menu > C26 AF Fine Adjustment > 2 On > Reset > OK > Menu > OK > Menu > Camera ready state

The read out after AF Reset should be:

Apply All.... +/- 0
Apply One.... Unset

--------

After the camera reset (Manual pp 335), I set the Program Line to MTF, selected "P" mode, then the C26 reset (above).

It may not be 'science,' but I see a HUGE improvement in AF accuracy. In fact, I no longer need to AF fine tune any lens... load, lock, and Go!

Cheers...

[dgaies]

Christine, you entered "0" for both, per lens AND global adjustment, when using "0", right?

No...

Do a global AF reset for your lenses...

Menu > C26 AF Fine Adjustment > 2 On > Reset > OK > Menu > OK > Menu > Camera ready state

The read out after AF Reset should be:

Apply All.... +/- 0
Apply One.... Unset

--------

After the camera reset (Manual pp 335), I set the Program Line to MTF, selected "P" mode, then the C26 reset (above).

It may not be 'science,' but I see a HUGE improvement in AF accuracy. In fact, I no longer need to AF fine tune any lens... load, lock, and Go!

Cheers...

I think falconeye was referring to her settings for the "AF adjustment ON" shots, to make sure there was no global nor individual adjustment on.

[Class A]

AF fine adjustment wiped, marked improvement in focus.

Does this hold true for "all" scenarios, i.e., yellowish low-light and bright daylight?

It would be good to explicitly rule out the possibility that cameras who see an improvement after wiping the AF adjustments don't have a natural back-focusing issue which is nicely compensated by -- under normal circumstances undesirable -- front-focusing.

[CorneliusK]

Really an eye opening experience reading about the 101 things which can affect focusing behavior.

Simply mind boggling. I hope that Pentax is able to get on top of this.

[MrPetkus]

I certainly see an improvement but the FF issue still persists when pushed. This is not a cure by any stretch but yet another variable that may need to be factored-in. It's dandy that forum members have the wherewithal to conduct their own little controlled studies but this isn't going to go anywhere until Pentax engineers are commanded to hunker down and deliver a fix for consumers.

We're starting to sound like Feynman determining the cause of the Challenger disaster only this time it's a $1500 consumer camera and we're armed with a patent application and photograph of a Pentax body cutaway.

[Class A]

I certainly see an improvement but the FF issue still persists when pushed.

Can you observe backfocusing under normal conditions? Surely, there must be a reason why you put in the AF adjustment values in the first place?

This is not a cure by any stretch but yet another variable that may need to be factored-in.

Ideally, it shouldn't be a variable if adjustment value "0" means "no deviation from the default setting".

It's dandy that forum members have the wherewithal to conduct their own little controlled studies but this isn't going to go anywhere until Pentax engineers are commanded to hunker down and deliver a fix for consumers.

Yes and no. No, because if something substantial is unearthed here which points to a manageable fix then the engineers might get the green light to actually implement it. As long as everyone (not only the colorimetric sensor ) is in the dark about the true cause, the engineering team may be forced to work on more promising issues rather than "wasting" time on something that might not have a cure.

We're starting to sound like Feynman determining the cause of the Challenger disaster only this time it's a $1500 consumer camera and we're armed with a patent application and photograph of a Pentax body cutaway.

We got a bit more than that but, true, we could do with a General Kutyna. Perhaps he is among us already!

[Ray Pulley]

WRT operating range. It would be a pitty if the AF cannot exploit its fabulous sensitvity just because some lenses are too slow.
[/B]

Ok, all of this shooting light into the eyepiece is interesting, but it really isn't very deterministic of much, IMO. The light that you pass through the eyepiece is likely ending up getting to the mirror and probably also to the PDAF module, but not in any sort of known or controllable manner, which means that you might see all sorts of odd results doing this depending upon many different variables.

Anyway, it seems to me that there is a good way to test the reduced light theory which also contols lens and light color variables, and which may also reveal a work-around, depending on the results:

1. Select the FA 50 f1.4, or a good fast prime lens with an aperture ring and max aperture under f2.0.

2. Setup a test lit with balanced light that is not dominated by one wavelength. It would be nice to know if the camera is calibrated closer to daylight or more like the older cameras at 2856K, but for this test, this is not so important, and we can retest in different color temps if we want.

3. I'm thinking that the light seen by the camera should be about EV3, f1.4. Set the shutter to get a good exposures based upon the test target chosen.

4. Set the lens to manual aperture selection, f1.4 and also set MF to on.

5. Manual focus the lens at f1.4 EV3 until the focus indicator light shows focus lock. Take test images.

6. Move the aperture to f2.0 and then re-focus, once again using the focus indicator to know when the AF system thinks the subject is in focus.

