DOF and Sensor Size

[PaulFurman]

Here's an interesting discussion on alt.photography:
Re: DOF and Sensor Size

The premise by my reckoning is that smaller sensors actually have an advantage for macro because they are not working in the macro realm as soon as a larger sensor (where effective aperture can limit things due to diffraction), assuming you fill the frame with the same subject. I suspect that in the real world this isn't really relevant since there are only so many lens possibilities but maybe there is an advantage one way or the other. In any case it's an interesting observation!

Previously I saw smaller sensors as a problem due to diffraction but this might provide a theoretical advantage. I think there is still the limitation that smaller sensors cannot make as large of prints. The discussion starts with the premise that he will ignore engineering limitations and that he will assume the possibility of making wider aperture lenses for the small sensor camera such that the depth of field is identical to a larger sensor setup. His goal is more to achieve shallow DOF for non-macro portraits and in a sense, that's the advantage I'm suggesting too; that retaining shallow DOF staves off diffraction losses, allowing better sharpness.

Sorry, I know this is all very confusing and I'm probably not helping things. Think of it more as just a mental exercise than having practical use and it might be more fun

One thing to clarify, that I commented on in the discussion: you'll notice he uses the term 'effective aperture' to mean something more like 'effective depth of field' relative to the image frame.

[elf]

It's not viewable unless you have a Google account.

[rjlittlefield]

As it happens, I do have a Google account so I read the thread. Well, maybe "skimmed" would be more accurate.

That discussion strikes me as a familiar situation. The OP posts an analysis that is essentially correct but is based on mathematical analysis, and the thread proceeds through several rounds of questions and dissent by people who don't follow the math, haven't done careful physical experiments either, and object on the basis of intuition and informal thought experiments, perhaps augmented by inappropriate or erroneous math of their own. Having been through that experience several times myself, I am sympathetic but reluctant to spend much of my own time trying to clarify the situation once again.

Nonetheless, since the question has been asked, I will try to summarize. It will be good exercise to see how far I can boil this down this time.

Let us agree to shoot "equivalent images" using sensors of two different sizes. "Equivalent images" means that the same size subject fills each sensor, with the same perspective and DOF, under the same illumination, for the same exposure time.

Under those conditions, it turns out that the key parameters are as follows:

- The lens entrance pupils must be at the same location and have the same diameter (not the same f-number).

- The effective f-number scales in a simple way: in direct proportion to the sensor size.

- The lens focal length is larger for the larger sensor, but not in any simple way. For infinity focus, it is in proportion to the sensor size; at high magnification it is almost independent of sensor size; at moderate magnifications the relationship is easily computed but not easily explained.

- The sensor must be set to an ISO rating that is in proportion to the sensor area (linear dimension squared).

When these things are done, then it turns out that diffraction and image noise due to photon sampling also end up being identical for the two sensors, in addition to the subject size, perspective, DOF, and exposure time that were held constant by specification.

A good intuitive way of thinking about this is that because each lens has the same entrance pupil, they admit the same light rays, the same number of photons with the same distribution, and so on. Sometimes I phrase this as "same light, same image".

The one key effect of sensor size is to limit the maximum diameter of the entrance pupil. Under "equivalent images" analysis, you inevitably end up using a shorter focal length lens with a smaller sensor. The shorter focal length in turn implies a smaller maximum diameter entrance pupil. Even when both lenses can reach the same diameter entrance pupil, the shorter lens will be working at a wider f-number and thus will be more vulnerable to aberrations.

As a result, in this analysis advantage always goes to the larger sensor, if there is any advantage at all. Barring artificial limitations in the diaphragm, you can always stop down the larger sensor's lens so as to be equivalent to the smaller sensor's lens. The reverse is not true -- the larger lens of the larger sensor can be adjusted to give more capability with wide apertures, to get less DOF and more sharpness at the same perspective. The larger sensor can also produce a quieter image by capturing more light, through using either brighter illumination or a longer exposure. (Any of these latter changes takes one outside "equivalent images" analysis.)

There are, of course, other considerations that may make a smaller sensor more appropriate in some situations. Their naturally short lenses provide great tools for doing "wide angle macro" work at short distances, a capability that is made even better by the way that automatic focus works under these conditions.

Think I got all that right...no doubt somebody will tell me if not...

I hope this is helpful. Please pass along my support to Pete at alt.photography.

--Rik

Edit: to expand the description of advantages.

[DQE]

Rik,

Thanks for providing such a well-written, "all in one place" explanation of this commonly misunderstood set of parameters and components.
----------

I recently enjoyed a photographic exhibit at our local art museum, by the "Group f/64", a well-known if not legendary group of photographers from earlier decades.

https://secure.wikimedia.org/wikipedia/ ... Group_f/64

Quotes from the above web page:

"The term f/64 refers to a small aperture setting on a large format camera, which secures great depth of field, rendering a photograph evenly sharp from foreground to background."

"The name of this Group is derived from a diaphragm number of the photographic lens. It signifies to a large extent the qualities of clearness and definition of the photographic image which is an important element in the work of members of this Group."

Am I correct in understanding that the huge size of their "sensor", a 4x5 inch or larger film, plays a role in their use of f/64 to maximize detail? In other words, would f/64 only work well, without grossly excessive diffraction, when using such a huge sensor? I am not familiar with their lenses but would guess that the lens elements are fixed, relying on a bellows to achieve focus. Of course they would also be able to use tilt and shift to further change their effective DOF and make the optical analysis even more complex.

