Sony 20-70G ...Bad Copy

Hi everyone, Got this new lens a couple of weeks back and have been using it on my new A7R5.

I started noticing that at 60-70mm the bottom corners were not sharp. So i setup a test chart printed out on my laser printer at A3 Size to test the lens at 70mm and what i was seeing in my photos is definitely showing up in the test chart too. So i had a look online to see if the DPreview test shots with this lens were also weak in the corners at 70mm, they are not.

Here is my setup:

Camera on a tripod, for the 70mm shots the front of the lens was approx 68cm from the test chart. I set the camera for recording large size JPG and did the following test shots

Focusing on the centre of the test chart i shot at

70mm F4, 70mm F7.1, 70mm F11

Result for these test shots is that the bottom corners are always blurry, but F4 is better than F7.1 which is better than F11

Setting focusing point on the bottom right corner of the test chart i shot at

70mm F4, 70mm F7.1, 70mm F11

Result for these test shots is that the bottom corners are even worse than focusing in the middle of the chart for the previous shots

Then as comparison shots I setup for 50mm and 35mm @ f7.1.

The were acceptably sharp (enough for me anyway)

I have included some 100% crops to show you and 1 full size image at F7.1 (which should be close to the maximum sharpness for this lens.

My Question is. Is this normal or do i have a bad copy? Has anyone else with this lens noticed this issue?















View attachment b8ea2c8f1f584373bf2cd2466d19984e.jpg

It doesn't look too bad. I guess it's slightly decentered, the top right is a bit sharper than the bottom left.

No lens is perfect especially zooms for the whole range.

I don't know if f7.1 is close to the maximal sharpness for the corners, it might be f5.6. Haven't seen comparisons of this yet.

I was comparing it to this (see link below). I have read a lot of reviews on this lens and at 70mm they all say very sharp corner to corner. If i wasn't noticing it in my photos i wouldn't care, but it is apparent which is why i setup the test chart to test it.

Sony FE 20-70mm F4 G Sample Gallery: Digital Photography Review (dpreview.com)

Interesting that at F4 it is better than F7.1

Thanks for the reply

Just adding the full size 70mm @ F4 shot for reference as well. I have never seen a lens that perform better wide open than slightly stopped down.



View attachment 3f3fea8364424e9eae5333c2def6e159.jpg

BigDaveE said:

Just adding the full size 70mm @ F4 shot for reference as well. I have never seen a lens that perform better wide open than slightly stopped down.

View attachment 3f3fea8364424e9eae5333c2def6e159.jpg

Click to expand...

Looks like this lens is sharpest wide open. In the center at least.

Yes for the centre, but not for the edges which is where my problem is, it is supposed to be sharpest between F5.6 and F8. I think i am dealing with a decentred lens from the research i have done and as suggested by the previous poster

Snip bellow from the LensTip link you provided

BigDaveE said:

Yes for the centre, but not for the edges which is where my problem is, it is supposed to be sharpest between F5.6 and F8. I think i am dealing with a decentred lens from the research i have done and as suggested by the previous poster

Snip bellow from the LensTip link you provided

Click to expand...

a decentered lens will illustrate one corner or side that is not sharp where the other IS. This is a sure fire way to detect a decentered lens. If a lens is simply NOT sharp in all it's corners say wide open (most likely at both extreme ends) then that is simply the design limitations of the lens. Over a week ago I took delivery of a Tamron 28-200 lens which I did expect to be near perfect. It became very apparent quickly that regardless of being wide open or stopped down, one side simply was NOT as sharp as the center or right side. I sent it back and received another copy. The replacement is sharp wide open right into the corners and both sides wide open at the wide AND long end and in between.

While some scoff at desiring or attempting to get a "perfect" copy of any lens, I've worked this way now for over 25 years. Prior to this decentered lens I've only had 2 other lenses of the dozens I've owned that the 1st copy wasn't perfect. So I'd get it replaced and see if the next time around doesn't yield you a better result. If as I say, BOTH corners or sides were identical then thats just what you can expect. If one is different than the other........get it replaced.

--
Name the gear and I've probably owned it and used it.

