Aperture and lens effects on stacking

A forum to ask questions, post setups, and generally discuss anything having to do with photomacrography and photomicroscopy.

Moderators: rjlittlefield, ChrisR, Chris S., Pau

rjlittlefield
Site Admin
Posts: 23561
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Aperture and lens effects on stacking

Post by rjlittlefield »

In a couple of recent topics (1, 2), George Dingwall touched on some questions of how to get best results when stacking, for example what lens and aperture works best, and how much improvement might a different lens give?

With luck, this post will answer some of those questions (while no doubt prompting others :D ).

Background is that over the years I've accumulated a variety of lenses that I've found useful for macro work, and I finally decided to set up a head-to-head test on a common subject and see what happened.

Test subject was the snout beetle face that I had previously posted.

Lighting was fiber optic halogen, with a pingpong ball diffuser.

I tested seven different lenses:
  • El Nikkor 50mm f/4 enlarging lens, reversed
  • Mamiya Sekor 55mm f/1.8 lens from 35mm film SLR, reversed
  • Olympus 38mm f/2.8 bellows macro lens
  • Olympus 20mm f/2 bellows macro lens
  • Zeiss Luminar 16mm f/2.5 macro lens (looks just like a microscope objective with an aperture ring)
  • aus JENA 10X NA 0.25 microscope objective (effectively 14.5mm f/2)
  • Edmund 20X NA 0.40 microscope objective (effectively 7.6mm f/1.25)
Due to the large range of focal lengths, I could not set up to get exactly the same field size (magnification) with each lens. So instead, I set up to get as close to the same magnification as I could without moving outside the lens' design parameters.

For lenses with variable apertures, I shot a series of single frames stepping aperture over each lens's full range. From those, I picked the aperture giving highest resolution, to use for stacking.

Along the way, I noticed that a couple of my lenses seemed to have a lot of flare when used without a macro lens shade, so I made one of those. It helped a lot.

Here are some pictures that summarize the results.

First, full frames illustrating the field size for each lens test. You can figure out the magnifications yourself, from the scale bar in earlier post. This image also shows the macro lens shade and its effect. You can see from the Photoshop histograms that without the shade, all the dark tones suffer pretty badly from stray light

Image

Second, here are some illustrations of what happens as various lenses get stopped down.

All of the lenses have lousy resolution when stopped down too far, mostly due to diffraction. But they vary in what happens at wider apertures.

The Mamiya lens in the middle is typical of what I find with "ordinary" lenses -- performance peaks around f/5.6 and falls off for wider apertures. There's also a significant shift in focus point over the first couple of f/stops, so it's a bad idea with this lens to focus wide open and then stop down to shoot. (I've seen the same effect with other lenses, including my Sigma 105mm macro.)

In contrast, the Olympus 38mm macro lens actually has highest resolution wide open at f/2.8, and starts to degrade noticeably at f/5.6.

By the way, it would be a mistake to compare across lenses from these pictures. These are actual pixels, but the magnifications are different.

Image

Finally, here is a direct comparison of center resolution from finely spaced stacks for all the lenses. What I've done here is to resize the result images so that they're all the same magnification. The Olympus 38mm image is just about actual pixels, and the others are scaled up or down to match. (There's a bit of detail lost in the 20X image, but I think the others keep everything that matters.)

Image

So what's the bottom line?

Well, it depends...

At one extreme, we've got the standard microscope objectives. They give great resolution and they're pretty cheap, but they give very short working distance and the DOF per frame is so shallow that you have to shoot really deep stacks, which costs time and wears out your camera. At the other extreme, the Mamiya and El Nikkor lenses give great working distance and they're cheap too, but the resolution isn't quite up to the dedicated macro lenses and flare may be more of a problem. In the middle there are those dedicated macro lenses, which are real nice if you've got 'em, but they're fairly expensive and can be hard to find.

