Sharpest two stops down....or not?

[Radtech1]

I know that anecdotally, we are told that "lenses are sharpest two stops down from wide open."

What I don't know is any studies/test/shootouts to demonstrate that. In fact, I seem to remember reading that lenses are sharpest wide open, but very few camera operators can focus with such an unforgivingly shallow DOF. (I just looked through Luminous Landscape - that is apparently not where I read it.)

What I am interested in is links, if you know of any, to empirical tests/data that show where lenses are sharpest.

Thanks in advance,

Rad

[Wilt]

You only need to look at MTF curves, or the numerical results of lens tests such as those published at photozone.de

Few lenses perform best wide open!

[rdenney]

What I am interested in is links, if you know of any, to empirical tests/data that show where lenses are sharpest.

Rather than provide links to online lens tests, I would rather address the underlying basis for that rule of thumb. That way, instead of trying to verify a rule, you'll know how to make your own rules.

Every lens has its own sweet spot. As you stop down, the fuzziness caused by diffraction increases. Most lenses show visible diffraction effects at the smallest apertures.

Most lenses also reveal their inherent optical faults when they are used wide open. As you stop down, the effects of these faults diminish.

The sweet spot is that point where both lens faults and diffraction effects are jointly minimized.

Some lenses perform best wide open. I have a Zeiss Jena Sonnar that is better at f/2.8 than at f/11. And I expect my 70-200/4L is nearly as good at f/4 as it is at f/8. I have a large-format Schneider Super Angulon that performs optimally at f/22-f/32, but at f/45 diffraction effects are clearly visible. Thus, there's no way to have a general rule of thumb.

The closest you can come is to say that most lenses will perform best at the smallest aperture that keeps the effects of diffraction below the threshold of acceptability. For some of my medium-format lenses, that's f/11 or even f/16. But some lenses are so well-optimized for wide-open usage that there is no effective improvement from stopping down except as a result of increased depth off field.

Just take some pictures of a detailed target over a range of aperture settings and look at the images in actual pixels on your computer monitor. That will tell you what the optimum aperture is for that lens. But remember that the optimum aperture isn't useful if it forces you to use a shutter speed that allows motion blur, or if it lacks sufficient depth of field.

Rick "who thinks other influences on aperture are more important" Denney

[Double Negative]

Each lens is different. Typically, shooting around f/5.6-8.0 will give you the best sharpness with any given lens. Shooting both wide open or stopped down fully are the least sharp and may introduce other issues.

[Tom W]

I would agree with the other 3 - all lenses are different. And all criteria are different.

The "rule of thumb" applies well to all lenses, but there are some lenses like the 200/1.8 or 300/2.8 that are at their sharpest wide open or 1 stop down. Other lenses perform their best stopped down more than 2 stops. My copy of the 16-35 at 16 mm on full frame performs its best at f/11 to 1/16 when corner performance is considered, whereas when center performance only is considered, it's sharp wide open and very good at f/4.

[Wilt]

Interesting how one's perceptions stack up (or not) against the lab tests. Photozone shows 300 f/2.8 performing best in MTF at f/5.6, and least amount of vignetting at that same f/stop...two stops down from max! And the 16-35 does best in center at f/4 and f/5.6, one or two stops from max, and at f/5.6 and f/8 at the edge (no figures for f/11 or f/16, admittedly!)

[Double Negative]

Wide angles are prone to fall-off in sharpness and even light at the edges way more than your average standard-to-telephoto focal lengths... So stopping down is more important.

With very bright lenses, typically designated as L - you're paying for not only the extra brightness but also being able to USE it. That's not to say stopping down won't help any, but they're much better wide open than their cheaper counterparts.

[Tom W]

Interesting how one's perceptions stack up (or not) against the lab tests. Photozone shows 300 f/2.8 performing best in MTF at f/5.6, and least amount of vignetting at that same f/stop...two stops down from max! And the 16-35 does best in center at f/4 and f/5.6, one or two stops from max, and at f/5.6 and f/8 at the edge (no figures for f/11 or f/16, admittedly!)

