p.3 #3 · FM Lens test for Centering (Mount tilt / swing)
s4ducati wrote:
Indeed the same test as FM has described and used here.
An interesting note however that I have note seen referenced to is the idea that the sensor may be mis-aligned too.
I wonder how often the sensor alignment is out of spec in higher end cameras?
Yes, sensor misalignment happens (or mount misaligned, depends on how you define things). I remember some stories about a recent Nikon DSLR camera where this was noticed in early production runs - probably due to the higher resolution which makes such things more obvious. But if you have just one lens and body it is difficult to say which one is the culprit, and even if you have several lenses it may not be entirely clear ;(
p.3 #4 · FM Lens test for Centering (Mount tilt / swing)
s4ducati wrote:
Indeed the same test as FM has described and used here.
An interesting note however that I have note seen referenced to is the idea that the sensor may be mis-aligned too.
I wonder how often the sensor alignment is out of spec in higher end cameras?
technic wrote:
Yes, sensor misalignment happens (or mount misaligned, depends on how you define things). I remember some stories about a recent Nikon DSLR camera where this was noticed in early production runs - probably due to the higher resolution which makes such things more obvious. But if you have just one lens and body it is difficult to say which one is the culprit, and even if you have several lenses it may not be entirely clear ;(
We suspected that my original A7r sensor was out of alignment when testing a number of lenses including my Zeiss 35mm f1.4 ZM and other lenses at or near wide open (when using 2 adapters) after the camera had a Kolari thin sensor stack modification (apparently the camera was originally out of alignment). All of the lenses tested showed the same corners being out of focus when focusing on the center as well as on the corners. This was confirmed and corrected by Kolari when they changed the V2 sensor cover to a V3 sensor cover. Kolari spent the time and aligned the sensor, calibrated the focus of the camera, and confirmed that both my original Phigment and Hawk's Factory Helicoid V5 Leica M to NEX adapters were fine at no additional charge.
p.3 #5 · FM Lens test for Centering (Mount tilt / swing)
I wonder if these newer autofucus lenses with the focus element or group riding on rails have difficulty keeping the focus group aligned? Seems like it would be difficult to avoid some play. I have tested a few lenses and gotten inconsistent results. My Zony 35 1.4 tested great at 1.8 and not so great at f/4 and f/5.6
p.3 #6 · FM Lens test for Centering (Mount tilt / swing)
It’s the first thing I test when getting a new body. I usually use a lens that I know is perfectly centered. I have seen bodies with tilted mount/sensor registration.
s4ducati wrote:
Indeed the same test as FM has described and used here.
An interesting note however that I have note seen referenced to is the idea that the sensor may be mis-aligned too.
I wonder how often the sensor alignment is out of spec in higher end cameras?
p.3 #7 · FM Lens test for Centering (Mount tilt / swing)
AZSteve wrote:
Fred, since this thread has been revived I want to mention that I recently had an experience with your centering test methodology versus Jim Kasson's Siemens-star test, where a lens that appears to pass the latter fails the former. (Your post of 1 Sept.)
B&H had a one-day deep sale on the Sigma Art 85mm and I took the bait. The first four images I took were Siemens stars in each corner, and then I went on to images of nearby mountains, a slight variation on your procedure. The stars looked passably similar and I was much relieved that I'd been sent a good lens. Then I looked carefully at the mountain images and discovered that the left edge was substantially softer than the right, and that the difference was still visible stopped down as far as f/5.6. I retested on the mountains and got the same thing. The lens would have been optically fine for portraits, sports, "street", etc, but not for landscapes. Back it went and I didn't try another.
Maybe it is easy to miss small differences in the imaging of the Siemens stars, I don't know.
p.3 #8 · FM Lens test for Centering (Mount tilt / swing)
David, Kasson's star test is to be conducted at what we might call "effective infinity", which he has an involved way of characterizing based on f-ratio, focal length, pixel size, etc.
p.3 #9 · FM Lens test for Centering (Mount tilt / swing)
naturephoto1 wrote:
We suspected that my original A7r sensor was out of alignment when testing a number of lenses including my Zeiss 35mm f1.4 ZM and other lenses at or near wide open (when using 2 adapters) after the camera had a Kolari thin sensor stack modification (apparently the camera was originally out of alignment). All of the lenses tested showed the same corners being out of focus when focusing on the center as well as on the corners. This was confirmed and corrected by Kolari when they changed the V2 sensor cover to a V3 sensor cover. Kolari spent the time and aligned the sensor, calibrated the focus of the camera, and confirmed that both my original Phigment and Hawk's Factory Helicoid V5 Leica M to NEX adapters were fine at no additional charge.
