This article originally appeared in the news section of byThom in May 2012.
It seems a lot of readers are grappling with which midrange zoom to stick on a D800. The commonly cited choices are 24-70mm f/2.8, 24-120mm f/4, and 28-300mm f/3.5-5.6.
Give me a moment. I need to put my flak suit on, buckle my helmet, and check the protective sandbags I'm crouched behind. Ah yes, all good.
All the things you read about resolution are both right and wrong. We've been in a tricky area with DSLRs lately, and it's going to stay tricky until we get far higher pixel counts. More resolution is always good, as you lose nothing and potentially gain something. But we're currently negotiating a specific rapid that has some hidden rocks we need to be aware of. (Can you tell I just got back from that Grand Canyon trip?)
In particular, diffraction is at the heart of some of the confusion. The irony is that compact cameras negotiated this rapid long, long ago. The small sensor size and huge pixel counts we've had most of this decade in compact cameras means that they're all pretty much efficient diffraction recorders. In other words, if you stop down the aperture at all, you record more diffraction on those compact cameras. The lens opening itself, even at maximum aperture, is already contributing diffraction to every compact camera's recording of detail. In fact, it acts like an AA filter, so guess what, you can remove the AA filter (or at least temper it considerably).
Back when we had 6mp DX or 12mp FX cameras, we didn't really have to worry much about diffraction impacts. At the extreme small apertures you could provoke diffraction far enough that it extended across the Bayer pattern enough to be fully realized, but at most apertures it wasn't an issue worth stressing over.
At 36mp on the D800, though, diffraction impacts start recording noticeably at f/8 and above. So we have about half the typically used apertures (f/2.8, 4, 5.6) where we don't have to worry much about diffraction, and half (f/8, f/11, f/16) where we do.
Now let's put another variable into play: maximum performance of a lens. Typically, that's one or two stops down from maximum aperture. And there's the rub. With the 24-70mm we're at f/4 or f/5.6 for optimal results and thus "okay." With the 24-120mm we're at f/5.6 or f/8 for optimal results and venturing into diffraction land. With the 28-300mm much of the range we're at f/8 or f/11 and definitely in diffraction land.
So here's why I put my flak jacket and helmet on and ducked behind the sandbags: different shooters are going to come to different conclusions about those three lenses on the D800.
Let's assume for a moment that they're all "equal" in performance at the same focal length. Let's pick 50mm and assume further that we need to stop down one stop for "picture perfect" results from our lens candidates. The 24-70mm is f/4, the 24-120mm is f/5.6, the 28-300mm is f/6.3. Yeah, I picked this case for a reason: the 24-70mm is not recording visible diffraction (it's there, it's just not a key influencer in final results). The 24-120mm got into the realm where some people say they see a difference on a D800, though a minimal one. The 28-300mm is further up that realm and more people will notice a small difference when pixel peeping.
Put a different way, faster apertures give you more leeway to avoid visible diffraction impacts. Couple this with the fact that the prices on these lenses predict a level of performance, too, and we get to this: the 24-70mm is the best mid-range zoom to put on your D800. Duck!
Phew, that one almost hit me.
Okay, but there's another way to use a D800: shoot at 36mp and reduce in size to 12mp or 18mp. Done properly, the diffraction problem pretty much goes away and these lenses are back to performing like they're on a D3 or D4, which is to say that even the 28-300mm looks pretty darned good. Duck!
Nah, Nah, missed me.
In reality, all three lenses are pretty good. The 24-70mm has some small faults, the 24-120mm very few more, and the 28-300mm surprisingly only a few more. But those faults do add up. On MTF tests, they array like this: 24-70mm best, 28-300mm worst. So how much worse is worse? As much as 20% in terms of absolute numbers averaged over the frame, but some of that is that the 28-300mm's best results occur at f/8, where diffraction is robbing it of a bit of performance, while the other lenses both top out at somewhere around f/5.6, where diffraction isn't yet a major factor.
I'm always reluctant to quote numbers, though, as my photographs don't consist of numbers ;~). If you can live with f/8 as your smallest aperture, the 28-300mm may be good enough that you won't notice the difference. Duck!
Phew, that one almost hit me.
Come on guys, I said "may not notice the difference." I didn't say there was no difference.
Back in the film days, Galen Rowell almost never shot below f/11 (e.g. f/16 or f/22). Why? Because of diffraction. I suspect that if we pixel peeped as thoroughly then as we do now, Galen might have upped that to closer to f/8. Basically, he was trying to balance depth of field with diffraction, and I think the D800 puts us right back in the same realm as film. If you're shooting at f/11 to get DOF, it might not make a huge difference which lens you're shooting with. Duck!
