Optimum bellows length

[rjlittlefield]

Why has this thread gone quiet?

Perhaps the main points had been covered?

I see that one of lauriek's questions remained unanswered:

I have a feeling that to get any benefit from a bellows much longer than that may require a substantially larger sensor

Perhaps a better way to look at this issue is that a substantially larger sensor will require a proportionally longer bellows to get essentially the same benefit.

The basic rules are simple. You want to fill the sensor with the subject field you care about. You also need to limit the angle of view, because lenses only deliver a sharp image within some cone that defines their coverage. Typical coverage for a good macro lens at low power might be 20 degrees; for a microscope objective it's more like 10 degrees.

Suppose you want to shoot a 10 mm subject.

Then putting the rules together, assuming 20 degrees angle of view:
- a "1.6 crop factor" sensor will require 2.3X with the lens at least 65 mm from the sensor;
- full-frame 35 mm needs 3.6X and at least 102 mm;
- 4"x5" sheet film needs 12.7X and at least 360 mm; and
- 8"x10" film needs 25.4X and at least 720 mm.

The lenses that go along with these widely different extensions are actually quite similar. They range only from 20 mm for the APS-C to 27 mm for the 8"x10".

Back in the old days, it was common to use large format film for shooting small subjects. This involved putting short lenses on long extensions. The main benefit of that approach was greatly improved gradation, compared to using available small-format films. There was not a corresponding improvement in resolution, because the small effective f-numbers in the large-format setups introduced correspondingly large amounts of diffraction blur. With the lens properly set, diffraction blur in the final image is essentially independent of sensor size.

These days, digital sensors of almost all sizes provide the smooth gradation that used to require large format film. But the digital sensors are much smaller, so the need for long bellows has dropped off accordingly.

--Rik

[DaveW]

Using the Nikon PB-6E extension bellows on the PB-6 bellows, Nikon's bellows system, originally designed for 35mm film cameras (now FX being the digital equivalent), go out to 438mm of extension. Which would seem a bit pointless a camera firm making them if such extension was unusable?

http://www.mir.com.my/rb/photography/ha ... index3.htm

DaveW

[rjlittlefield]

Nobody said "unusable" besides you, Dave.

That 438 mm of extension would be great with a reversed 135mm f/4.5, giving around 2.5X magnification with a working distance of 190 mm and an effective aperture of f/16.

Completely usable, especially on full-frame film. It's just a pretty specialized application.

--Rik

[DaveW]

I did not say anybody said they were unusable Rik, but got the impression they were implying there was a lesser maximum usable bellows length than the Nikon set-up.

As to how many sets you can couple together on the old 35mm film format. I know many have Lester Lefkowitz's book "The Manual of Close-Up Photography", so see his combination of three sets of standard bellows plus extension tubes on p.86.

Note he also makes his own ultra long extension tubes from lengths of plastic drain pipe on p.88. As he claims his centre drainpipe extension tube illustration is 27" long, that makes the longest one in the same illustration over 4ft long!

DaveW

[puzzledpaul]

No comments on the use of multiple bellows rigs, other than to say I tried a setup with 2 once, when I was using the old D30 - and knew even less about stuff then than I do now

Thing I find interesting about bellows, is that many ppl who seem to have bought them in film days to 'try macro' hardly ever used them - hence a typical ebay listing stating 'hardly/little/never used etc ... and the 'shot in the arm' their use has been given by the arrival of digital and the associated reduced (running) costs of experimentation...

... although it does make me smile when I see in such (ebay) listings, the reason for sale of said bellows is that the user has 'gone digital' ...

Still, their loss is our (bd members) gain

pp

[rjlittlefield]

I know many have Lester Lefkowitz's book "The Manual of Close-Up Photography", so see his combination of three sets of standard bellows plus extension tubes on p.86.

Note he also makes his own ultra long extension tubes from lengths of plastic drain pipe on p.88. As he claims his centre drainpipe extension tube illustration is 27" long, that makes the longest one in the same illustration over 4ft long!

That's an excellent reference, Dave.

For those who don't have a copy of the book, let me quote the introduction of Lefkowitz's discussion (page 85):

SUPER EXTENSION

There are special cases when extremely long extensions are necessary. Examples include: (1) a need for unusually long working distances and high magnifications, such as when photographing dangerous or remote industrial or medial procedures, or when recording behavior of misanthropic or skittish animals and other biological species; (2) a desire for high-resolution photography (scientific, forensic, medical, or professional) at magnifications greater than 10X (10:1) without buying the expensive true macro lenses.

It's important to note that the previous page is titled "WARNING: DIFFRACTION!" (in caps and boldface, just like this).

The example on page 85 shows an 8X image that is already quite soft even at its published size of 3 x 4.5 inches. The example on page 86 shows a chain of 3 bellows, with a 105mm lens on the front of it. The caption also describes the picture next to it, which was shot with a 200 mm lens. That picture is of the "Star of India -- world's largest blue star sapphire", shot in situ at the American Museum of Natural History, sitting "6 inches behind a very thick slab of impenetrable glass", and requiring a working distance of 28 inches even at 0.5X magnification. The example on page 88 shows a 22X shot of a dollar bill. That image is reproduced at 2-1/4" square and is barely sharp even at that small size.

What Lefkowitz is demonstrating is the use of long bellows to produce images that are large enough to fill a 35mm frame, regardless of how fuzzy those images might be.

With film, there are compelling reasons to do this. Projecting a cropped slide doesn't work very well, and with any film, using the whole film area produces significantly better gradation than using only a part of it.