Keep in mind that photozone tests these lenses on a 350D, so corner performance is largely untested. The 16-35 lens is interesting on a 5D as you can literally stop down and watch the area of good sharpness grow from the center at f/2.8 to the corners at f/11 when shooting at 16 mm.

[JaGWiRE]

Each lens is different. Typically, shooting around f/5.6-8.0 will give you the best sharpness with any given lens. Shooting both wide open or stopped down fully are the least sharp and may introduce other issues.

What if your shooting a lens that is wide open at f5.6 at say 300mm? If you shoot 300mm and at 5.6, would it still be considered sharp, or would it fall under the rule where if it's wide open it's the least sharp.

I'm also curious about how this rule applies to the cheaper super telephotos like the sigma bigma and tamron 200-500 (similiar to the queston above I guess.)

[Tom W]

What if your shooting a lens that is wide open at f5.6 at say 300mm? If you shoot 300mm and at 5.6, would it still be considered sharp, or would it fall under the rule where if it's wide open it's the least sharp.

I'm also curious about how this rule applies to the cheaper super telephotos like the sigma bigma and tamron 200-500 (similiar to the queston above I guess.)


Again, it depends on the lens, and your requirements. My 100-400 zoom at 400 mm shows a noticeable improvement when stopping down from f/5.6 to f/7.1. Stopping down further might have a measureable difference, but it generally doesn't matter to me at that point.

My 300/4 IS prime was sharper wide-open at f/4 than the 100-400 was at f/5.6 & 300 mm, but both showed improvement by stopping down.

[JaGWiRE]

Again, it depends on the lens, and your requirements. My 100-400 zoom at 400 mm shows a noticeable improvement when stopping down from f/5.6 to f/7.1. Stopping down further might have a measureable difference, but it generally doesn't matter to me at that point.

My 300/4 IS prime was sharper wide-open at f/4 than the 100-400 was at f/5.6 & 300 mm, but both showed improvement by stopping down.

Hehe. I see. If I do ever get a super telephoto (probably Tamron 200-500), I probably wouldn't care much about stopping down as I'de be shooting outdoors and with a tripod anyway.

[Double Negative]

Hehe. I see. If I do ever get a super telephoto (probably Tamron 200-500), I probably wouldn't care much about stopping down as I'de be shooting outdoors and with a tripod anyway.

Exactly... And basically, what Tom said. If you start off at f/5.6, sure - you'll need to stop down to say, f/8.0 or f/11 to reach the sweet spot. I was hesitant to agree with stopping down two stops as a generic rule - because what if you have a bright prime?

I guess the point is, stopping down a couple of stops and into the middle of the range is where you'll generally find the best performance of any lens.

[JaGWiRE]

Exactly... And basically, what Tom said. If you start off at f/5.6, sure - you'll need to stop down to say, f/8.0 or f/11 to reach the sweet spot. I was hesitant to agree with stopping down two stops as a generic rule - because what if you have a bright prime?

I guess the point is, stopping down a couple of stops and into the middle of the range is where you'll generally find the best performance of any lens.


You know I think shooting wide open has done more harm to peoples photos then good (for me atleast.) For example, say you own a 70-200 f2.8 (choosing a random lens here that is pretty wide wide open), and you shoot wide open outdoors out of habit. Chances are you'll get some shots with too shallow a depth of field required for whatever you were shooting.

Anyway, as much as I'de love to be able to get beautiful bokeh whenever I want, it seems that you need glass that costs a lot of $. I have a 50mm 1.8 (going to return it) and have serious problems shooting wide open with it (like many others.)

[JNunn]

Since I started in photography shooting medium format where f/2.8 is considered fast, I'm often amazed at how much concern there is over shooting wide open. Or the absolute complusion some people have for only the fastest lenses. Some of the sharpest photographs I've ever taken were taken with ASA 32 or 64, or 80 (almost never past 100) film at f/21 or f/32 or even f/45. That said I almost never use flash, preferring natural light in almost every case.

Shooting at f/2.8 or lower actually is a bit confining in that the DOF is so minimal. Let's see...not good for close portraits (unless you don't care that the ears aren't in focus while the nose is)...not too good for macro (unless your bug or flower is flat!)...not good for landscapes, etc.etc.