Phew, you almost hit me again.
Of course, the reason to buy a D800 is...well, what is the reason to buy a D800? If you answer "resolution," then I suspect that you need to stay under f/8 and you will definitely find that the 24-70mm puts up slightly better numbers than the 24-120mm, so you automatically put yourself in the "must buy 24-70mm" camp. Duck!
Ha Ha! Didn't even come close to hitting me.
I suspect that people are stressing over the wrong thing. Yes, you can select a lens that takes away some of what the D800 gains you, but if you're jumping from a 12mp camera to 36mp, is that really going to hurt you?
But so far we've just talked about resolution (and its cousin sharpness). Don't these lenses have other attributes that come into play?
Sure. Some of you will reject the 24-120mm because of its vignetting, for example, which is at least a stop wide open (and two stops at 24mm) and at least two thirds of a stop at f/5.6. Remember, you're going to be avoiding f/8, Duck!
Wow, that one came in fast and low.
The linear distortion and chromatic aberration production is relatively high on all these lenses. The 24-120mm isn't particularly better or worse than the 24-70mm. The 28-300mm is slightly worse at a few of those things (linear distortion at the extremes) and slightly better at others. Personally, none of those attributes bother me all that much as they're pretty all correctable after the fact.
Hey! That was a cheap shot, I didn't even say Duck! Fortunately it bounced off me.
For me, it all comes back to center and edge performance, and there the lenses stack up exactly as you'd expect: (1) 24-70mm, (2) 24-120mm, (3) 28-300mm. In particular, the 24-70mm has extremely strong central performance, weak corner performance at maximum aperture that rapidly gets to very good. The 24-120mm has surpisingly good center performance wide open, though at most focal lengths the corners are far weaker, but those corners are quite good up to about 50mm once you get to f/5.6 or smaller. The 120mm performance is good in the center, weak in the edges, pretty much no matter which aperture you pick. The 28-300mm is the weakest in the center but still very good (better than I expected and strong enough for some to consider even on a D800), but at anything less than f/8 the corners are very weak. At above f/8 the lens is diffraction limited in what it can do and isn't far off from the rest.
So my preference is exactly as you might think it is and the price points suggest: (1) 24-70mm, (2) 24-120mm, and (3) 28-300mm.
Let me check my helmet for a moment. I think I'll put on another layer of flak protection.
Buy the 24-70mm. While it has plenty of modest weaknesses, that extra stop actually buys you a bit of performance at f/4 (plus, of course, you have f/2.8 ;~). If you're on a budget or can live with the edges, okay, buy the 24-120mm. It's surprisingly good on the D800, enough so that the extra focal length starts to make it highly tempting, even if that focal length is the weakest part of the lens. Plus you've got VR.
Whoa, almost didn't see that shot coming. Fortunately I'm immune to that shot, as I've long said that the 24-70mm needs VR.
The 28-300mm? No, I'll pass. I know a lot of people like the notion of one-lens-does-everything, but that seems to me to be in direct contradiction to the core of what the D800 is. I really don't see the point of buying the DSLR with arguably the best image quality available and then slapping it with the penalty of less capable glass (probably with a protective filter out front that further reduces capability).
Hey, was that a Swiss Army Knife someone attacked me with?
There's another option out there, though. Get the 28mm f/1.8G, the 50mm f/1.8G, and the 85mm f/1.8G. Yes, f/1.8. Another hint at what I'm likely to write in future reviews: I like the 85mm f/1.8G better than the f/1.4G, and I've already written that I prefer the 50mm f/1.8G over the f/1.4G because of focus speed. Other than some field curvature you have to work around, the 28mm f/1.8G is right up there, too. And these lenses will chase away most of those demons you see in the corners of the zooms. If you're not on a budget and don't mind manual focus, the Zeiss ZF.2 primes will polish those D800 pixels even more. A 21/25mm, 50mm, 100mm Zeiss combo basically nets you the best you can do right now on a D800.
This also brings me to my last thought before I duck down deep behind my bunker: maybe the 24-120mm f/4 is the right choice. Coupled with three primes, what would you be missing?
