EF 24-70 f/2.8L Replacement [CR2]

[neuroanatomist]

So you don't agree that 50mm/2.8 on a FF is identical to 31.25mm/1.75 on 1.6 APS-C? Or is it something else?

Yes I don't agree with that, f/2.8 is f/2.8 regardless of the crop factor

Ok, good, then I understand. You are right, that for the same lens the aperture doesn't change when you go from FF to crop. That would be weird. But now think of two different lenses that both share the same entrance pupil - say 18mm in both cases - but where the focal length is different, namely 50mm in one case and 31.25mm in the other.

Yes, f/2.8 is f/2.8 and the light per unit area of the sensor is the same, independent of sensor size. Likewise, the focal length of a lens doesn't change on crop vs. FF, just the angle of view (i.e. the magnification is the same). The effect on DoF is an apparent one, seen in practice, even with the same lens. If I frame a head shot with an 85mm f/1.2 @ f/1.2 on my 7D, then I say 'whoops, what am I thinking' and move the 85/1.2 onto a 5DII without moving my feet or the subject, instead of a nice, tight head shot I'll have a lot of background around the subject. So, I'll take a few steps forward to get that nice tight framing again, and that decreased distance-to-subject translates to a shallower DoF - by a factor of 1.6x.

 

[neuroanatomist]

The saving grace of the 24 - 105mm is the IS which the 24 - 70mm doesn't have, when it finally gets it I think the 24 - 105mm will be going on Ebay!

With this, I agree. If Canon ever gets around to releasing the 24-70mm f/2.8L IS, my 24-105mm f/4L IS will go on the auction block...

 

[epsiloneri]

Photozones test of the 24 - 105mm f/4 IS on FF wasn't what I wanted to see, it has problems with edge softness vignetting, distortion, and its minimum focus distance and magnification leave a lot to be desired. It's useless for close up work, the 17 - 55mm is much closer to the 24 - 70mm here.

As I already wrote, according to the photozone tests themselves, 24-105mm/4.0L on 5D2 is at least as good as 17-55mm/2.8 on 7D, at least if we're talking IQ (7D has other advantages). I confirmed it with my own tests. In other words, the 17-55mm/2.8 on 7D is just as useless as 24-105mm/4.0L on 5D2.

f/2.8 on APS-C gives about the same viewfinder brightness as f/4.5 on FF. Why is that? Because in order to view the full field of view in the same solid angle, the FF viewfinder needs to compress the image (almost) 1.6x more than the APS-C viewfinder -> increasing the surface brightness. This is easy to test yourself: mount the 17-55/f2.8 on a 7D and the 24-105/4L on a 5D2. Which viewfinder is brightest? They're about the same (the 5D2 has a slightly larger apparent field in the viewfinder, but it's compensated by the lens being 4.0 instead of 4.5). If you don't believe me try it! (Of course, your 24-70mm/2.8L lens will be much brighter on the 5D2.)

 

[epsiloneri]

So, I'll take a few steps forward to get that nice tight framing again, and that decreased distance-to-subject translates to a shallower DoF - by a factor of 1.6x.

Yes, but the perspective will have all changed, so it's not equivalent. I think we both agree that, in contrast, the imaged formed by a 50mm/2.8 lens on a FF sensor is actually equivalent to the image formed by a 31.25mm/1.75 lens on a 1.6x APS-C sensor. Actually equivalent (identical!), in terms field of view, photon collection rate per solid angle, depth of field. No difference.

 

[kubelik]

dilbert, the guys over at luminous landscape would have to disagree:

http://www.luminous-landscape.com/reviews/lenses/24-70-review.shtml

at least to me, I value the opinion of actual in-the-field shooters a heck of a lot more than I value DXOmark scores, which, to put it politely, I consider a bunch of hogwash