7. Without disturbing the focus, carefully move the aperture back to f1.4 and take more test images.

8. Repeat for f2.8, f4.0, and f5.6.

This method removes the lens and color temp variables and simply progressively delivers 4 less stops of light to the AF and color sensing system during the manual focus process. The result will be focus tests taken from EV 3 to EV-1 which are based upon what the AF system thought was in focus at each reduced light level but at the same color temp, but with the same lens.

You can repeat the test with a different color light source to see if the result is the same or similar.

Most interesting would be the result that the system indicates the proper focus points this way, especially if the test results are similar at a higher and lower color temp.

If the result is that the MF focus lock is fine this way or the step to FF is not present, then one answer might be that the color sensor is not engaged when the AF lever is set to AFM.

That would be a solid work-around were it to turn out to be true.

If the color sensor is not engaged, I would stiil expect that the AF point would move towards FF as the color temp is reduced, but in a more progressive fashion just like every other Pentax model of the past that I have used and also like many other brands. The reasons for this shift (and the reason Pentax designed in a color sensing system) are described in the patent, and make sense optically, but the effect is progressive based upon wavelength and does not have the step seen with the K5.

I no longer have a K5 or I would have set this up and tried it already.

Ray

[ovim]

[b]1. Select the FA 50 f1.4, or a good fast prime lens with an aperture ring and max aperture under f2.0.

2. Setup a test lit with balanced light that is not dominated by one wavelength. It would be nice to know if the camera is calibrated closer to daylight or more like the older cameras at 2856K, but for this test, this is not so important, and we can retest in different color temps if we want.

3. I'm thinking that the light seen by the camera should be about EV3, f1.4. Set the shutter to get a good exposures based upon the test target chosen.

4. Set the lens to manual aperture selection, f1.4 and also set MF to on.

5. Manual focus the lens at f1.4 EV3 until the focus indicator light shows focus lock. Take test images.

6. Move the aperture to f2.0 and then re-focus, once again using the focus indicator to know when the AF system thinks the subject is in focus.

I think you'd need a preset M42 lens to do this. FA (as well as F, A, M and K- lenses) always focus wide open regardless of what the manual aperture setting is set to. So you are not reducing the light getting into the camera while focusing by closing the aperture with FA or any other "modern" lenses.

[dlacouture]

Tried the Reset trick... Did nothing for me...

I'll try the flashlight trick (either in the eyepiece or somewhere in the scene) in real world shots, to see how it goes with faces... But in my test scene, it's a 100% hit (both trick)...

[raphtze]

tried the reset trick....and it does work a lot better, but it's no means fool proof. that being said....it's time to say good bye to my K-5.....sending it in for the sensor issue....

-raph

[raphtze]

oh and i'm a little late to this, but super duper big thanks to falk for doing such a comprehensive test!!! it is humbling to see the dedication + work you put into this. the ball is definitely in pentax's court now.

-raph

[Rin]

i really hope a firmware can fix this, the testresults seem to be consistent enough for it, at least it should be possible to minimize this problem.

[Ray Pulley]

I think you'd need a preset M42 lens to do this. FA (as well as F, A, M and K- lenses) always focus wide open regardless of what the manual aperture setting is set to. So you are not reducing the light getting into the camera while focusing by closing the aperture with FA or any other "modern" lenses.

You might be right now that I think about it, or at least a non-A lens, as the aperture lever of the A lenses is held wide open mechanically once attached to the camera.

Thanks for pointing that out.

Ray

[Duplo]

You might be right now that I think about it, or at least a non-A lens, as the aperture lever of the A lenses is held wide open mechanically once attached to the camera.

Thanks for pointing that out.

Ray

Iwill not be able to confirm until tonight, but IIRC onecould use the aperture ring on the lens to stop down, to avoid what you describe above.

The aperture level is only controlling the aperture if you set the aperture ring in the "A" position.

[JeffJS]

Iwill not be able to confirm until tonight, but IIRC onecould use the aperture ring on the lens to stop down, to avoid what you describe above.

The aperture level is only controlling the aperture if you set the aperture ring in the "A" position.

No. What you Can do is to rotate the lens on the mount so at f22 (or whatever the small aperture is) is completely stopped down. Then turning the aperture ring will operate the blades as it would with an M42 lens (set to manual). You can also use this to shoot Av with K/M lenses. Otherwise, the aperture remains full open until the shutter is pressed (or optical preview is activated or the green button is pressed).