Pentax 67 Viewfinder Geometric Accuracy

[yessammassey]

I have a Pentax 67, and a Pentax 6x7. The 67 had its focusing screen replaced long ago, before I started using it. Architecture is a frequent subject for me, and looking back at the several rolls I've shot on the 67, there seems to be a recurring theme where it looks like I've got everything squared up in the viewfinder, only to have crooked lines in the final images. E.g. the general trend for horizontal lines on the tops of buildings is that they appear parallel to the top of the frame in my VF, but come out looking like this:

Questioning if it was just a case of technique, I set my 67 up on my tripod, pointed it at a rectangular building, squared up all the lines, then popped it off and replaced it with my 6x7, transferring over the prism and lens from the 67, and making sure it sat exactly the same on the tripod. The view through the 6x7's finder seemed to confirm that something was amiss with the 67, as the building now appeared crooked, with lines no longer parallel to the VF edges, and the roof line sloping downward to the right just like it did in so many of my negatives. It was subtle, but I switched out the cameras several more times and always found the same discrepancy in perspective: the camera with original focusing screen disagreeing with the camera whose focusing screen was replaced.
The mirror and prism seem to be in perfect condition and alignment on both cameras, and considering that the viewfinder had been replaced, that's what I suspect is the problem. Pentax 67 viewfinders are not supposed to be user-replaceable, maybe it wasn't properly aligned when replaced. I took a look. After removing the prism and protective plate, I'm met with a bracket with three screw adjustments:

It seems like there's not much here that could throw the view out of whack. The focusing screen looks to be perfectly level. The screws hold it snugly. Focus is accurate; always has been. Feeling bold, I turned the right screw 180 degrees clockwise. It seemed to have no effect on the image on the screen at all. Turned another 180, nothing. I reset by turning it 360 degrees counterclockwise, and that's where I left it. I took it to a small-time local camera repair shop and explained my issue. The technician made a cursory examination of the focusing screen and said they couldn't see any problem with it. I saved two frames on my next roll for head-on shots of buildings and found the same slightly askew lines on the negatives, as before.

Considering that I can see a difference between the two cameras' views, I think the issue must be in the optical path to the VF, not the film back.

Does anyone with decent technical knowledge of these cameras (something I am obviously lacking) know why I might have this issue? Is there a repair service that might be able to diagnose and fix this problem? Is there any chance that I might be able to fix this myself by adjusting the focusing screen until I have agreement between my problematic 67 and 'known good' 6x7?

 

[John Koehrer]

Is bottom of the mirror square in it's mounting frame? foam under the mirror?

 

[yessammassey]

Is bottom of the mirror square in it's mounting frame? foam under the mirror?

It looks to my eyes to be square. That is, the silvered plate is square in its metal holder, and the bottom of the holder is square with the lightbox. The foam looks good, but I'm wondering... is the mirror the more likely cause here?

Had a spirit level taped to the wall and the camera on a tripod head with spirit levels as well. Made sure the camera was level (for 'pitch' and 'roll' - to use aeronautical terms) and aligned with the level on the wall. But it still looked skewed, like the camera wasn't facing the wall dead-on. Problem is, no matter how I move the camera's 'yaw' on the tripod, I can't get the spirit level on the wall to show horizontal alignment in the VF while also keeping the camera's 'roll' level. Frustrated, I started going a little crazy with the focusing screen adjustment screws... making it tilt to the left or right, with no real effective difference in how the level taped to the wall appeared in the VF.

Infinity focus is now way off thanks to the screwing around. I'm thinking I'll just pop the whole focusing screen out and see if the springs underneath are OK.

Whatever the problem is, I don't think the adjustment screws on the focusing screen are going to be the answer.

 

[yessammassey]

I took the focusing screen all the way out, made sure the spring was in good shape, and put it all back, making sure to get everything aligned as exactly centered and squared possible. Put it all back together and shot a test roll. Same issue. Yes, I think the problem isn’t the focusing screen. If the issue was caused by imprecise focusing screen placement, I doubt I would’ve been able to duplicate it exactly with a fresh install. There’s a lot of wiggle room in the focusing screen bay; no way I placed it exactly to the mm as it was before.

Wishing I’d checked the mask that sits below the screen while I had it out, though.

Still scratching my head over what it could be.

 

[yessammassey]

Another update on this. I got my hands on another Pentax 67, in apparently great shape (just like mine), took it out to a wall and went thorough a roll, just keeping the camera as square to the wall as I could. Didn't have time to set up a tripod. Developed frames show the same thing, a sloping to the right.

