Jeff
Offline
• • • • • •
Upload & Sell: On
|
 p.3 #19 · 24-70/2.8L II field curvature (is this normal?) | |
So, in answer to my own question: "Is this normal", after Roger's input and further testing of two different (new) lenses, I can say that it appears that it is indeed, 'normal', with qualifications. I spent 8-10 hours over the last two days taking test shots between the two lenses, and one thing I'll say is that, albeit with a sample of only N=2, Canon's quality control appears to be very good. The original lens and the replacement lens performed nearly identically, even with some quick-and-dirty astro shots to evaluate coma.
Although the replacement lens appears to be ever-so-slightly decentered (I'd say roughly similar to Roger's sample) compared to the original which was perfectly centered, I ended up choosing the replacement lens after doing a very controlled test this morning. My gut feeling with the original lens was that, *even when properly focused*, the majority of the depth of field tended toward the fore-, rather than background. The result was as originally posited, that is, people to the sides in a group portrait would end up significantly out of focus, at 50mm and f/5.6. I've just never seen this before with any lens I've used, which led me to post this thread.
I was beating my head against the wall trying to pick one of these two lenses, until I finally found the combination of conditions that made my decision easy. Focusing on a water bottle in a dirt parking lot at a distance of ~20m and 24mm at f/2.8, the A:B comparisons were focused ever-so-slightly differently (both within tolerances of AF systems, I'm sure), but when I then looked at the hill behind (effectively at infinity), the difference in sharpness was remarkable. In the screenshot below, note the clear difference in the clarity of the towers; I'd guess that image #3796 was focused ~3" behind the bottle, while #3813 was perhaps 5" in front of the bottle (see screenshot, tough to tell). At a distance of 20+ meters and a focal length of 24mm, I personally wouldn't expect to see such a big difference in sharpness at infinity, as the example shows.
Anyway, I just wanted to update everyone on where this ended up, and thank everyone for their help (I hope Roger still posts his field curvature results to this thread for the wider angles). I do still have reservations about this lens, however I'm sure being new to the 5Ds (as we all are) is related. I shot some sequences with the 1Dx this morning as well, and as expected, some of this was a bit harder to see in the lower-resolution images.
These two 24-70/2.8L II's are indeed insanely sharp in the center of the image, but never get insanely sharp to the corners, as I've heard/read. The lens is quite soft at f/2.8, even in the center, but sharpens up nicely from f/4 on. You can clearly see the observed visual relationship manifested in the file sizes of images captured on a tripod at different apertures (see below), but keep in mind this analysis used a 5Ds, so all differences are effectively magnified.
Anyway, I will *not* be using this lens for group portraits the way I used the Mark I for many years, but it will have some value. I'm a bit bummed about paying an astronomical price for a lens that doesn't perform similarly, but I'd have thought the field curvature issue would have been discussed more than it has been; I believe the 5Ds is to blame for that, as most other users might not notice, as Roger pointed out. I think this is one of those cases where the incredible resolution of the 5Ds definitely brings out the worst in a lens design, and perhaps f/4 lenses will be the norm for me in the future, esp. considering how thrilled I am with the performance of the 16-35/4 IS. Brandon's post (thanks!) regarding how correction of spherical aberration in a given lens effects the depth of field was the kicker that allowed me to pick one of these lenses, and gave me the clarity to understand what I was seeing in the first lens, without there being any 'obvious' problem.
If any of you run across situations where this field curvature issue manifests itself, feel free to post here in the future, I'd be interested in seeing it.
Thanks for all the input, and cheers...
Jeff
full-frame image
Focal point
Background comparison
File sizes
|
