Lenses and Optics

Stop it Down. Just a Bit.

Published November 13, 2011

If you read much that I write, you know I love’s me some wide-aperture prime lenses. I like the narrow depth of field that isolates my subject, and the ability to shoot in low light.

More than anything else, though, I love a nice sharp lens with superb resolution and excellent contrast. A lot of people I know shoot their wide-aperture primes wide open all the time (“That’s why I bought them!!”, they say). They realize they’re giving up some sharpness doing that, but most of them aren’t sure how much. I usually shoot mine stopped down just a bit, but I have to admit I wasn’t really sure how much sharpness I was gaining. So I thought maybe I’d investigate that.

Sharpness and Aperture

The truth is wide-aperture primes are usually not exceptionally sharp wide open. They appear to be sharper than they really are because the in-focus subject stands out so much from everything else (assuming the subject is near the center) given the narrow depth of field wide open. But if you measure the absolute resolution numerically, a prime lens shot at f/1.4 is no sharper (and in many cases less sharp) than a top quality zoom shot at f/2.8.

Let’s look at a very good wide-aperture prime, the Canon 35mm f/1.4 “L” lens. Here are the Imatest results for fifteen Canon 35mn f1/.4 lenses measured at both f/1.4 and f/2.8 apertures. (The number on the vertical axis is the MTF 50 weighted average for the entire lens, that on the horizontal axis is MTF50 at the center point).

At f/1.4 (blue diamonds) the lens is very good for a wide-aperture prime. But for a reality check, the 24-70mm and 16-35mm f/2.8 zooms both have higher resolution (shooting at f/2.8, of course) than the prime does shooting at f/1.4. But when the 35mm is tested at f/2.8 (red boxes) it is much sharper than either of the zooms. (I didn’t want to crowd the graph, but if I had put numbers for the two zooms at 35mm, their dots would have filled the white space between the two groups above.)

In fact, when shot at f/2.8, the 35mm L is almost exactly as sharp as our current f/2.8 resolution champion, the Canon 100 f2.8 IS L.

What about other lenses?

Well, that was really useful for those of you who shoot Canon 35mm f/1.4 lenses. But there are probably a couple of you out there who shoot other prime lenses. Since I’m supposed to be using our optical testing equipment to make sure our rental lenses are working properly, I don’t really have time to perform MTF tests for every lens at various apertures (at least not until I finish testing another 2,500 more rental lenses).

Luckily a number of thorough and reputable review sites do evaluate lens MTF at various apertures and publish their results. I love it when other people have already done the work for me.

Since most reviewers test just one copy, and copies of lenses vary, I don’t take any single review as gospel. Testing methods of the various reviewers differ slightly, too, as does their data presentation. And, of course, not every lens has been tested by every reviewer. But I’ve gathered data from the following review sites: DxOmark.com, SLRGear.com, Photozone.de, and DPReview, compared their results as best I could, and tabulated the average findings below.

Lens Best Center % MTF Increase Best Corner % MTF  increase

Canon

14mm f/2.8 II f/4 15% f/8 25%
24mm f/1.4 II f/3.2 15% f/5.6 30-50%
35 f/1.4 f/2.8 33% f/5.6 >50%
35mm f2.0 f/5.6 15% f/5.6 30%
50mm f/1.4 f/2.8 30-50% f/6.3 50-100%
50mm f/1.2 f/2.8 30-50% f/.5.6 50-80%
85mm f/1.8 f/3.2 20-50% f/8 33-100%
85mm f/1.2 f/2 15% f/4 33%

Nikon

24mm f/1.4 G f/3.2 15-25% f/5.6 >60%
35mm f/1.4 G f/3.2 15-25% f/5.6 33%
35mm f1.8 G f/2 15% f/2.8 25%
50mm f/1.4 G f/4 25% f/5.6 33%
50mm f/1.4 D f/4 25% f/5.6 >50%
85mm f/1.4 G f/3.2 15% f/4 25%

Other

Zeiss 21 f2.8 f/4 5% F/5.6 10%
Sigma 50 f1.4 f/4 25% f/4 50-100%
Sigma 85 f1.4 f/4 15% f/5.6 33%
Zeiss 35 f2 f/4 10% f/5.6 15%
Zeiss 85 f1.4 f/4 25% f/5.6 33%

When there was a big disagreement between the various reviewers, I’ve listed a range. But for most lenses they generally agreed and I  averaged out their results as a single number. Take the numbers as estimates: in some cases the site doesn’t present actual MTF or Blur Index numbers, so I’ve taken the data from their graphs. But overall, the agreement between the various reviewers was pretty reasonable.

The variation did surprise me: there are some lenses that benefit hugely by stopping down, others barely at all.

So, What’s the Point?