With digital, the advantage is quite a bit less because resizing is easy and the images are not so noisy to start with. Given that the underlying image is fuzzy, it's hard to tell the difference between 2000x1500 pixels rescaled from a full-sensor image of 4000x3000, and 2000x1500 pixels simply cropped from an image at half as much magnification.

Finally, Lefkowitz's book deals primarily with 35mm film -- a 36 x 24 mm sensor. But elf asked about the Olympus e330, which has a sensor barely half that big, only 18 x 13.5 mm. The acceptable size of the Airy disk, and the corresponding bellows length, simply scale to match -- half as big, half as long.

The bottom line, in the context of elf's question, is that long bellows could be very useful if he wanted to make high magnification fuzzy pictures with really excellent gradation. Barring that, shorter bellows will work just fine.

--Rik

[NikonUser]

Thought I would add a practical experience to the above discussion.
Top: Tip of FW of noctuid moth - full frame, width of frame = 2.5mm (=9.5x mag on sensor)
Middle: 800 px selection of actual pixels
Botton: Set up

Schneider Apo 40/2.8 @ f/2.8. 380mm total extension, 59 frames @ 0.005mm, HF stack. Working distance from lens collar to moth is 30mm. When photographed the paper lens collar was placed around lens.

I could have got the same mag using a 10x ELWD microscope objective but the working distance (10mm?) would have been considerably less.

[elf]

NU: Can you post a similar image using the 10x ELWD microscope objective?

I ended up with 210mm bellows draw and I could increase that with other extension tubes to 315mm. I can't say that I have any usable images yet with the maximum draw.

This one is with reversed El-Nikkor 50mm 2.8 at 175mm bellows draw:

[DaveW]

Elf,

Your image appears to be at an angle to the lens? At that sort of magnification it will need to be exactly parallel with the front element. That is unless your bellows has movements and allows you to use the Scheimpflug principle. Otherwise you will need to photostack to cover the depth of field required.

DaveW

[rjlittlefield]

NikonUser, great illustrations!

Here's my take on this setup.

You're running an f/2.8 lens at 9.5X. Assuming no issues with pupilary magnification factor, that turns into effective f/29 from the standpoint of the camera.

Your camera, I believe, is an Nikon D2X, 4288 pixels across a 23.7 mm sensor.

As shown at http://www.cambridgeincolour.com/tutori ... graphy.htm, that combination makes a pretty large Airy disk. Most of its energy is spread over an area about 3 pixels across.

That's quite a lot of diffraction blur. Aggressive sharpening can back out some of the blurring, but even so, you're probably left with empty magnification.

To determine how much empty magnification, I ran an experiment with your second picture, the "800 px selection of actual pixels".

The experiment was this. I pulled the posted image into Photoshop and cropped it to 798 pixels square to make the dimensions exactly divisible by 2 and by 3. Then I made a copy of the image, reduced it to 50%, expanded it by 200% (thus restoring the original size), and layered it over what you posted. The result was a slight reduction in visible detail. I did the resizing again, this time reducing the image to 66.66666% and expanding it by 150%. The result the second time was no visible difference from your posting except for a slight reduction in pixel noise. Initially I thought there were no differences at all and that I had botched the experiment. To see that in fact the images were not the same, I had to have Photoshop explicitly compute the differences of pixel values, and multiply them by 10X.

The only way I see to interpret these results is that the pictured setup has at least 50% more magnification than is justified by that lens & sensor.

The extra magnification is not quite a complete loss. If you reduced the extension to give only 6.3X instead of 9.5X, your working distance would increase slightly to maintain focus. Because the diameter of your lens is fixed, the increased working distance would slightly reduce the NA of the lens, which would slightly reduce its resolution on the subject. However, that reduction is less than 5%, and in exchange you would capture over twice the subject area. (1.5*1.5=2.25)

This issue of bellows length is packed with tradeoffs.

If you care about an image area of exactly 2.5 mm wide, and you need 30 mm working distance, and you have this 40 mm lens handy, then 380 mm extension is just the ticket.

Or if you need the working distance, but there's added value in imaging twice the area with a 5% resolution loss, then a shorter extension with the same lens would be better.

Or if you need more working distance, then you need an even longer extension -- and a really good longer lens too.

Or if working distance is critical, and you want maximum resolution, and you have many dollars to spend, then you would do better to buy an ULWD objective. At 10X, an ULWD objective will have about 30 mm working distance also, and almost twice the resolution of that Schneider Apo. But it may also cost close to $1000, unless you're very lucky with some auction.

Or if the working distance is not so critical, you could use one of those Nikon CF Plan objectives and again get almost twice the resolution, at the cost of working at 9 mm instead of 30 mm.

Choices, choices... They're a burden sometimes.

--Rik

PS. Do you know what species the moth is? It looks very much like one of the Yellow Underwings, Noctua pronuba perhaps?

[NikonUser]

The shot of the noctuid wing (Syngrapha octoscripta) with that ridiculous bellows extension was taken in desperation. A few days before you explained why extra long bellows were unnecessary for the setups most of us use I made a non-cancellable order for a Nikon PB-6E bellows extension. So when it arrived I just had to see if it was useable. If only I knew then what I know now ......

Anyway, thanks Rik for your detailed explanation and possible options.
As I am still using a single flash I do like 'long' working distances and so 30mm is preferable to 9mm; and 25cm extension (to give 6.8x mag) is preferable to 38cm. I guess I will re-box the PB-6E and keep a look out for a ULWD.

[rjlittlefield]

If only I knew then what I know now ......

You got off cheap. I spent hundreds of dollars to acquire one of the legendary Luminars, only to discover that it gave less resolution than an inexpensive 10X objective I already owned! I was in denial for quite some time over that one...

--Rik