This suggests to me (especially if I was reading this instead of writing it) that there's something off in the user's perspective, and it's not a fault with the camera. The 67's viewfinder is huge and it is hard to see the frame edges even with the user's face smashed all the way up against the eyepiece. This could introduce a tendency to error, but the sloping to the right is weirdly consistent.

I think I need to check again with my 6x7, actually run some film through it too and see if it really doesn't show the same effect... but in my experience it was never a problem before. Maybe something new is wrong with my eyes haha!

But if there is a physical, camera-based reason for the discrepancy between viewed-image and taken-image, here are my best guesses as to the cause:

- Mirror tilted. There is a mirror adjustment screw (or maybe two, it's not crystal clear in the repair manual) so obviously a tilted mirror can occur. It's also kind of odd that the 67 only has a mirror bumper foam patch on one side of the upper lightbox; maybe the mirror can start to tilt as a result after many thousands of cycles. Mirror looks fine, though.

- Focusing screen distortion. Maybe not the focusing screen alignment, but flaws in the acrylic molding. I think it's less likely than the mirror, but possibly something that happens often enough to be seen in two different copies of the same part.

- I don't think it's a viewfinder mask issue. The mask is rigid and part of the light box. Plus, to say that the images tend to show sloping to the right is a simplification; the whole alignment between camera and subject plane also appears tilted a few degrees 'to the left' of perfectly orthogonal in the image, while appearing dead-on in the viewfinder.

 

[grain elevator]

I think any misalignment of mirror or focusing screen will have much more severe effect on focus than on what you call geometry, if any.
Could it be the film? If it sits at an angle... like the problem that is known in screwmount Leicas. I can't imagine it could be angled by much, but you haven't told us how large the error is...
You can rule it out by checking the projected image at the film plane with a piece of ground glass, or scotch tape. If it differs form the VF image, there's a VF issue.

 

[yessammassey]

As far as I can tell from switching back and forth from viewing the image behind the lens with ground glass and looking through the focusing screen (straight on and through the prism), the problem lies in the viewfinder. There’s no discrepancy in focus between the two, but there is a difference in the appearance of straight lines.

I can't really come up with a value to quantify the degree of perspective error, but here is an example. (Sorry to bring scans into the analog forum, but just to illustrate an issue that one can easily seen in the negatives in real life. This is a problem that's way harder to satisfactorily correct for with optical printing... while it's simple with Photoshop. This is an analog issue.) Each one of these compositions was lined up perfectly level in the viewfinder, but as you can see, the vertical and horizontal lines are skewed in the final image.

Going on the idea that what I’m seeing is not the result of a tilted mirror (and the mirror looks as straight as any other 67 I’ve examined), and from my experience that it can’t be corrected by adjusting the seating of the focusing screen, I placed an order for a new focusing screen. My best guess is that the focusing screen currently in the camera is affected by some kind of distortion.

Although one other thing that’s kind of odd is that if I compare the viewfinder image in my Pentax 6x7 to that of the 67 in question, the entire frame is offset downwards. I checked this by replacing the bodies on a tripod, same lens, same quick release plate. Not a very well-controlled comparison, but doing it repeatedly always gave the same result. The 6x7’s view is as if the tripod head had been lowered a few inches compared to what the 67 sees. It’s like this on all three of the tripods I own. That seems more like something that would be the result of a difference in mirror angle, but both cameras have perfectly accurate focus even wide open with the 90/2.8.

 

[yessammassey]

Update: preliminary results appear to show that the new focusing screen has fixed the problem. Shot through an old roll of TMAX 100 and developed immediately; eight of the ten frames were spent on head-on walls, storefronts, etc., just to test it out. What looked square in the VF mostly came out square on the negative. There was even one where I misjudged or held the camera just a little wrong and the horizontal lines were tilted the other way. I never had a single shot that came out where horizontal lines were accidentally tilted up on the right with the old screen.

But there is one odd thing about the new focusing screen. It renders out of focus areas quite a bit differently than any other 67/6x7 that I've used. Essentially, there's 'bloom' around unfocused areas, and the fresnel itself looks more 'textured' than the old screen when inspected with a strong light. After installation, I thought the reason for this could be that I put it in upside-down, but I flipped it over and found the same effect (necessitating flipping it back to the correct orientation once more).

My description doesn't make it clear, and these pictures only work marginally better. It's hard to get good pictures of the viewfinder image, but hopefully you can see what I mean. Unfocused objects retain their hard edges in areas of contrast, while they simply melt away in every other focusing screen I've used. Compare the features near the corresponding colored dots on the matching pictures. Any ideas why it's like this? It seems work just fine, but I wonder if there's a functional impact it could have that I haven't considered or encountered yet.