Consensus on 24-70/2.8 II field curvature at infinity focus?

I purchased my lens in a shop where they let me test it on the 5D3 I purchased at the same time. I was able to at least detemine that it focused quickly and quietly and no strange noises or slop in the focus or zoom ring existed.

I insisted on at least this test since the 5D2 with 24-105 arrived with the dreaded err01 issue. It had to be replaced. Since then I have shot thousands of images and after pixel peeping my landscapes and comparing them to the ones I took with the 24-105 can report that this lens is extremely sharp right across the frame at infinity focus.

I think you should try your lens on another body.

I think you should try your lens on another body.

Confirmed. I just tried the 24-70/2.8 II on my other body (1Ds mk3) and see precisely the same behaviour. Sigh. It behaved well in the shop (I specifically tested it for this known artifact), and only later did it become apparent that it just isn't as sharp all over when focusing at infinity. Lesson learned: don't test a lens in the shop, even if you can focus on something "reasonably" far away (e.g. 10-20 m). Instead, shoot something that is truly close to infinite focusing distance. Otherwise the lens will just look better than it really is.

Of course, for all those who shoot mostly at closer distances, this lens is still great.

Press Correspondent wrote:

wazu wrote:

I think you should try your lens on another body.

Yes, try putting it on Canon. Nikon D800 has a left focusing point problem

I use a Nikon lens hood on my Canon 40 mm STM

Any further opinions and experiences from landscape shooters needing good sharpness on distant subjects?

Considering the outstanding reviews that this lens has got, I would expect better than I have seen in the copies I have tried. Especially the 35-50 mm range seems to have significant curvature.

My lens has field curvature, with soft edges as a result focused at infinity.

And the “cure” is to focus beyond the infinity mark. For a start, focus using live view, aperture set at 8, place the focus point in one of the far edges, stop down and manual focus until the edge is sharp. You will by trial and error find the “magic focus point” that will bring both the center and edge/corners pin sharp.

This is what I tried to achieve too - using somewhat offset microadjustment. However, I was unsuccessful at getting a sharp center at the same time. Tweaking focus using live view across the frame is of course possible, but kind of annoying.

Another copy I tried was sharp as a knife both in the center and at the right edge, but mushy on the entire left side... There is just a lot of variation going on between copies. The excellent AF of the 1D X just makes any such inconsistencies even more obvious.

It seems to me that the 24-70 L II is not really such a good landscape lens after all.

Benedict Slotte wrote:

My lens has field curvature, with soft edges as a result focused at infinity.

And the “cure” is to focus beyond the infinity mark. For a start, focus using live view, aperture set at 8, place the focus point in one of the far edges, stop down and manual focus until the edge is sharp. You will by trial and error find the “magic focus point” that will bring both the center and edge/corners pin sharp.

This is what I tried to achieve too - using somewhat offset microadjustment. However, I was unsuccessful at getting a sharp center at the same time. Tweaking focus using live view across the frame is of course possible, but kind of annoying.

Another copy I tried was sharp as a knife both in the center and at the right edge, but mushy on the entire left side... There is just a lot of variation going on between copies. The excellent AF of the 1D X just makes any such inconsistencies even more obvious.

It seems to me that the 24-70 L II is not really such a good landscape lens after all.

I suspect you are right and I rather doubt Canon did design this lens with landscape photographers primarily in mind. I know that the original brief for the original 24-70/2.8L and its 28-70/2.8 predecessor was to produce a fast zoom for photojournalist use and I doubt that brief has changed significantly with the MkII. As such some field curvature at infinity would probably be considered a reasonable trade off against other performance characteristics that a photojournalist might consider more important.

I'm certainly not implying this is the case with you but I fear some photographers do buy lenses because they are the fastest or most expensive, rather than with a realistic assessment of what they are actually going to use the lens for, and so can end up making expensive mistakes, perhaps like this. It would be nice if a manufacturer could list all the things a lens is not so good at but naive and unrealistic to expect that of course.  In the meantime we rely on test reviews and hope the reviewer is able to cover the aspect we are particularly interested in.

Benedict Slotte wrote:

I have tried a few copies of this lens, and settled on the one that was best among them.

Benedict, thanks for binging this up. I'm in a similar situation as you are. I always do buy several samples of a zoom lense and pick the best. I have seen great sample variation with tamron (as expected). With the Nikkor 80-200 2.8 there was non at all ( 3 samples), and the Nikkor 18-35 2.8 had extensive variation also. As i'm just switching platform, i have no experience with canon (actually i have, but thats 20 years back). Can you desribe how the samples differ and did you test several focal lengths or just one? I expect corner resolution to vary mostly, right?

In spite of this, I do see significant field curvature in landscape shots (read: focused close to infinity) even at f/8, meaning that not even at f/8 will it be possible to have decent sharpness from one edge to the other.

To my best knowledge field curvature is not really subjected to sample variation (lense alignment), but is a design limitation.

The only thing you can do about it is to harmonically optimize diffraction and hyperfocal distance.