Like all tables, the one above compresses the data quite a bit. You might want to visit the various websites to get more detail on your own lenses. But in looking at the data I reached several conclusions:
  1. Some lenses benefit greatly from being stopped down, others very little.
  2. For those lenses that do benefit, stopping down just to f/2.0 provides the majority of resolution improvement. The difference between wide open and f/2.0 is generally much greater than the difference between f/2.0 and the maximum resolution.
  3. Getting the edges and corners sharp requires stopping down to at least f/4 for most wide-aperture primes, and some really need  f/5.6. Stopping down to f/2.8 may maximize center sharpness but often makes only a slight difference in the corners, at least on a full-frame camera.
  4. None of the lenses performed any better after f/5.6 (for the center) or f/8 for the corners. Most were clearly getting softer at f/11.

Of course, stopping down the aperture counteracts the narrow depth of field that wide-aperture primes offer, to some degree. But by how much? Well, the simplest and most accurate answer is “get a depth-of-field calculator for your smartphone”. Once you use one for a bit you’ll wonder how you ever got along without it.

But in general stopping down doesn’t increase the depth of field as much as you might think. An 85mm lens on a full-frame camera shooting a subject 15 feet away (a pretty common portrait distance)  has a depth of field of 9 inches at f/1.4; 14 inches at f2.0; and 19 inches at f/2.8. So stopping down to increase sharpness still leaves a nice, narrow depth of field.

Depth of field at the same shooting distance is larger with shorter focal lengths. At the same 15 foot distance a 50mm lens has a  2.3 feet depth of field at f/1.4 (increasing to 4.7 feet at f/2.8), while a 35mm lens focal has a 4.7 foot depth of field at f/1.4  (10.5 feet at f2.8). DOF also increases the further the subject is from the camera and narrows when the subject is closer.

The takeaway pont, though, is for many wide-aperture primes, stopping down slightly, just to f/2.0 or so, will dramatically increase resolution. For others, stopping down makes little difference, especially in the center of the image. Like everything else in photography, the more you know about the specific lens you are shooting with, the better you can use it.

Roger Cicala

Lensrentals.com

November, 2011

Author: Roger Cicala

I’m Roger and I am the founder of Lensrentals.com. Hailed as one of the optic nerds here, I enjoy shooting collimated light through 30X microscope objectives in my spare time. When I do take real pictures I like using something different: a Medium format, or Pentax K1, or a Sony RX1R.

Posted in Lenses and Optics
  • Roger Cicala

    Hi Roger,

    The 135 is a beautiful lens and makes amazing portraits. But it’s certainly not the sharpest lens in your bag – it’s just not designed for absolute sharpness. I would say if it’s as sharp at f/4 as your two excellent zooms are at f/2.8 it’s probably about normal.

    Roger

  • Roger Knight

    Roger –
    I’m real late to this discussion, but I’d appreciate your opinion, if you have one, regarding Nikon’s 135, f/2 DC lens. My 135 is very soft until f/4-5.6, in fact my old 80-200 and 28-70 lenses are as sharp at f/2.8 as the DC is at f/4. Do you or anyone at Lensrentals have experience with 135?
    Great site and great articles.
    Thanks.
    Roger Knight

  • Colin

    OLD photographic saying: “For the best pictures, be there at f/8”. Meaning, be at the right place, at the right time (for the light), and shoot at f/8.

  • Roger Cicala

    Michael,
    The telephotos in general benefit less from stopping down than the standard range lenses. Most are a bit sharper one stop down but few require 2 stops for maximum sharpness. The 200 f2 is a good example, sharper at f/2.8 in the edges but not much difference in the center.

  • Michael

    Hi Roger,

    Great article, I understand the stopping down a little principle on most lenses but would like your opinion on lenses like the Nikon 200 f2 which is reputed to be at its best wide open or 2.8.

    Thanks
    Michael

  • Carl

    Frans, very interesting indeed. I was aware of the high sharpness of the Nikon 35mm f/1.8, but you really are on top of it. I’ll check out your software.

    I notice my Voigtlander 58mm has noticeable “bokeh fringing” along high contrast elements in images at f/1.4, but other than that, the combination of sharpness, color, and contrast is very satisfactory, even to the corners on my 1.6x crop camera (on a full frame I’m sure there’s noticable softness and vignetting towards the corners). I have rented the Canon 24mm f/1.4, the Zeiss 35mm f/2, and the Canon 85mm f/1.2 ii, amongst others…and the Zeiss 35 f/2 was the only one that didn’t have noticable bokeh fringing, wide open at f/2. Both it, and this Voigtlander, seem to let so much contrast in (when wide open), that I wind up needing to under-expose by 2/3, to keep the highlights from blowing out so much (or else use highlight tone priority, or similar). Stopped down, the contrast is less of a problem, obviously.

