[stoppingdown]

The other strange thing is that I don't see that levels of CA in my photos. Of course, 99% of times I have Lightroom correction automatically kicked in, but when I ran some test shots just after buying the lens I disabled it and pixel-peeped. Actually I ran the test shots with the NEX-6 (the a6000 came later), so less megapixels make CA less evident. I'm going to look at it again.

[Klaus]

I think the lens is "reasonably" well centered so Sony Australia may just take that as a reference.

 

Honestly, astigmatism alone doesn't explain those borders.

[Rover]

Ouch. It's been a while since you churned out a 1,5* review (though there have been quite some on Markus's side, like the Sigma 20mm f/1.8)

On a side note, I wonder how well the 16-35 FE fares on APS-C? Is there going to be a test of that?

[Brightcolours]

I think the lens is "reasonably" well centered so Sony Australia may just take that as a reference.

 

Honestly, astigmatism alone doesn't explain those borders.

Well, a few test images will tell a lot. 

Consider this lens:



In your and other MTF based reviews it does pretty badly in the corners, so much so that the lens has a bad reputation. But in real life usage, its results are pretty good (obviously not the best lens in the world at this focal length).

Shooting at MFD reveals very strong astigmatism, which explains the poor border/corner results in MTF tests.

 

Here an example past MFD (with 12mm ext. tube) to show the problem very clearly:



 

And here the same lens at a more normal (for the lens type) focus distance:

Quite nice performance of this little pancake lens.

 

I am not saying that astigmatism IS the big bad bug in the MTF results for this Sont 16-70mm f4, I am just suggesting that it just might be an explanation for the difference in the MTF results and the images. Yes, the images never are tack sharp in the borders, but they are also not as bad as the MTF results suggest. A few real life images at MFD should show if astigmatism muddles the results.

[Klaus]

Honestly I could show you a gazillion sample images where the borders are poor at 70mm. Visually I would also rate the results as shown in the review.

 

We do not measure the tangentially and sagittally resolution independently nor does this may any sense from a non-academic perspective. We are looking for the combined maximum of those two values. Thus the result doesn't have to be on the same focus plane as the center (which is, in fact, rather rare). This is also how we can make statements about field curvature - thus the relative distance (Z-axis) of the local maximums between center, border and corner.

[Guest]

For quite a while I have been evaluating mirrorless solutions to replace my trusted Canon Fullframe System. I even bought a Sony A7 only to find that the standard zooms available to this body are are rather miserable performers in comparism to my Canon 24-105. In paritcular the Zeiss 24-70 is a disgrace for the price...... After that experience I even considered going back to APS-C sensor size in order to reduce the weight of my kit.

 

Now it is revealed that the only "high quality" standard zoom for the Sony system is dismal as well.  

 

So what to do?

 

Fuji is no solution, since they seem to stick to this idotic x-trans sensor which  renders landscape details and foilage like watercolour paintings in all user-friendly raw converters (raw-converters like iridient and ninja seem to work better, but  are unaccpetable for  an efficient work flow).  On top the "better" noise performance of the x-trans sensor appears to be due to some on-sensor NR judged by the compartively soft high-iso raws, making Fuji high iso shots no better than those from other manucaturer with some careful NR applied in post.  

 

Samsung cameras  appear to be very nice, but the NX 1 plus a 16-50 zoom are neither small nor light, so it does not really make sense to  sell a fullframe kit for the minor savings in weight and size. On top the lens selection is not very broad.

 

M4/3 does not have the low light performace required for my stock photogprahy, so it is no option altogether.

 

I guess  end up with the EOS M3, which will hopefully have better AF than the original Eos M and a better dynamic range. At least the (too few) EF-M lenses seem to be quite good and  i could use some of my existing lenses.  I am just wondering how  well the Canon adapter works with  large aperture lenses EF lenses (like the 50mm 1.4 and 85mm 1.8.)

 

Looking at the bare technology apparently availble to manufacturer today, I am  just surpised how  every single  manufacturer has so far managed to  keep himself from conquering the enthusiast market with at least one major mistake:

 

Fuji: slow  AF in earlier models and  x-trans sensor which does not work properly with the most  widely used raw converter. Did Fuji really think  everybody would be prepared to change their established workflow and picture library just for a new camera in times when new cameras come and go faster then software is updated?

 

Sony: Bad quality of the most important lens of every camera system: the standard zoom. Zeiss 24-70 and 16-70 are just  a disgrace compared to solutions available for DSLR systems. Having some really nice special purpose lenses  for inflated prices (like 55mm 1.8 or FE 16-35) doesnt help, if the standard zoom is bad.

 

Canon: Slow AF in earlier M model, hopefully now corrected, limited lens selection, no built-in viewfinder, dynamic range not as good as sony sensors.

 

Nikon: No real mirrorless offering for enthusiats. 1-inch sensor is way to small.

 

M4/3: Some very nice cameras and lenses availbale. But too many compromises in respect of low light performance and sensor resolution, if you need to make big prints. Still probably the most well rounded  system for people happy who mostly shoot in good light and do not need to print very large.

[Brightcolours]

The Canon EOS M3 does have better AF than the M, but is not as fast as the quickest Sony/Fuji/Olympus models. The AF in video is unusually well implemented, with smooth, natural transitions.

If you find yourself to be a shadow puller (never was one myself) you might notice the DR in the dark end is less than for instance Sony Exmor sensors, but if you are no shadow puller, the IQ is pretty good (was already pretty good in the original M, and M2).

The external EVF is of good quality, but makes the package more expensive. The 18-55mm STM standard zoom is pretty ok. The AF accuracy of most EF mount lenses is good, the AF speed too (much better than on Sony (F)E with AF capable adapter).

Pretty off topic, all this, though

[Brightcolours]

Honestly I could show you a gazillion sample images where the borders are poor at 70mm. Visually I would also rate the results as shown in the review.

 

We do not measure the tangentially and sagittally resolution independently nor does this may any sense from a non-academic perspective. We are looking for the combined maximum of those two values. Thus the result doesn't have to be on the same focus plane as the center (which is, in fact, rather rare). This is also how we can make statements about field curvature - thus the relative distance (Z-axis) of the local maximums between center, border and corner.

I understand that the normal distance images you shot are not to your liking. However, they should appear quite a bit worse than they do, considering the MTF results. That was my point, I guess. 

 

Close up images would maybe explain the difference.

[Klaus]

I agree that it may have a certain impact at 20mm. However, we are nowhere near close up magnifications.

 

At 70mm, however, the argument isn't really valid anymore.

[Rover]

By the way... is it just me or does the 24-70 suddenly look not so bad in comparison?