Note that center resolution doesn't tell the whole story either. Go very far off center, and some lenses start to lose resolution and pick up color fringes. The dedicated macros seem to shine in this regard, showing essentially no change in image quality across the whole frame (as shown by other tests, not these).

I'll stop here, 'cuz I really don't know what might be of interest to anybody reading this.

Questions, comments? George?

--Rik

Carl_Constantine
Posts: 304
Joined: Tue Aug 01, 2006 8:02 am
Location: Victoria, British Columbia, Canada
Contact:

Post by Carl_Constantine »

I'm kind of surprised at your peak performance results. I would think that at F11 or F22, the depth and clarity would be better (though a significantly longer time to shoot) than the lens would be wide open, or even at F5.6. Also, you don't say what camera you are using (though it doesn't really matter) as I'm curious if any of these lenses work with mine ;-)

Yes those that have the microscope objective are definately out of the question for me. I don't want to have to be right on top of my subject to get a picture, particuarly if I have to do any stacking.

The 20mm Olympus and 16mm Luminar didn't do too bad at all. After playing with my sister-in-law's 90mm Tamron macro, my heart really is set on a macro lens. It may just be a while before I can buy one. I'm also interested in similar tests of:

Tamron 90mm Macro Di
Sigma 150mm Macro
Sigma 105mm Macro
Canon 100mm Macro


Thank you for taking the (what must have been) incredible amount of time to create this mini report Rik.
Carl B. Constantine

rjlittlefield
Site Admin
Posts: 23561
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Post by rjlittlefield »

Carl_Constantine wrote:I would think that at F11 or F22, the depth and clarity would be better (though a significantly longer time to shoot) than the lens would be wide open, or even at F5.6.
That's a common belief, Carl, but it's completely wrong. (Not to worry, you're in good company! :D )

That why I keep harping about stacking being a good thing, if you can afford the limitations.

To get the sharpest image, you need to open your lens a lot farther than the standard recommendations for macro work. That's because the standard recommendations for macro work are based on balancing the effects of geometric blur and diffraction in one shot. Stacking lets you shift the balance and get the same DOF with a lot higher sharpness, in addition to the obvious possibility of getting a lot more DOF with the same sharpness.

I have posted some other stuff on this topic that you may find helpful. Try:The third ref in particular addresses the Sigma 105mm.

The camera was a Canon 300D, the Digital Rebel.
Thank you for taking the (what must have been) incredible amount of time to create this mini report Rik.
You're welcome. I seem compelled to try for "correct and complete", though of course this is impossible. It's a curse. :lol:

--Rik

rjlittlefield
Site Admin
Posts: 23561
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Post by rjlittlefield »

Carl,

You might be amused by the following story...

I started doing photomacrography back in high school, probably '66 or thereabouts. 'Twas all 35mm film at the time, of course. Well, I realized quickly enough that to get large magnification, I had to use short focal length lenses, and all that I had available were microscope objectives. But dang! Every time I tried to use a microscope objective, the DOF got so small it was hopeless. Stopping down a microscope objective is fairly tedious (I machined aperture plates and dropped them in), so I went looking for a short focal length lens with an iris diaphragm. Pretty quick I found a surplus lens from an 8mm movie camera -- a 12.5mm f/1.9-16 Rodenstock-Ronar, serial number 4193938. (How do I know this, you ask? Keep reading.) Cheap, the lens was, and soon delivered.

Well, the lens was a little tricky to mount, but after a couple of hours of lathe work I had a cute little adapter with Pentax M42 threads on one end and some ludicrously fine thread on the other end to fit the lens' filter threads. (21 x 1/3mm, I think it was.)

Mounting the adapter on my bellows, I eagerly proceeded to shoot a test series of frames, and scurried off to the darkroom to develop them. What a disappointment! Even at marked f/16, the DOF was still not enough to be useful, and the resolution was ghastly. I gave up in despair, blaming a cheap lens and looking forward to the day when I could afford something better...a lot better.

