inksandpaper
Offline
•
Upload & Sell: Off
|
 p.3 #6 · Sensor and lens tilt and how to DIY adjust your equipment | |
Fred Miranda wrote:
I did notice a small but noticeable tightening when mounting/unmounting the lens+adapter. That's probably due to mount bend as you mentioned. I now understand how important it is to shim in pairs. Lesson learned.
Great that you noticed the tightening, that's a very good indication that the free play in bayonet CNC machining is less than the user-introduced warpage, so it is better to re-do the adjustment just to feel reassured. Well done!
BTW is there a way to see previously edited versions of a post on the forums? I thought I've seen that function before, but now I can't find it. I recall seeing different versions of your messages on this thread at least. The paper roll question is now gone, except where I've included it in a quote in my reply. Also, I seem to recall you mentioned not needing the paper rolls but only tweezers (something to do with not having coffee  ) and that it was a smooth 5-min job. Isn't it fun to take off the bayonets of these old lenses? No electronics in the way. Bayonet on and off, super quick to make any readjustments. Canon EF lenses are no fun at all. I guess you now see how it's an iterative process, especially at the start there's a bit of a learning curve. I tore my hair out a lot over the months we spent trying to figure this out. I suppose I could have peppered the article with more tips about the shimming work, but oftentimes I can't imagine what obstacles one might encounter so any advice is better crafted as the questions arise. It depends.
I've removed the shims and started over by adding 100 microns to both screws (top and bottom left) and the result is very similar. Correction for tilt was not needed. At least now, the mount is flat.
Oh, so no more tilt correction. Hmmm. Here's another tip that should be useful. You like to rotate the camera in the roll direction, not yaw, so your subject is captured at a slanted angle as you've always done in your lens tests because you want to put the subject right into the far corners (that's fine for that purpose), though I see you are not keeping the rotation consistent like in the illustration you provided (it's not important in your lens comparison tests but it is when you are working out focus swing and tilt). For example, the roll rotation in the unshimmed pair of crops for the #2 bottom left image corner is 38.5 degrees, as measured using the ruler tool in Photoshop off the guard railing of the building. It's 33 degrees (remarkably close to optimal) for the opposite #4 or top right corner. Your illustrations actually call for a consistent 33.6 degree roll rotation. And you need to compute what that skew would be with the rotation undone, with the camera upright (viewfinder pointing straight up), to be able to say 'the lens is swung'. And at right angles to that for 'tilt'. Otherwise what you are seeing is not swing, but rather some combination of swing and tilt, which makes it harder to decide how much to shim and where.
Here's a pro tip. I've very often noticed that the apparent depth of field in the corners, especially of lenses which have more optical aberrations, is seemingly greater than in the center and mid-field (mid-field can sometimes be strangely thin). And the S and T field curvatures often diverge, to add to the trouble. When you have organic detail or structural detail at a rotated angle, a wide range of field tilts combined with the depth of field can mean anything, in your case, from 80-120 microns of adjustment can still look quite balanced. You don't actually need to see into the corners for this kind of adjustment, it's not useful and blindsides one actually. I only ever look at the center portion of the image top, bottom, left and right edges. The camera is always viewfinder pointed straight up, or rotated 90 degrees anti-clockwise to the left to shoot the test scene. The test scene must have continuous detail near to far for this to work. Here's the tip: don't look only at where it's sharply in-focus, look at the near edge and far edge of the depth of field. Observe the gradient of in-focus to out-of-focus, use the blur gradient to pinpoint the true plane of focus. It is much easier this way. You might find then that you're at least 10 microns off from an optimal adjustment, because the depth of focus (not field) at f/2.8 (for any focal length) is about 23 microns, and optical aberrations in the corners increase that to perhaps ~30-40 micron's worth where it all stills looks OK.
An optimal alignment might not so important for a 200mm lens with no field curvature, but if you're looking to adjust one which has field curvature, it's much better to ensure the tilt alignment is as good as you can get it. Because then you can use compromise focus (e.g. focusing on the mid-field zone) reliably. Or you can actually focus right on the image edge and know that the other side will be perfectly balanced in focus too (even flat field lenses can benefit then). And don't worry about not looking at the corners initially, you would find they will follow and be in balance too, assuming the lens is reasonably optically centered.
Before the shimming, when I focused at center, the top/bottom left corner were OOF. When focusing at the corner, I noticed that I was focusing closer, so it made sense to add the shims there instead of the other side.
Did you mean you had to turn the lens focus ring to focus further (past the lens barrel infinity focus marker) for the left corners?
The subject was at 3.1 miles so your calculation is correct.
Wow! Lucky guess haha!
Thank you for your comments. I've learned a lot from this thread.
You're most welcome! I'm very thankful that you've so enthusiastically investigated it for yourself
|