good to hear that Kolari spent the time for correcting that and did a good job
In most camera's this is a lot of work and often it is impossible (at least without dedicated factory equipment) to align the sensor exactly with the camera partly disassembled to access the sensor alignment parts (usually some small screws/clamps/shims etc.). Using trial and error often means more risk of sensor dust etc. ...
p.3 #10 · FM Lens test for Centering (Mount tilt / swing)
I guess I am old and not very smart. could someone post 4 photos taken with a decentered lens and tell us how you know that.
anyway I took some photos with my samyang 35 f2.8 and with my batis 135 f2.8 and the same object I focused on look about the same amount of sharpness (use 1/400 sec iso 400 overcast and kinda rainy).
these are the 4 corners from each shot with the 135, raw 7rm3, crop then export jpg full sized
**sorry -- I think these are 32meg files***
OK now just 4 meg files (full size for cropped portion). I did mention my problems above.
p.3 #11 · FM Lens test for Centering (Mount tilt / swing)
AZSteve wrote:
David, Kasson's star test is to be conducted at what we might call "effective infinity", which he has an involved way of characterizing based on f-ratio, focal length, pixel size, etc.
Hi Steve.
Ok. but a lens can still be different cantered or decentered depending on the focus distance ;-)
If the lens elements rotate during focusing the unsharp area will change place too.
p.3 #12 · FM Lens test for Centering (Mount tilt / swing)
David Cartagena wrote:
Hi Steve.
Ok. but a lens can still be different cantered or decentered depending on the focus distance ;-)
If the lens elements rotate during focusing the unsharp area will change place too.
David
That's definitely true but from my experience testing "prime" lenses, if it's well-centered at infinity, it's also great at other distances too.
p.3 #13 · FM Lens test for Centering (Mount tilt / swing)
Fred Miranda wrote:
That's definitely true but from my experience testing "prime" lenses, if it's well-centered at infinity, it's also great at other distances too.
OK. I Understand. My Sony 55 f/.8 are sharper in the corners at short distances than at infinity. So it's definitely not getting worse at shorter distances;-)
Prepare yourself to be disappointed as unfortunately it's uncommon for a lens to be equal at all four corners, especially when testing it on a high MP sensor. My copy of the Loxia 21/2.8 used as sample here is one of them...but I went throught 3 copies.
IMO, testing at infinity is easier and preferred as the alignment variable is removed. Trustworthy MTF charts like LensRentals and Zeiss are tested at infinity distance.
Of course there are lenses with low variance but I think it's more related to optical design complexity than tolerance although the latter plays a role. That's even for primes and from any brand. I'd say it's pretty unlikely a zoom lens is perfectly centered at all focal lengths.
p.3 #17 · FM Lens test for Centering (Mount tilt / swing)
Fred Miranda wrote:
Prepare yourself to be disappointed as unfortunately it's uncommon for a lens to be equal at all four corners, especially when testing it on a high MP sensor. My copy of the Loxia 21/2.8 used as sample here is one of them...but I went throught 3 copies.
IMO, testing at infinity is easier and preferred as the alignment variable is removed. Trustworthy MTF charts like LensRentals and Zeiss are tested at infinity distance.
Of course there are lenses with low variance but I think it's more related to optical design complexity than tolerance although the latter plays a role. That's even for primes and from any brand. I'd say it's pretty unlikely a zoom lens is perfectly centered at all focal lengths....Show more →
Lucky (if want to call it that) my 99II allows all 4 corner micro adjust...it would have to be really bad for me to go in and adjust a zoom wide then tele and then the 4 corners at each distance.
p.3 #19 · FM Lens test for Centering (Mount tilt / swing)
Fred Miranda wrote:
Hi Roger,
What's the variance frequency for primes?
It depends greatly on which prime and how sensitive we look. Our standard is <10% variation at 30 lp/mm is perfect (I really careful pixel peeper with a high resolution camera might be able to see that, just barely, in some shots. But they would have to work hard to do it.)
At that standard there are a few primes where none are perfect. A really good prime would be over 90% perfect rate though, and most get to 95%.
There are good zooms where 95% are perfect at one focal length but never seen one perfect at all 3 we test (both extremes and the mid point). There's just too many things moving.
p.3 #20 · FM Lens test for Centering (Mount tilt / swing)
Fred Miranda wrote:
Hi Roger,
What's the variance frequency for primes?
It depends greatly on which prime and how sensitive we look. Our standard is <10% variation at 30 lp/mm is perfect (I really careful pixel peeper with a high resolution camera might be able to see that, just barely, in some shots. But they would have to work hard to do it.)
At that standard there are a few primes where none are perfect. A really good prime would be over 90% perfect rate though, and most get to 95%.
There are good zooms where 95% are perfect at one focal length but never seen one perfect at all 3 we test (both extremes and the mid point). There's just too many things moving.
