To my eyes, the biggest difference is between the upper right and the lower right. If this were curvature, should they not look identical?
Yes this looks much worse than the previous test. How is it at 24mm?Did quite a lot of tests today, and as the sunlight faded and air cooled I could do more distant subjects on the horizon without hot atmospheric air in the way. I think the bad upper right corner is pretty obvious here:
100% crops:
ISO100 1/2000s f2.8 70mm
By the way, doing tests at different focal lengths is quite confusing with this lens. While by far the worst performance is at 70mm in the upper right corner, at 50mm both the upper and lower right are soft while the left side is sharp (at 70mm lower right is excellent, as you can see). And at 35mm all corners except for the lower left are good. I don't see the logic in such varying performance across the focal range, it'd have expected the upper right to be the worst across the range but clear up a bit as I zoom out. Perhaps someone more knowledgeable can offer an explanation for this...
www.facebook.com
There is a some decentering but if you read his article the amount you see isn’t that unusual, it’s not perfect not close but also not outrageously bad. He shows good example of near perfect and acceptable and when it’s bad, from what I see yours isn’t outrageous bad but not great either. it’s not perfect but that’s really also rare in zooms, all lenses have decentering to some extent and your never truly gonna get 4 equal corners especially not in a zoom and especially not in a 24-70mm one these should be very hard to make cantered… if you send it in they most likely send it back and say it’s within specs, or that would be my concern.Did the test in line with the method recommended by phillipreeve.net. Here's the result at f2.8:
ISO100 1/250s f2.8 70mm
And at f8:
ISO100 1/125s f8 70mm
All 100% crops, view at original size.
--
It’s not that simple though. You have to assess if those corners are oppositely front or back focused. If they’re all slightly back focused (like in this case) while the centre is in focus, then it’s field curvature. If some are front focused and others back focused, it’s tilt.
Roger has written a lot of articles about field variations, including curvature, tilt and decentering. He also has done some real measurements on copy variation.
Also note that zoom lenses vary over entire zoom range. And every single GM 24-70 zoom lens behaves slightly different because mass produced optical and mechanical precision is within limits.
Sometimes we should just stop looking for slightly imperfections because with that mindset we are never gonna be happy. I tend to call it the pixel peep disease.
Tilt or decentered in these discussions (while technically correct to distinguish) is not so important for the buyer, cause either way it is a lens flaw and lens should be returned or repaired, if it bothers.
Per johnNEX's post above I looked at the method of a decentering test by phillireeve and found it interesting.
Here is the morning test of a distant object (severe crop of a chimney) focusing in the center of the frame and then imaging the corners (using the phillipreeve method). Tested at f/2.8 and a second for curiosity at f/5.6. Sony A1, ISO 50, on a tripod, electronic shutter. My take, as long as I was following the method as described, the lens is soft in the corners at f/2.8 (expected) but also appears to be slightly decentered using the example at f/5.6. The question and statement posed on "phillipreeve" is all lenses are decentered but is the user willing to live with it?
At this point I would need to see another copy and do a similar test using the same method. Not willing unless my local store will lend me one.
I realize there may be errors here and I do not at all say my testing is ironclad valid. However I was consistent in the 2 sets of images.
Comments Encouraged.
Jim
Decenter Test f/2.8
decenter test f/5.6 Center Image is the same f/2.8 as above.
Thanks for the suggestion. Time to respond to comments in the near future is limited so I thought posting in this particular thread may be of value. Wish I had more time to test. The bottom line for me, sans posting in the other forums, is to obtain another sample and test when I do have the time. If I find the other copy is significantly better I will send mine to Sony for evaluation and possible repair/replacement. It is still under the warranty period and I am not in short supply of quality GM lenses in its stead.
I would say that’s perfectly acceptable, again generally I would not do these test unless I notice something really wrong in normal shooting. Because the only thing it leads too is that once you know it’s there it cannot be unseen, sometimes ignorance is a bliss.Per johnNEX's post above I looked at the method of a decentering test by phillireeve and found it interesting.
Here is the morning test of a distant object (severe crop of a chimney) focusing in the center of the frame and then imaging the corners (using the phillipreeve method). Tested at f/2.8 and a second for curiosity at f/5.6. Sony A1, ISO 50, on a tripod, electronic shutter. My take, as long as I was following the method as described, the lens is soft in the corners at f/2.8 (expected) but also appears to be slightly decentered using the example at f/5.6. The question and statement posed on "phillipreeve" is all lenses are decentered but is the user willing to live with it?
At this point I would need to see another copy and do a similar test using the same method. Not willing unless my local store will lend me one.
I realize there may be errors here and I do not at all say my testing is ironclad valid. However I was consistent in the 2 sets of images.
Comments Encouraged.
Jim
Decenter Test f/2.8
decenter test f/5.6 Center Image is the same f/2.8 as above.
I assume you were responding to me. In normal shooting I am satisfied. If I come across another copy I will borrow/test it. Not on my "urgent" list.Out of curiosity , I just did the same test with my Tamron at 75mm f/2.8 and f/5.6. I can't see any difference between the 4 corners.
I usually don't do this test unless I suspect an issue in my pictures. I guess all/most lenses have some degree of imperfection. I would expect a good quality control of the GM lens though.
Your de-centering issue doesn't seem severe. But it's a very expensive lens and there shouldn't be any compromise. Since the return window is closed, just try to enjoy the lens and not to focus too much on the top right corner at 70mm
In normal shooting with the new beast I have found little to be concerned about. The few times I peruse this forum certain threads pique my interest. This morning was one of those times so I tested the lens. Luckily it didn't take too much time. But now curiosity has entered in and I will probably test my 1st generation 24-70GM as well as other lenses. But not so much interest to obsess.
Thanks. I think it does as well. I plan on keeping it and if I can get another for testing I will. And I will report back to the forum in a new thread since it appears I may have hijacked this one a bit.
When the first version of this lens was current, I ended up with one where the plane of focus was kind of substantially tilted. I sent it to Sony under warranty for evaluation and they concluded that it was not within spec and returned it repaired. I think repair for these means replacing the whole optical block inside, but at any rate the difference afterward was very obvious and unmistakeable and it's been totally satisfactory since.
blog.kasson.com