-Armin

I was beginning to think I was the only one in the world who had noticed it.

Micro adjustment helped mine a little because not only was the focus way too far forward at the edges it was a little too far forward in the middle. Its still obvious though. For my usage and priorities the lens works well but if I was a landscaper I'd say it was a good only as a 24mm prime lens. The lack of CA is impressive. Is there a better single lens alternative?

I feel lens designers are prioritising sharpness and CA correction over less obvious issues like field flatness and coma (24mm f/1.4 L II). If lens designs like the 35L disappear in favour of a higher resolution lower CA mkII it will be very sad if that destroys the beautiful way it renders a scene.

Andrew

Hi.

The 24-70L II is superb at 24mm. Probably why it reviews so well. I was pretty stunned by it. The field curvature is less that at longer focal lengths and the DOF helps hide it. As the OP has said it can do really weird things in the middle of the focus range and mid-long distances and getting an infinity scene sharp at the edges at 70mm is very difficult.

The 40mm pancake is sharper at the sides of the frame at the same apertures. If I was a landscaper I'd put the 24-70L II money into some primes without the field curvature issue.

Do you mean 24L has field flatness and come or does not have coma or...? Could you clarify?

The 24L II has coma issues, the samyang 24/1.4 is a better lens for astro photography. The 24L II is a lot sharper than the 24L. They are getting one thing better at the expensive of something else. Most people will like that direction but if you have special uses, want flat field, want low coma, then you'll need a very good reviewer to tell you. These things aren't noticed by most reviewers.

If lens designs like the 35L disappear in favour of a higher resolution lower CA mkII it will be very sad if that destroys the beautiful way it renders a scene.

Could you please elaborate on the properties of 35L you touched upon? Also, do you have an opinion on how Sigma 35/1.4 compares to it in this regard?

35L makes beautiful images, it does have problems with CA on high contrast edges wide open. I hate that when it occurs and process it away but I'll accept it if the alternative is a highly corrected lens with a less pretty image overall. Sigma 35mm looks to the the lens if you want extra sharpness at the expense of less smooth bokeh. Not everyone will care, some people think the 50/1.4 has good bokeh and I think its awful. The Sigma 50mm looks better in that regard.

Cheers,
Andrew

That's not how it shows itself, 24mm usually is no/less problem.

http://www.photoluminaire.co.uk/posted/6B9C4262.jpg

Explain to me why the center of the far middle tree is sharp and yet the base of it isn't.
Then look at how sharp the corners are of the foreground tree that were just the other side of a fence from me. No other lens I have does this, the old 24-70 actually curved the other way!

Luckily it doesn't affect my intended use, I tested for field curvature at a few meters and its minimal.

Andrew

Which lens is better?

Primes (but that's self-evident). I don't have the old 24-70 anymore, so I can't make a good comparison. I don't care if there is visible curvature at f/2.8, but by f/5.6 and f/8 it would be nice to get rid of it by absorbing it completely into the DOF.

I would guess that the 24-70 II is not necessarily that much "better" than the original, but rather just "different", when doing landscapes.

I suspect you are right and I rather doubt Canon did design this lens with landscape photographers primarily in mind. I know that the original brief for the original 24-70/2.8L and its 28-70/2.8 predecessor was to produce a fast zoom for photojournalist use and I doubt that brief has changed significantly with the MkII. As such some field curvature at infinity would probably be considered a reasonable trade off against other performance characteristics that a photojournalist might consider more important.

Yes, that's what I start to think as well. However, the major increase of field curvature when focusing at infinity was a surprise.

I'm certainly not implying this is the case with you but I fear some photographers do buy lenses because they are the fastest or most expensive, rather than with a realistic assessment of what they are actually going to use the lens for, and so can end up making expensive mistakes, perhaps like this. It would be nice if a manufacturer could list all the things a lens is not so good at but naive and unrealistic to expect that of course. In the meantime we rely on test reviews and hope the reviewer is able to cover the aspect we are particularly interested in.

I upgraded partly because of the much more convenient size of this lens. The design of the old one was a bit annoying in comparison. Also, the sharpness seen in reviews was just as important as a factor, but resolution tests shot at a few metres obviously say *even less* about curvature at infinity than I would have expected. Note to self: next time, test before buying at genuine infinity rather than e.g. 10 m.

Yet another technical hypothesis I made was the following: since phase detection AF relies on light passing close to the edges of the lens, chromatic aberration will supposedly make AF ambiguous, meaning that the AF system can front-focus or rear-focus depending on the colour of the subject. A lens having less CA (read: 24-70 II compared to the original) should, technically, also have more consistent and accurate AF. I noticed this also with the 24 f/1.4 II vs the original version. Less CA, more consistent AF.

More consistent AF in itself can be reason enough to upgrade. I wasn't happy with the inconsistency of the original 24-70. I don't care about CA otherwise, since it is easily processed away in Lightroom, but I do suppose it matters for AF consistency.