Of course, the real problem was that at the time I did not understand about diffraction. And if I had understood about diffraction, I still would not have been able to do anything about it, because at the time computers were Really Big and Really Expensive and still were nowhere near powerful enough to do anything like stacking.

I kept the lens, though. It didn't take much space in one of my junk drawers.

A few minutes ago, I put that old "junk" lens on the front of my current setup. (The old adapter still works fine, since now I've got an M42 --> Olympus adapter for the front of my bellows.)

A quick aperture series showed that f/2.8 was good, and then I ran a stack.

Here's the result.
Image

Is it a great lens? Nope, and for under $10 even then, you wouldn't expect it to be.

But is it a piece of junk, as I thought? Certainly not! Actually it gives a bit higher center resolution than the Mamiya 55mm, though not quite as good as the Olympus 38mm.

The problem was not the lens, it was simply that I was asking light to do something contrary to its nature.

If you cram light through a small hole, the images get fuzzy.

It took years for me to learn that fact so that it's now intuitive.

I suppose next somebody figures out how to change the rules so that it's no longer true. (That'll be an excellent day. :D )

--Rik

georgedingwall
Posts: 207
Joined: Tue Aug 01, 2006 11:15 am
Location: Invergordon, Scotland
Contact:

Re: Aperture and lens effects on stacking

Post by georgedingwall »

Hi Rik,
rjlittlefield wrote: Questions, comments? George?

--Rik
This is a really interesting topic to me. It illustrates well that each lens can have it's good and bad points and shows the need to understand your equipment when working at these extreme macro setups.

Since building my home made focussing rail, I've been working through my lenses to try to find which ones are best, and at what settings I should use them.

So far, I've established that My Nikon 105 mm Micro is best at F5.6, My Sigma 150 mm Macro lens is best ay t F8 and the 50 mm F2 Nikkor EL enlarging lens is best at F5.6. All 3 lenses seem to give better results when reversed.

I think also that a key point you touch on is the use of a diffuser. While one isn't necessary on all shots, a diffuser seems to give a sharper, clearer image when the subject has reflective surfaces.

I don't have the ability to attach a microsope objective to my camera, but I can attach my camera to my telescopes. I may try a few more high magnification shots this way. The main problem I have with the stereo zoom microscope is that it does not have a fine focus, so it is difficult to be consistent in the step sizes. I'll post any results I get.

Thaks for doing all this hard work. I'm sure it's of interest to lots of people.

Bye for now
George Dingwall

Invergordon, Scotland

http://www.georgedingwall.co.uk/

Charles Krebs
Posts: 5865
Joined: Tue Aug 01, 2006 8:02 pm
Location: Issaquah, WA USA
Contact:

Post by Charles Krebs »

Whew Rik! You are a glutton for punishment! Thanks for putting it together.

An anecdotal thought on the design of special "photomacrography" lenses. Most of the good ones from 25mm to 50mm use a 6 element/4 group design. This includes the 20/2 and 38/2.8 Olympus, Canon 35/2.8, and Leitz Photar 25/2.5. I suspect you would find that the 50/2.8 El Nikkor which also has the 6/4 construction would be noticeably better than the 4 element "tessar" type 50/4.

As you have illustrated, flare is a big issue with this type of work. As I mentioned to George elsewhere, it seems even greater care must be taken with digital SLRs (compared to film cameras) because of the flat, highly reflective surface over the sensor. In addition to proper "shading" up front, I have been able to really cut flare by inserting thin, opaque, matte black "apertures" inside the various adapters I use, with the opening cut just large enough to allow the image to hit the sensor with no vignetting.