It is only on the bottom, bottom left is worse than bottom right, top left looks ok, and top right is acceptable

BigDaveE said:

It is only on the bottom, bottom left is worse than bottom right, top left looks ok, and top right is acceptable

Click to expand...

Sounds like you can certainly do better based on that. I wouldn't settle. Go for another copy and hopefully it's an improvement (which I'd be willing to bet it will)

BigDaveE said:

Result for these test shots is that the bottom corners are always blurry, but F4 is better than F7.1 which is better than F11

My Question is. Is this normal or do i have a bad copy? Has anyone else with this lens noticed this issue?

Click to expand...

Are you sure the camera/sensor is perfectly aligned to the test target?

Actually, this seems very good for a zoom lens ranging from super wide to short tele at test chard distance. Try other distances.

Do a test at a longer distance using the "Gletscherbruch test" (basically focusing in the center, then reframing so each shot has the center in a corner) and a subject such as tree branches or a building having fine details.

A nearby chart would have to be PERFECTLY perpendicular to the camera for a test to be meaningful, something very difficult to achieve in practice.

Yes, it is difficult to get right however the reason I started using the chart is that I first noticed it in pictures taken of subjects far away (about 1-2 kms).

also my pictures taken of the chart at 50mm are very good and the weirdest thing is that it is much sharper in the corners at f4 than f7.1 so if it was camera misalignment with the chart wouldn’t f4 be worse as it has a smaller DOF?

Ill try the test using some distant buildings as you suggest and see how that goes.

thanks for your reply.

BigDaveE said:

Yes for the centre, but not for the edges which is where my problem is, it is supposed to be sharpest between F5.6 and F8. I think i am dealing with a decentred lens from the research i have done and as suggested by the previous poster

Snip bellow from the LensTip link you provided

Click to expand...

That was measuring lens edge, not lens corner. Corners will be significantly worse. Besides Lenstip measured a single lens and they did not measure the corners separately. Just from what they published, their lens might as well have a corner worse than yours.

BigDaveE said:

Click to expand...

BigDaveE said:

Yes, it is difficult to get right however the reason I started using the chart is that I first noticed it in pictures taken of subjects far away (about 1-2 kms).

also my pictures taken of the chart at 50mm are very good and the weirdest thing is that it is much sharper in the corners at f4 than f7.1 so if it was camera misalignment with the chart wouldn’t f4 be worse as it has a smaller DOF?

Ill try the test using some distant buildings as you suggest and see how that goes.

thanks for your reply.

Click to expand...

That does seem to be unusual behaviour.

Try also carefully manually focusing (with enlargement)" to rule out any AF problems.

Distant' shouldn't be kilometers distant, because then you have atmospheric haze and thermal distortion adding their bit. I'd recommend about 50-200m or so ( the tree line I like to use is about 70m away). If using tree branches, watch out for wind: that's given me some false positives at times.

There are many reviews with samples on the internet of this lens performing really well at 70mm F5.6 to f8 in the corners of the frame, not just the edges.

you can search YouTube for Christopher frosts review as an example.

Mine was super soft at top left, soft bottom left, ok on the right side. Centre though, crazy sharp!

Stopping down helped a tiny bit. Manual focus also helped a bit but still soft.

When I say soft, I mean soft to the point it's smudgy.

I took a picture of a poker chip with texture on a tripod.

Not really scientific, but convinced enough that the lens isn't centred.

I'd make some tests around infinity focus. That way you can get a better idea or real world results

I use Phillip Reeve's decentering test whenever I get a new lens. It's fairly easy to do, and does help to eliminate most variables (like not having your camera perfectly aligned with your test chart):

https://phillipreeve.net/blog/how-to-check-how-decentered-your-lens-is/

WoodWorks said:

I use Phillip Reeve's decentering test whenever I get a new lens. It's fairly easy to do, and does help to eliminate most variables (like not having your camera perfectly aligned with your test chart):

https://phillipreeve.net/blog/how-to-check-how-decentered-your-lens-is/

Click to expand...

Good decentering test.

It also says:

• Zooms are more prone to decentering, compared to primes you should lower your standard a bit.

=> that's why I think what the OP showed is not bad at all.

Thanks for the link. I’ll give it a try on the weekend and report back