So playing around with my new Sigma lens I found this really strange behavior:
(Shot at 26 mm, F/4.5, focusing about 100 m away at a building, all objects in the scene about 50 to 300 m from me).
I can’t post any of these photos, the scene I’m testing on right now, even if heavily cropped would reveal the location of my home.
Focusing through viewfinder AF results in a blurry right side of the photo and overall rather low sharpness. This is most noticeable when using the center point, but it happens for some other as well, when I focus on the same spot and readjust the frame so all photos are roughly the same.
However, if I focus through live view, the problem is gone. The focusing is really slow, but it is correct. The photo is as sharp as I expect it to be everywhere.
Note: the camera’s viewfinder AF system should be ok, other Sigma 105 mm does not have this issue and on old 18-55 mm the difference is minimal.
Now where’s the problem? Is my camera incompatible with the lens if the viewfinder AF is so unreliable? Or is it just a faulty lens?
Well, 100 m should be infinity, as indicated on the lens. No adapter is used. The lens should be compatible with DX Nikon DSLRs (e.g. my D3500), as stated on MPB.
I’ve got it for 2 months and only noticed now. But why does it not happen with live view? It doesn’t make sense to me why the viewfinder AF focuses wrong.
This is strange. I’m not very well versed in optics but couldn’t your mirror be off? It would explain the live view being good but viewfinder being off. I can’t explain why the same issue wouldn’t appear with other lenses though…
Well, manually focusing through optical viewfinder is really tricky.
If I get it right, the image is ok. But focusing slightly nearer the image looks blurry on the right side, more than on left – again, suggesting asymmetric field curvature.
It seems like the focus is wrongly measured with AF using the viewfinder. But I’m not having issues with other lenses though, not that I noticed something obvious so far.
True, but their numbers and description doesn’t match my (in my opinion) rather intense asymmetry and blur when slightly out of focus.
EDIT:
It’s more got to do with actually not being properly focused, as when I actually focus properly, the issues are much less noticeable, but still present. not too bad, but yeah…
I had a similar issue with a sigma 30mm lens (which apparently is well known for its inaccurate AF) on a D5100, that af through the view finder was always off a bit. In my case, I managed to mostly fix it by calibrating with the usb dock. Not sure if there is any alternative way, if your lens is not supported :(.
As to why it miss-focuses with the view finder but not in life view: I think these two modes use completely different focusing methods. With the view finder, phase detection AF is used, which, to my understanding, is an open loop one shot (so no loop actually) process, i.e. the sensor tells the lens to correct focus by amount x and then takes the shot without further checking if the lens did the right thing. What you can calibrate through the usb dock, is the response of the lens to that command from the camera.
In live view, contrast based focusing is done, which is an iterative closed loop process, i.e. the camera adjusts focus a bit, then checks again, re-adjusts, etc. until focus is good. So here, the camera can correct if the initial response of the lens is not correct.
At least, that is my understanding, please correct me, if anything wrong.
Right, that might explain why it takes soo long.
Thanks for your input, I appreciate it. I can theoretically ask repair shops I know about, like what can be done… or contact MPB if all else fails.
So, your sample of one doesn’t match Imatest’s sample of one … no surprise, statistically speaking and my 17-50mm could easily be different to both of y’alls.
Seems different, idk how is the graph oriented but my lens suffers right asymmetry, rather than on the left (as i said, might be upside down), but the 28 mm F/4 seems to be pretty bad in my case too.
Stopping down to F/5.6 helps though. If I solve the focus problem, I’ll just have to watch for the lens behavior… Thanks for the resource. Very intuitive and interactive graph viewer is a great thing to have at hand.
I would recommend downloading a depth of field app for your phone. The hyperfocal distance for a 26 mm lens mounted on your camera is just under 8m. So if you had focused on an object about 10 m away then everything from about 4.5 m to infinity would look in focus. This may hopefully also assist if there is asymetric field curvature, but you would need to confirm that by doing some tests.
Also the live view focus and the viewfinder focus are very different in how they focus and in your case the live view is giving better results. @luator described this correctly.
I woud recommend try taking these shots again and instead of focusing at 100M try a range of closer distances and see if the depth of field benefit overcomes the focusing issue with this lens. Only testing will see if this is the case or not.
On a side note I have a Nikon 18-300mm f/3.5-6.3G ED VR lens and I have noticed some softness to the focus on the left side. I presumed I had damaged the lenses during my many travels, but I am now wondering if I am just picking up asymetric field curvature. So I might look into this issue and I too might see how live view compares to the viewfinder. So overall an interesting thread to read.
