[firefly]

Hi all,

I thought I'd pose a question to the forum, since I'm unable to understand why it happens. I hope someone has the knowledge to answer me. First, some background.

I have a Pentax K100D, going on on year now. The camera seemed to function well with both the 18-55 DA and 50-200 DA kit lenses. Yet, after some time of use I noticed that some shot aren't as sharp at certain focal lengths as others. With experience I became more familiar with how to judge image quality in a photo. This led to me inquiring with my father about an old Pentax film SLR that I remember from my childhood. He went through the closet, pulled out an old Pentax ME body with the lens still attached.

This turned out to be the Pentax-M SMC 50 f/1.7 lens. I suddenly began to learn photography 'old school' and began using the lens almost exclusively for some months. I became quite frustated with the focus consistency, as the focus screen did not have the ability to distinguish sharp focus at such wide aperture, so I purchased a split screen from eBay and installed it myself. Success! Now all my photos come out in sharp focus with the manual fifty.

I began to notice however that 'perfect' focus in the split screen did not light up the AF green haxagonal focus indicator in the viewfinder. After some testing, I discovered that indeed the split screen was correct, and assumed the AF sensor plane was not quite the right distance from the mount. After much research I found that the K100D has a service mode which you can enter and input an AF bias, which basically corrects AF in software. After some tinkering with values and a focus chart, a value of +30 micro meters was used, which led to both the AF indicator and split screen calibrated perfectly together... or one would think. The manual lens was working perfectly, the kit lenses became sharper, everything was great. (*NOTE* I do not believe the +30 value actually shifts the sensor plane at all, but I do believe it creates an offset in software to correct for the AF - so in essence it would be as if the plane was moved)

This is where I became lost.

Growing confident in my ability to handle and calibrate equipment, I purchased a Tamron SP 28-75 f/2.8, to be able to auto-focus in low light and have a very versatile 28-75 zoom while sacrificing very little in resolution. The lens just arrived and I am totally confused as to what I am seeing.

First, the lens front-focused terribly in AF. I thought to myself, no big deal, I'll just MF the lens. That ALSO did not work, even though the split screen showed the scene was perfectly in focus, upon reviewing the photo, it is obviously front focused. So both the AF sensor (which is calibrated using the 50 to the split screen) and the split screen are not able to focus this lens. In a last ditch effort I was able to get the lens to autofocus properly using the service mode... using a value of -250 micro meters as the AF bias.

So now, the lens will autofocus properly. However, the viewfinder shows a very large discrepancy in the scene, which is now visibly 'split' in the split screen. How can this possible by? Since the camera was calibrated with the original +30 micro meters I've ensured that the optical length between the mount and the AF sensor plane is the same as the distance between the mount and the split screen - which produces SHARP photos when focused with any of my other lenses.

I could understand if the AF was misfiring and not actually focusing the lens, which could then be correctly using the split screen image... but a perfectly focesed viewfinder image will still produce backfocus. How is this lens able to project light correctly in one optical path but not the other? I sincerely hope someone has an answer and explanation to this, as it is beyond what I can figure at the moment.

Geoff

---

Last edited by firefly; 10-22-2008 at 06:05 AM.

[wcleong]

There's the potential that you didn't see any problems with the kit lenses because they are slower (smaller apertures) and have larger DOF, so if there is a little BF/FF it still may be within it's DOF.

Now, if you see a discrepancy between the focus screen and the CCD, ie. when the viewfinder is in focus the recorded CCD image is not, and vice versa, then it would indicate that the lens distance to the focus screen/CCD are not the same (they should be for the image to be in focus at the same time for both). And AFAIK this shouldn't matter what lens is attached. If the distance is the same, what you see should be what you capture (except for the stopping down of the aperture).

At least this is my understanding of how these cameras work.

It might be a long shot, but I've read that some split focus screens come with shims, is it possible you didn't install this? (or potentially put it on the wrong side of the screen?)

[firefly]

Thanks for your intput wcleong. It did occur to me that the larger DOF of the slow lenses may have been masking a problem all along. However I do have a 50 f/1.7 with a DOF that was razor thin.

I determined that the CCD was about the right distance away since the split screen focused this lens perfectly. The only problem I had with it was that the AF focus indication did not light up when the lens was perfectly focused. I was able to correct this with a +30 um adjustment in software. Now the 50 f/1.7 focuses perfectly in accordance with the split screen value... and the AF focus indicator lights up at that point as well.

I will go home later today and take a few pictures of a focus chart and try to use my point and shoot to take a photo of the viewfinder for each shot as well, to better explain the problem.

But you are right... I should return the focus bias to +30 um (the prime 50 calibration) and manually focus again using the focus chart, and see what I get. Does this sound right to you? Also, no the split screen did not have a shim, but I used the original one in its original location.

[konraDarnok]

I have experienced a very similar story. I have come to the conclusion that the outlier lens is the aberation. I'd send the tamron back and request a new one.

[wcleong]

I see it as two separate problems, one with the AF indicator being wrong, and the other being that the viewfinder and actual captured image is different.

As for the AF, I'm not really familiar with the physics behind it so I'm not sure what affects it. konraDarnok may be right saying that it is just an aberrant lens.

Manually focusing in the viewfinder and comparing focus in the captured image image for different lenses might be a good place to start (disregarding the AF indicator), though as you stated you have to keep in mind that the original focus screen can't handle the shallow DOF.

What confuses me is that one of your lenses (disregarding the AF indicator) shows that the viewfinder and CCD in agreement, and that the other lens shows it way off. That doesn't make sense to me, though it wouldn't be the first time

[firefly]

Thanks to both of you...

Yes wcleong, this is exactly what has confused me as well. I was doing a very similar test, and it seems that using only the optical path (and disregarding AF, as you say)... I get two different results between the two lenses. I'm going to run a second round of tests tonight and record in detail what's happening, to help myself find any patterns.

It had crossed my mind that any axial abberation should be caught in the optical viewfinder, which really is not problem, but if there is any radial misalignment, then it could cause divergent paths once the light hits the mirror (i.e. the mirror is presenting an inappropriate angle to the incoming light, since the light entering is slightly skewed). This could cause two different light paths (to CCD and to focus screen, at least the way I'm picturing it in my mind).

konraDarnok, what is the outlier lens? Could a misalignment there lead to this phenomenon? Could you give me a briefing of your similar story?

[firefly]

So, I got home and carried out the manual focus only tests. I documented the viewfinder image with an old point 'n shoot. Here are the results:

view of viewfinder split screen while the Tamron is attached:

File Date

View Picture EXIF

and this is the resulting photo with the Tamron at 50mm, at f/2.8:

File Date

View Picture EXIF

I notice a degree of front focus with the Tamron.

Now onto the 50 prime. This is the split screen viewfinder:

File Date

View Picture EXIF

And here is the photo at f/1.7:

File Date

View Picture EXIF

Well looky here, front focus! I guess I was too quick to panic. This is actually good news... since both lenses are now front focusing using the viewfinder only all is well with the world. I only need to modify the thickness of the shim in the viewfinder to properly calibrate the optical distance of the split screen to the mount. This should allow me to have an extremely accurate manual focus system for all lenses.

I am still left with a front focusing Tamron, but now it all seems logical, since the optics are reflecting the results I get in my photos.

[wcleong]

That's good to hear. Good luck with adjusting the shim. I have in the past used clear tape (like scotch tape) applied to the metal "U" shim that already is in the camera to make some very fine adjustments. It probably will take several layers but I found it easier than making a shim. I don't know how durable this method would be, I only had it in temporarily.