    When photographing stars or other pinpoints of light, it’s hard to tell what types of aberration there is, wide open, because the bokeh fringing gives them a soft pink or purple glow.

  • Although it is reasonable to summarize lens sharpness with a weighted average across the field of view, I have seen some very interesting behaviour on a Nikon 35 mm f/1.8 by visualizing the sharpness as an image.

    You can try my (free as in beer) software yourself:
    http://sourceforge.net/projects/mtfmapper/

    My copy of this lens sharpens very well on stopping down, going from very good at f/2.8 to phenomenal at f/5.6. It is hard to describe this in words — images (produced by MTF Mapper) show that Sagittal MTF50 figures are better than 45 lp/mm in the centre at f/2.8, but somewhat softer on the edges and corners (around 35 lp/mm). Stopping down to f/5.6, however, flattens out the Sagittal MTF50 image tremendously, with extreme corners showing values of 50 lp/mm. (All MTF50 values measured on a D7000, using unsharpened raw images processed with dcraw). At f/1.8, the corners are decidedly soft, dropping to 25 lp/mm in places, and the centre only reaches about 40 lp/mm.

    The Meridional MTF50 images were equally interesting: they all had a typical radially decreasing pattern, i.e., high in the centre, dropping off rapidly towards the corners, regardless of how much I stopped down this lens.

    So what I found fascinating was that Sagittal resolution appeared to change shape by flattening out on stopping down, while Meridional resolution retained the same “centre hotspot” shape throughout the aperture range.

    I have yet to test all my other lenses to see if they behave similarly.

  • Carl

    I bought the Cosina Voigtlander, and couldn’t be happier. The photozone.de review/test does not lie. It’s sharper wide open, and at least as sharp, if not a tad sharper when closed down…than the 85mm f/1.2L…unbelievable! Bokeh is beautiful and extremely buttery smooth (almost as good as my 135L). Color is very neutral, yet nicely saturated…contrasts are extreme…and near ABSOLUTE ZERO CA and distortion. Now if adapter rings were only priced relative to what it cost to make them… !! I tried the cheaper ring, and am now returning it…very loose!

  • Dawei Ye

    Roger, great article as always!

    I think it comes down to priorities. I am a sharpness freak, but I rather sacrifice some sharpness for a “WOW” shot. A “WOW” shot to me is a greatly diffused background as a result of using a lens like the 85L or 200L wide open. Yes I sacrifice some sharpness from using f/1.2 instead of f/2 (such as in the case of the 85L) but that to me is worth the extra blur in the background.

    Heck, I even photograph group photos wide open at f/1.2 for that background blur. For me, even someone slightly OOF is worth the blurrier background (I make sure nobody is grossly OOF by using focus stacking)

    So definitely different priorities for different photographers

  • Carl

    My profound apologies for spelling “aperture” wrong…evidently I have always pronounced it wrong as well! Cheers!

  • Carl

    Has anyone ever seen odd things when, say…taking a shot that includes mostly sky (not the sun, just sky), while the lens is closed down a lot? It’s disturbing and I know what it is…but wondered if anyone else has seen it. I’ll just bet Roger has.

    Out of curiousity, does anyone have an opinion on the Voigtlander 58mm f/1.4 SLii Nokton? Photozone’s test certainly looks like it is the sharpest lens around 50mm, maybe on earth ?? !!…both wide open and stopped down…but especially wider than f/2.8. Certainly way sharper than either the Canon f/1.4 or 1.2. And not that I dislike Sigma lenses, but their 50 and 85mm f/1.4’s seem to be designed to have a steep dropoff in sharpness for a sensor any larger than their own 1.7x crop cameras. Probably no accident…

    It’s a shame Cosina only makes the Voigtlander in the Nikon mount now. I still think I’d like to have one of these Noktons at some point.

    I rented the Canon 85mm f/1.2L last year, and while pixel peeping, found it to be sharpest across the (crop) frame when closed to f/5.6. The Zeiss 100mm Makro Planar had maybe slightly more contrast, or rather dynamic range between dark and light, but also had a very warm color cast, where the Canon was neutral to cool. The Zeiss 35mm f/2 had even more dynamic range and was more neutral in color than the 100…I didn’t compare it with any other 35’s. I know Roger likes their new 35mm f/1.4 more.

    I must say, after finally renting the Canon 200mm f/2 L last month, it simply did a combination of sharpness, color, contrast, and perhaps “micro” contrast…better than any other lens I have tried. It made my camera take pictures better than I ever thought it could. They seem to have less noise and a more film-like color saturation, with a natural progression toward the brighter end without washing out, or losing detail within colors. And the IS was just awesome! I was able to achieve usable shots with exposures over 1 second, while only mounted to a monopod! The only aspect my 135 f/2 does better (and this surprised me), was background bokeh, especially at longer distances nearer infinity. The 200 f/2 had kind of a busy, almost doughnut thing going on, where de-focused objects behind the focus plane, like tree branches, made a double image of themselves, rather than dissolving into creamy soft focus the way they do on the 135.