DaveW
Posts: 1702
Joined: Fri Aug 04, 2006 4:29 am
Location: Nottingham, UK

Post by DaveW »

Carl

To understand diffraction try these sites:-

http://www.exploratorium.edu/snacks/diffraction.html

http://en.wikipedia.org/wiki/Diffraction

Theres a nice little interactive one to play with here:-

http://www.phys.hawaii.edu/~teb/optics/java/slitdiffr/

This one explains it in relation to photography:-

http://www.cambridgeincolour.com/tutori ... graphy.htm

DaveW

rjlittlefield
Site Admin
Posts: 23561
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Re: Aperture and lens effects on stacking

Post by rjlittlefield »

georgedingwall wrote:I don't have the ability to attach a microsope objective to my camera
George,

A plastic body cap can become the adapter you need. Just drill a hole in it large enough to accept the objective's threads, and cobble up some scheme to hold the objective in place while you play with it.

For the setup shown in Figure 6 here, I used duct tape. (It's carefully trimmed to be unrecognizable in the photo. :wink: )

If you have a lathe handy, just bore and thread the hole to match the objective's threads. They're probably RMS thread (Royal Microscopical Society), which is 0.8" diameter, 36 tpi.

That scheme (body cap, bore, thread) is how my M42-->Olympus adapter got created, after searching the Internet failed to find what I needed.

--Rik

Carl_Constantine
Posts: 304
Joined: Tue Aug 01, 2006 8:02 am
Location: Victoria, British Columbia, Canada
Contact:

Re: Aperture and lens effects on stacking

Post by Carl_Constantine »

rjlittlefield wrote: If you have a lathe handy, just bore and thread the hole to match the objective's threads. They're probably RMS thread (Royal Microscopical Society), which is 0.8" diameter, 36 tpi.

That scheme (body cap, bore, thread) is how my M42-->Olympus adapter got created, after searching the Internet failed to find what I needed.
Some of you just have more "gusto" than I do. Playing with things to make them fit/work even though they are not designed to do so. I'm still at the "it smells new" stage on my Canon 300D so I won't be putting duct tape on it anytime soon, let alone bore holes in a handmade adapter.

Thanks again Rik for all your hard work on this. Thanks Dave for the extra links.
Carl B. Constantine

rjlittlefield
Site Admin
Posts: 23561
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Post by rjlittlefield »

Charles Krebs wrote:You are a glutton for punishment!
Charlie,

That's true, but in this case I really wanted to know the answers and I couldn't think of a much easier way to get them. :?

What surprises me, in these results, is how much better the microscope objectives did than even the dedicated macro lenses.

According to theory, the resolution of a diffraction-limited lens system depends only on its effective aperture -- the angle of its entrance cone.

That theory predicts that an NA 0.25 microscope objective and a really good f/2 macro lens (at high magnification) would give pretty similar resolution, since NA 0.25 is very close to f/2.

Yet in these tests, an inexpensive objective (10X NA 0.25) gave much sharper images than two much-touted macro lenses of similar focal length and nominal aperture (Luminar 16mm f/2.5 and Olympus 20mm f/2).

Why would that be?

Reviewing the pictures this morning, I think I see one important clue. The full frames from these three lenses show similar magnification. But the blur of the background antenna is much wider for the objective than for the two macro lenses. That seems to indicate that the objective actually has a much larger effective aperture than the macro lenses, markings and calculations notwithstanding.

I'd be interested to hear your thoughts about this. If you see some mistake in my thinking or experimental procedures, please, let me know!

--Rik

twebster
Posts: 442
Joined: Wed Jul 26, 2006 8:02 am
Location: Phoenix "Valley of the Sun", Arizona, USA

Post by twebster »

Hi y'all, :D

Let me add a little more confusion to the subject of dedicated macro lenses. :D

I had once read a book on optical design by Alfred Blaker, one of the optical design gurus of modern times. Diffraction effects increase not only with how small an aperture becomes within the lens but also increases with the distance the aperture is from the film or, in our case, digital sensor. Diffraction effects increase with increasing distance between the lens and film plane. As the distance between the lens and film plane is increased, the perceived size of the aperture is decreased. The farther you move the lens from the film, the smaller the aperture appears. Crazy as this sounds, the effects are quite real.

