I'm getting the feeling that there is a reason many sites (as well as myself) mostly to stick to lab tests or a ton of general shooting to get a general feel for a lens and don't try with a lot of careful real world comparisons as between slightly differing true f-stops, different field curvature, the surprisingly large differences in various parts of the frame that the TINIEST nudge of the focusing makes when shooting complex scenes with tons of detail at all sorts of depths up and down and side to side and back to front (yes even at 16mm and 24mm and f/8 or even f/10!), it's hard to carry out a decent test, one minute I am like oh man the 24-70 II truly is much sharper dead center than the 16-35, the next minute, hmm maybe it's actually close, one minute man the 16-35 does the near corners better, the next hmm maybe not because in that frame what I thought I focused on looks worse but this other focusing trial makes the corners a touch worse when the main subject is more similar in micro-contrast and yet at the same time dead center bottom frame near subject the 16-35 is crisper and same for far subject upper top center and on and on, just a little tweak, even at 24mm and f/8 where people talk about extreme DOF and a touch focusing difference making no difference, well it makes a lot of difference.
And you really need to shoot on a 100% cloudless day near noon in an open area where you won't get brightness constantly subtly changing and changes in shadowing and angle of light hitting the scene every few seconds. As how bright the light and what angle it glances off things can change apparent relative micro-contrast, etc.
It's almost easier to get a feeling by just randomly popping of tons of shots and eventually noticing that one lens, say a 24-105, just seems to never nail certain things, while some others do or doing that and sticking to easier scenes with less depth and just getting a general feeling after a while whether a lens can ever make far edges or corners sharp. Or a careful indoor chart test (but these are a bit tricky for wide angles, it's hard to test the edges in a normal fashion and it doesn't tell you as much about FC). It's almost like either strict test chart constant indoor lighting procedure or just a ton of random snapping makes it easier to get a sense than attempting a too quick supposedly careful real world tricky outdoor test.
But these tests where you have a scene with all sorts of depth and can't find a 100% cloudless day, etc. makes it tricky. And you really need to take like at least 12 focusing attempts, trying to exactly focus on the same spot and maybe another set where you keep trying to get best overall balance of all things in focus at once. Tricky. I have shots that I could show that would make the same lens appear to have somewhat noticeable different corner performance and peak center sharpness and difference in focusing was considerably small. If I labelled on lens A and one lens B you might think boy those lenses perform quite differently
.
One curious thing is that both 24-70 II and 16-35 f/4 IS set to 24mm and f/8, it seems like I can get center frame top distant and center frame bottom close both a touch crisper with the 16-35 than with the 24-70 which seems to hint that the 24-70 II is truer to the f-stops and gives a touch less DOF (and more brightness) for the same aperture, OTOH the extreme corners might be a touch crisper on 24-70 on the torture test scene (maybe a sharpness difference or maybe a field curvature issue or maybe just a nudge of the focus as I'm still not sure I quite got the placement of focal plane quite exactly the same for both, it's very tricky to do, maybe with some giant slanted ruler or an array of cereal boxes to get a better sense of exact focusing, but just based off focusing on a real world scene, it's very tricky, even with more than one attempt, I bet you need a good 12 tries at least).
I also notice that even with copies of the same exact lens and with the camera left locked down tight on a tripod it seems like the framing slightly chances, perhaps hinting at slight tilts of the lens mounts copy to copy being common, which makes it all the more tricky to compare real world.