Small clarification. There was one post about this article on dpreview that pointed out a point I didn't make fully clear here. I'm writing about peak lens performance in this article (and peek performance, as in peeking at pixels). Some people do indeed get confused, as that poster wrote. If we print a D3, D3x, D4, and D800 image all at 14" wide, diffraction isn't going to limit the D3x or D800 image (it's buried too deeply in the print's pixels where you can't see it). As I noted at the beginning of the article, more resolution is always good. While I write for sophisticated shooters who know these things, I forget sometimes that there are others trying to get up to that level that might not have yet learned everything I assume or are confused by something I write because they're missing a piece of information.
One reason why Galen was often criticized by other photographers during his career was that those (medium and large format) photographers felt that he was limiting his potential gallery print size by using a smaller capture format. The diffraction would show up in very large prints because you were magnifying the small frame more to get the big print (compared to 645 or 4x5). Yet he was very careful to understand just how far he could go, and pretty much went right up to the edge by practicing great shot discipline and understanding the balance between DOF and diffraction in the equipment he was using. I'm reminded of that when I use my D800: with care I can push into what used to be the sole territory of medium format. But push too far and things like diffraction starts to show.
So let me state this: there will be times when it is worth going into diffraction-limited apertures to capture more DOF. As I noted, more resolution is always good, and done right you should gain more from the DOF than you lose from the diffraction. But once diffraction is the primary limiting factor of your len's resolving power, the gains from the extra resolution start to fall. A lot of people are buying D800's to "print big." Those people, especially, need to understand the gains and losses they're going to encounter, and to do what Galen did: carefully balance them.
Nikon's Recommended D800E Lens Set
The following lenses are ones that Nikon NPS has published as having excellent resolution, and by implication (they don't actually come out and say these are the recommended lenses for D800E users), work well with the D800E. Indeed, I'd tend to say these are the best lenses Nikon has produced in terms of optical quality, though I'd probably add the 70-300mm f/4.5-5.6G VR to this list.
- AF-S NIKKOR 14-24mm f/2.8G ED
- AF-S NIKKOR 16-35mm f/4G ED VR
- AF-S NIKKOR 18-35mm f/3.5-4.5G ED
- AF-S NIKKOR 24-70mm f/2.8G ED
- AF-S NIKKOR 24-85mm f/3.5-4.5G ED VR
- AF-S NIKKOR 24-120mm f/4G ED VR
- AF-S NIKKOR 70-200mm f/2.8G ED VR II
- AF-S NIKKOR 70-200mm f/4G ED VR
- AF-S NIKKOR 80-400mm f/4.5-5.6G ED VR
- AF-S NIKKOR 200-400mm f/4G ED VR II
- AF-S NIKKOR 24mm f/1.4G ED
- AF-S NIKKOR 28mm f/1.8G
- AF-S NIKKOR 35mm f/1.4G
- AF-S NIKKOR 85mm f/1.4G
- AF-S NIKKOR 85mm f/1.8G
- AF-S NIKKOR 200mm f/2G ED VR II
- AF-S NIKKOR 300mm f/2.8G ED VR II
- AF-S NIKKOR 400mm f/2.8G ED VR
- AF-S NIKKOR 500mm f/4G ED VR
- AF-S NIKKOR 600mm f/4G ED VR
- AF-S NIKKOR 800mm f/5.6E FL ED VR
- AF-S Micro NIKKOR 60mm f/2.8G ED
- AF-S VR Micro-Nikkor 105mm f/2.8G IF-ED
- PC-E NIKKOR 24mm f/3.5D ED
- PC-E Micro NIKKOR 45mm f/2.8D ED
- PC-E Micro NIKKOR 85mm f/2.8D
One thing to note about this list is Nikon has included lenses that are f/5.6 at maximum aperture, and then makes the somewhat ambiguous statement "greatest contrast…can generally be achieved at an aperture setting two to three stops down from maximum aperture." You could be at f/11 with some of these lenses if you followed that advice, and diffraction would most likely render the D800 and D800E results as equal at that aperture (there's still a very small measurable difference, but not a visual one).
I would phrase things differently than Nikon: if used at maximum aperture or any aperture of f/4 or faster, these lenses will attain more acuity on a D800E than a D800. The f/1.4 and f/1.8 lenses can be stopped down to f/2.8 or f/4 and get better than wide open results. The f/2.8 lenses generally can be stopped down to f/4 and get visibly better results on a D800E than they would wide open (the exceptions might be the 300mm and 400mm). But once you're above f/5.6, diffraction begins stealing acuity. At f/11 and above that acuity loss is going to be clearly visible if you're a pixel peeper.