    The 135 L is still my favorite lens (that I can afford for now anyway) by far, but it doesn’t zoom, have IS, or go as open as f/1.4. I know the Canon 100 f/2.8 IS is popular now, but I definitely wouldn’t ever own it instead of the 135. The 100 on paper, looks like it’s sharper when wide open (it is a macro!), but I bet the AF isn’t as fast, and certainly I’d bet the color saturation doesn’t compare…not to mention the shallow DOF effect. Before renting the 200, my 135 had the most natural, unhyped, yet extreme contrast of any lens in that focal length range.

    Canon may replace the 135 with an IS version at some point, but that means a totally different lens design, with different elements and groupings. I still think Canon should make a wide arperture zoom, say a 60-160mm f/1.6 with IS. The problem is, if they ever did, based on their price structure now…it would cost $35k! And since they are now going to ISO 200k, is ISO 1 million far off? No…so I guess there’s no incentive to make super-fast megazooms. ISO is getting cheaper, where cutting-edge lens design and construction gets costlier.

  • Flavio Rose

    Thanks. A further hypothesis I offer is that your lp/mm are for MTF = 50%, whereas dxomark.com’s are for MTF = 20%.

  • Roger Cicala

    Flavio,

    Exactly a factor of 2 (and some semantics).we use “Line pairs per mm” and I believe DxO are reporting “Lines per mm”, which should be double. You’ll also notice some sites measure jpgs which seem to show much higher resolution than the unsharpened raw images we test.

    Roger

  • Pingback: You Can Increase Sharpness a Lot by Stopping Down Just a Little()

  • Flavio Rose

    A curious question since you mention dxomark.com. That site gives lens resolutions typically on the order of 40 lp/mm for good lenses. Your used lens sales data gives lens resolutions typically on the order of 20 lp/mm. The graphs in this post give lp/sensor dimension, which assuming full frame also turns out to be on the order of 20 lp/mm. I am wondering why the factor of two difference. Is it just a factor of 2, or is there something I’m missing?

  • DW

    How about labeling those axes?

  • Ian

    Keep up the great work with test tools. As a fellow prime lover, I’m enjoying these articles.

  • teaman

    Thanks very much Roger! Good article.

  • Great post. This is a topic I have been wondering about for a while. Thanks for the breakdown.

  • Roger Cicala

    Teaman,

    Never the 1.0, it’s a specialty lens and very soft wide open. The 1.4 is a good little lens and I don’t hesitate to use it. The f/1.2 is better, but also rather flaky and difficult to use. For the money, unless I really needed the effects of the f/1.2 I’d invest in the 1.4. Or consider the Sigma 50mm f/1.4.

  • CarVac

    I don’t have that many lenses, but I know by heart how much to stop them all down. I know, for example, my C/Y Planar 50/1.4 and Tessar 45/2.8 and Sonnar 135/2.8 are best 2 stops down, my Sonnars 85/2.8 is fine wide open and gets slightly better 1 stop down, and the Distagon 28/2.8 somehow gets sharper and sharper to an extreme at around f/8. Yay for aperture rings and muscle memory!

  • teaman

    What Canon 50mm (or similar) lens would you recommend from these findings for a low light lens? 1.0, 1.2, 1.4, 1.8

  • Pingback: Stop it Down. Just a Bit. | Photography Gear News | Scoop.it()

  • if you’re going to buy a new lens, go to the review sites

    but if you already own it, don’t!! run your own tests, and be sure about what worries you and what doesn’t; know your gear, and it will be become much more useful

    that’s why I ran all these sharpness and bokeh tests:
    http://www.similaar.com/foto/lenstests/lenstestsa.html
    http://www.similaar.com/foto/lenstests/bokehtests.html

    and I carry a short note on my phone telling me how much I have to stop down each lens to achieve optimal image quality

  • Tenisd

    Great post 🙂 Thank You.

  • 60Dx1528x2818x100x28

    Very genuine article! Thanks for that.
    How about adding the Canon 28mm 1.8 to the chart?

  • Aidan

    Another reason to stop down is that the bokeh of some primes is more attractive when they’re not wide-open. For example, the Canon 35mm f/2.0 can have sharp edges to bokeh at f2.0, but magically that’s gone at f/2.2.

    It can also be a reason NOT to stop down, if the bokeh is more attractive wide-open.

    My statements assume that bokeh is what you’re after over sharpness of course 🙂

Follow on Feedly