This is one of the reasons dedicated macro lenses, such as the OLympus 20mm and 38mm lenses, have short focal lengths and are best corrected to deliver maximum resolution at maximum apertures. Short focal length macro lenses do not have to be moved very far from the film plane to increase image magnifications. The apparent size of the aperture doesn't decrease as much when you extend a 38mm lens 76mm from the film plane to achieve 2:1 image magnification as would a 100mm lens set at an aperture of f 5.6 and extended 200mm from the film plane to achieve 2:1 image magnifications. Diffraction effects become greater from the 100mm lens because the apparent size of the aperture decreases much more.

When reading the specifications on true macro lenses, such as Zeiss Luminars and Leitz Photars, you will find that each focal length has been assigned optimal image magnification ranges. These ranges are based on 1) the optical correction of the lens for these image magnifications and 2) the least amount of apparent change of the aperture at the extensions required to produce the optimal image magnifications. "Regular" camera lenses are designed to give acceptable image resolution over a wide range of focusing distances, even what are offered as "macro" lenses. However, compromises have to be made in these designs. True macro lenses, however, are optically corrected to give the best possible image quality within a narrow range of focusing distances.

Just how narrow is the optimal magnification range? Let's look at Leitz Photar lenses, for example:

12.5mm f 1.9...15x to 30x
25mm f 2.5...7x to 16x
50mm f 2.8...3x to 8x
50mm f 4...3x to 8x
80mm f 4.5...1x to 4x
120mm f 5.6...1/2x to 2x

I'm sure you can see the pattern. The longer the focal length, the smaller the magnification range. Longer focal lengths need more extension to achieve high image magnifications than do shorter focal lengths. These longer extensions place the aperture much further from the film plane resulting in greater effects from diffraction.

(Rik...As an aside, Zeiss Luminar lenses are corrected to be used with tungsten illumination for best sharpness and color rendition.)

Asymmetrical lens designs can also increase diffraction effects when used with extension. You can tell how asymmetrical a lens is by viewing the aperture from the front of the lens and from the back of the lens. Stop the lens down to a mid-range f stop. View the aperture from the front of the lens and from the back of the lens. If the apparent size of the aperture changes, then the lens is asymmetrical. The aperture viewed from the back of the lens is known as the "exit pupil". The smaller the exit pupil the greater the effects diffraction will have on the image when the lens is used outside of its intended focusing distances and image magnifications. Reversing asymmetrical lenses can actually improve lens performance when used for macro photography as the exit pupil will now be the larger of the two views of the diaphragm in relation to the film plane. This is also why dedicated or "true" macro lenses have nearly symmetrical lens designs. John Shaw in his book, "Close-ups in Nature" touches on this subject on pages 141 and 142.

Have I got you all confused now :?: :D

Best regards to all, :D
Tom Webster

Phoenix "The Valley of the Sun", Arizona, USA

The worst day photographing dragonflies is better than the best day working! :)

Charles Krebs
Posts: 5865
Joined: Tue Aug 01, 2006 8:02 pm
Location: Issaquah, WA USA
Contact:

Post by Charles Krebs »

Let me add a little more confusion to the subject of dedicated macro lenses....
The apparent size of the aperture doesn't decrease as much when you extend a 38mm lens 76mm from the film plane to achieve 2:1 image magnification as would a 100mm lens set at an aperture of f 5.6 and extended 200mm from the film plane to achieve 2:1 image magnifications. Diffraction effects become greater from the 100mm lens because the apparent size of the aperture decreases much more.
Hi Tom. You are trying to confuse me, aren't you!? :-k

The effective aperture (fe) is determined by magnification (m), “marked” f-number (f) and P, the pupillary magnification factor. It is not dependent on focal length. A symmetrical 100mm lens extended to 3:1 will give the same effective apertures (at a given “marked” aperture) as a 20mm symmetrical lens extended to obtain 3:1


The formula for a lens mounted “front forward” is:

fe=f((m/P)+1)


The formula for a lens mounted “reversed” is:

fe=f(Pm+1)

Where P is the ratio of exit pupil over entrance pupil. Nearly all special “photomacrography” lenses (and enlarging lenses) are symmetrical, so P=1 and the formula is simply fe=f(m+1))

A strongly retrofocus wideangle, if used reverse mounted will require somewhat less extension to arrive at a given magnification, but because of the effect of "P", there is no advantage as far as DOF or effective aperture in concerned.