I guess that's why I don't find my 70-200mm f/4 or 17-40 f/4 painfully slow as do some people. I almost always need DOF and that means stopping the lens down.

[JaGWiRE]

Since I started in photography shooting medium format where f/2.8 is considered fast, I'm often amazed at how much concern there is over shooting wide open. Or the absolute complusion some people have for only the fastest lenses. Some of the sharpest photographs I've ever taken were taken with ASA 32 or 64, or 80 (almost never past 100) film at f/21 or f/32 or even f/45. That said I almost never use flash, preferring natural light in almost every case.

Shooting at f/2.8 or lower actually is a bit confining in that the DOF is so minimal. Let's see...not good for close portraits (unless you don't care that the ears aren't in focus while the nose is)...not too good for macro (unless your bug or flower is flat!)...not good for landscapes, etc.etc.

I guess that's why I don't find my 70-200mm f/4 or 17-40 f/4 painfully slow as do some people. I almost always need DOF and that means stopping the lens down.

I agree with most of what you said. I don't have a flash yet (going to buy one), but I also really like shooting with natural light.

[PetKal]

Rather than provide links to online lens tests, I would rather address the underlying basis for that rule of thumb. That way, instead of trying to verify a rule, you'll know how to make your own rules.

But remember that the optimum aperture isn't useful if it forces you to use a shutter speed that allows motion blur, or if it lacks sufficient depth of field.

Rick "who thinks other influences on aperture are more important" Denney

Now, I agree with that, Denney.:)
However, if you have some aperture latitude in the field, the lens optics principles are as useless as a third tit. Unless one knows/remembers the lens sweet range, one should stay in the vicinity of the generic sweet spots....like f/4.0-f/8.0 or some such.

Petkal who thinks simple heuristics rules.

[Tee Why]

check out photozone.de

[incendy]

Yep, the data supports it! The 135L holds it's sharpness edge to edge and throughout it's focal range though:) Damn impressive lens

[Tom W]

check out photozone.de

check out that they don't test with a full-frame camera. Good information, but note that the corner information isn't there.

[Double Negative]

Yeah, they're consistent with the 350D... But it would be nice to update everything with a 1Dxx or 5D maybe.

[rdenney]

Shooting at f/2.8 or lower actually is a bit confining in that the DOF is so minimal. Let's see...not good for close portraits (unless you don't care that the ears aren't in focus while the nose is)...not too good for macro (unless your bug or flower is flat!)...not good for landscapes, etc.etc.

I guess that's why I don't find my 70-200mm f/4 or 17-40 f/4 painfully slow as do some people. I almost always need DOF and that means stopping the lens down.

I agree with what you are saying, but I would offer a bit of debate. If you use f/32 or f/45 on an APS sensor, your ability to enlarge will be greatly hampered by the effects of diffraction. An 8x enlargement of 6x4.5 medium format is about 13.5x18 inches. From the 15x23mm format of an APS sensor, it's a 5x7-inch print. If you make a 12x18 print from the 15x23 format, the effects of diffraction will be three times larger and more apparent. Fortunately, depth of field is better with the shorter focal lengths permitted by the small format, so it compensates.

I can get razor-thin depth of field in medium format using my favorit portrait lens, a CZJ Sonnar 180/2.8. At 180mm, f/2.8 is fast, even by small-format standards. Getting DOF that thin with a lens that is 2.4 times normal would require something like a 70mm/1.4 lens in APS format, or 100mm/1.8 in 35mm format. That's one reason why small formats have such fast lenses.

Another reason is that small cameras are intended for hand-held photography, particularly for reportage. Most serious landscape photographers would think of medium format as too small, and a compromise at best. (A compromise I'm certainly willing to make, heh, heh). I'll use f/45 on a 4x5 camera, but if I can't get sufficient depth of field at f/32 in medium format, I look for a different composition. In fact, I think f/32 is the smallest aperture I have in any medium format lens--most only go down to f/22.

I once did some copy work with the 10D using the 50mm/2.5 compact macro. That lens has really small apertures, and because of my copy stand lighting, I used f/22. I had to do it over--too much diffraction effect.

With a view camera, you can control depth of field with tilts, and that's something I wish was more readily available in the smaller formats.