Rik… the DOF you saw in your individual images is, as you suspect, probably the clue to the performance of the objectives. As you know, the f-number of photography lenses (even the special photomacrography lenses) is calculated using the ratio of focal length to aperture size, as if the lens were focused at “infinity”. The photomacrography lenses can’t be used that way, so the effective apertures are always much smaller than the marked aperture.

Now…
I would need to examine the spec for NA of a microscope objective and the “conversion” factor used to arrive at a comparison "equivalent" f-number. The logical explanation for the observed DOF and higher resolution is that the “effective” aperture of the objective is, in reality, far larger than a roughly "equivalant" photomacrography lens. (Perhaps the conversion from NA to f-number is for when the objective is used at its designed tube length – that is—already at “magnification". Don't know.)

It would be interesting to use the 10X and the Olympus 20/2 at the exact same magnification just to see how exposure times compare. This would allow you to do an empirical check of "effective" apertures between the objectives and macro-lenses. (Your Luminar would be a better focal length match for the 10X, but it is not marked in f-numbers)

rjlittlefield
Site Admin
Posts: 23561
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Post by rjlittlefield »

Charlie,

It's always like this, isn't it? Every detailed experiment produces some intriguing results that simply cry out for explanation -- typically requiring another experiment. A perverse form of "job security" applied to a hobby, I suppose. :?

Regarding NA versus F-number, I've been using the simplifying assumption that because this setup gives magnification much greater than 1, subject-to-aperture distance is pretty close to the lens focal length, in which case F-number and NA should be related as F-number ~= 1/(2*NA). Hence, my statement that NA=0.25 is roughly equivalent to f/2 (F-number=2).

That relationship is not exact, but at 10X and NA 0.25, I calculate the difference to be only about 15%, much smaller than what I think I'm seeing in the pictures.

I'll take some more measurements and recheck other stuff as time permits.

Thanks for thinking about this.

--Rik

Charles Krebs
Posts: 5865
Joined: Tue Aug 01, 2006 8:02 pm
Location: Issaquah, WA USA
Contact:

Post by Charles Krebs »

Rik… put this in the hopper as well… :wink:

I just finished a shot and decided to do a little test (why should you have to do all the work?!). I dug out the 35mm f2.8 Canon photomacrography lens, and the 4/0.16 Olympus S Plan Apo. (Olympus specs give the focal length of this objective as 36.71mm, so they are very close).

I photographed the exact same magnification/subject/light with each. The Canon was used a maximum aperture-- a “marked” f2.8. The Olympus was 2/3 stop faster based on obtaining nearly identical (as close as possible) histograms.

Using the f-number= 1/(2*NA) relationship (which is also the one I have seen) the f-number of the Olympus should be f3.125. In a “real world” shot, it behaved more like f2.25, nearly a full stop discrepancy. (This is based on the f2.8 marked on the Canon as being accurate).

I haven’t tried to figure out why the math does not seem correct, but the actual results seem consistent with your tests… higher resolution and less DOF for the objectives.

Charles Krebs
Posts: 5865
Joined: Tue Aug 01, 2006 8:02 pm
Location: Issaquah, WA USA
Contact:

Post by Charles Krebs »

Oh yeah.. also in the FWIW department... I wanted to mention that when looking at the focal lengths for the various Olympus 10X objectives (LB series, 160mm tube length) I noted 6 different 10X objectives, with focal lengths of:

16.92
18.96
17.69
16.9
18.98
15.69

Post Reply Previous topicNext topic