Edit: I forgot to mention that the old photographers thought of thin depth of field as providing sharply focused eyes, slightly fuzzy nose, and out of focus ears. I can do that with the Sonnar, and with the 85/1.8. The old guys did it with f/6.3 lenses on 8x10" format, or with f/4.5 lenses on 4x5" format. I have an old Ilex Paragon 8.5"/4.5 lens for 4x5 that is an example of what was once used. Different formats have a big effect on the apertures one needs for a particular effect.

Rick "who has shot thousands of images on Kodachrome 25 and Panatomic X" Denney

[rdenney]

Petkal who thinks simple heuristics rules.

The heuristic that I'm suggesting is that we practice enough with the lenses we use so that we don't have to have general rules of thumb to guide our practice. For lenses where I don't have that experience, I just use f/8, as in the old saying "f/8 and be there!" Most of the time, though, I have depth-of-field or shutter speed requirements that dictate what aperture I must use to get the effect I want. If the results are unacceptable, I cross that lens off the menu of available optics for those sorts of situations in the future.

Rick "who thinks instinctive responses based on experience are the ultimate heuristic" Denney

[PetKal]

The heuristic that I'm suggesting is that we practice enough with the lenses we use so that we don't have to have general rules of thumb to guide our practice. For lenses where I don't have that experience, I just use f/8, as in the old saying "f/8 and be there!" Most of the time, though, I have depth-of-field or shutter speed requirements that dictate what aperture I must use to get the effect I want. If the results are unacceptable, I cross that lens off the menu of available optics for those sorts of situations in the future.

Rick "who thinks instinctive responses based on experience are the ultimate heuristic" Denney

Gotcha....I am with ya now.;)

[JaGWiRE]

The heuristic that I'm suggesting is that we practice enough with the lenses we use so that we don't have to have general rules of thumb to guide our practice. For lenses where I don't have that experience, I just use f/8, as in the old saying "f/8 and be there!" Most of the time, though, I have depth-of-field or shutter speed requirements that dictate what aperture I must use to get the effect I want. If the results are unacceptable, I cross that lens off the menu of available optics for those sorts of situations in the future.

Rick "who thinks instinctive responses based on experience are the ultimate heuristic" Denney


That makes great sense. I should create like a list of questions I need to answer before choosing my camera settings like:

What are the lighting conditions you are shooting with?
Is what you are shooting in motion? If so, is it in motion very fast or slow?
Are you trying to isolate your subject from your background?
Do you want to freeze motion?

[ChopstickHero]

i find my 17-55 lens sharp at f/3.2 to about f8 or so... after that, you can really tell that it's not as sharp. not saying it's bad, but there's no real reason to shoot from f/9 - f/22 unless you are really looking for some kind of motion blur control.

[JNunn]

ffect.

Rick "who has shot thousands of images on Kodachrome 25 and Panatomic X" Denney

Excellent description! I think(?) my Zeiss 40mm Distogon had f/45, I know my 50mm had f/32. You're also right about the change in perspective in dealing with the different formats and hand holding versus tripod mounted.

Again, excellent description!

[rdenney]

Excellent description! I think(?) my Zeiss 40mm Distogon had f/45, I know my 50mm had f/32.

It's my 180/2.8 and 300/4 Sonnars that have f/32. That's effectively a 5.6mm and 9.4 mm aperture, respectively. But f/45 on a 40? That would be .9mm, which is a tiny, tiny aperture. I would expect lots of diffraction, not to mention the difficulty in making an iris accurately for a hole that small. I'm impressed by the engineering that would be required (and think that Zeiss Oberkochen might be one of the few factories that could pull it off), but I think I would avoid it, heh, heh.

My 30mm Arsat fisheye, 45mm Mir and 50mm Flektogons only go down to f/22. The Mir and the Fleks are old-fashioned retrofocus designs that need to be stopped down to perform well. I don't make really big prints, though. If I did, I'd be having to pull those Super Angulons out all the time.

Come to think of it, I'm going to check to see how small the aperture gets on my 47mm Super Angulon. It might indeed stop down that far.

Rick "who needs to get back into large-format photography" Denney