As usual, ignorance limits my understanding of their optics and images. With the many parameters that must be specified and fixed in order to make the specific comparisons provided in your post, I fear that simply saying that they maximize their detail and DOF by using f/64 is not technically complete.

In any event, the exhibition of their (contact) prints was very enjoyable, as was the description of their art and technology.

Perhaps they should have used say f/8 and learned how to focus stack! (insert many grins here) Perhaps we will rediscover 4x5 or larger camera sensors when technology can provide them at an affordable price. Terapixel images anyone?! (more grins)

[rjlittlefield]

Thanks for providing such a well-written, "all in one place" explanation of this commonly misunderstood set of parameters and components.

Thanks. The explanation gets a little more concise each time I write it. Practice makes perfect...or at least better...

Am I correct in understanding that the huge size of their "sensor", a 4x5 inch or larger film, plays a role in their use of f/64 to maximize detail? In other words, would f/64 only work well, without grossly excessive diffraction, when using such a huge sensor?

Correct.

To keep things simple, let's imagine that their film was actually 4x6 inches. That's almost exactly 4 times larger than full-frame 35 mm film. When shooting landscapes, effective f-number is just equal to the lens's nominal f-number. So, analyzing in terms of equivalent images, f/64 on 4x6 inch format gives the same DOF and diffraction blur as f/16 on 35 mm format. If you stopped down to f/64 on 35 mm film, you would get 4 times the DOF, but only at the cost of 4X bigger diffraction blur.

Group f/64 certainly did make notably sharp images. Compared to other images shot on large film, the higher sharpness was due to relatively smaller apertures, sufficient to get more DOF while still not suffering from diffraction blur. Compared to images shot on small film at equivalent apertures such as f/16, the higher sharpness was due to secondary characteristics such as relatively flatter film, thinner emulsions, and smaller grain size, all measured with respect to the image size.

Most of the secondary advantages of large format have disappeared with the advent of solid state sensors. However, large sensors still have the advantage of being associated with larger lenses, which allow higher resolution if it's OK to sacrifice DOF in proportion.

You're right, a focus-stacked f/8 image from a large format high resolution sensor would be a feast for the eyes.

--Rik

[Craig Gerard]

You're right, a focus-stacked f/8 image from a large format high resolution sensor would be a feast for the eyes.

How many slices and at what magnification?

Craig

[DQE]

You're right, a focus-stacked f/8 image from a large format high resolution sensor would be a feast for the eyes.

How many slices and at what magnification?

Craig

I don't recall anyone at the macro forums I participate in trying macro or high-mag macro with a very large-format camera (i.e., 4x5 or larger). The only example that comes to mind is a macro photography book from the film era where the author used a Hasselblad (or some other brand with this format). Can't recall the author's name but I could probably find the book if anyone's interested.

Here's a link to the Hasselblad 50 megapixel camera back, about 30K US dollars:

http://www.hasselbladusa.com/products/d ... fv-50.aspx

There's something strange about a camera body costing more than a nice car. Lenses are of course quite a bit extra.

[PaulFurman]

The Nikon Multiphot was designed for large-ish format macro work.

Bjørn Rørslett's page:

Macro-Nikkors are excellent performers and within their specified magnification range will cover up to 4x5" format with ease.

BTW, Thank you Rik for commenting!

[SONYNUT]

are you trying to say this is a bit much..lol

[Craig Gerard]

SONYNUT,

Is that piece operational......fire at will!

Which PhaseOne digital back is that...a P65+ P45+? I can't quite make out the details in the image.

It reminds me of this:
http://luminous-landscape.com/reviews/c ... view.shtml


Craig

[SONYNUT]

P45..

i was working on stills for new computer servers.
15 degree stepped shots all around

[DQE]

Hmmm....there used to be a scanning slit, step-and-repeat camera by Leaf. I think it's not around now but am not sure. I am not sure if Sinar supports such a camera.

Seems like you could end up with a gigantic digital photo that way, although the relatively slow speed of the scanning slit mechanism has limitations, of course, unless the subject is static. Gigapixel images might result!

[elf]

I found this a few years ago: http://www.photomacrography.net/forum/v ... highlight=

I think macros are far simpler to do with a DSLR. Can you imagine dragging an 8x10 around in the field with a 360mm lens (equivalent to 50mm lens on 35mm). Besides the weight and bulkiness, you would need nearly 3/4 of a meter of bellows to get to 1:1.

A quick check on film costs for 4x5 (just purchase and developing) shows about $5.00 per image. My Heather pano would have cost about $12,500, but would have printed really huge

[ChrisR]

My Heather pano would have cost about $12,500

Well there you go - $30,000 for a digital back is remarkably cheap, if that's the work you're using it for.
If "1:1" means macro then I've done professional large format macro work. I can't actually remember how large, the "camera" took rolls of Cibachrome as big as was needed. I think it went up to 900 mm wide. My input was little more than to press the button and listen to the whirring.
I had the sad task later, of dismantling it to get the only valuable parts out - which were enormous Nikkors. All the lovely rack and pinion gear got melted down for scrap.
The "f64" group, if I remember right, was generally using 10"x8" glass plate cameras. It was said around the time how remarkable it was, that Ansell Adams could get comparable quality from a 5" x 4" plate, enlarged, as was possible from a 10" x 8" contact print. That says a lot about the shortcomings of the business at the time, surely, if anyone would expect to tell the difference? I wonder if part of their reason for stopping down as far as f/64 was just to get decent performance out of their lenses.

I have used 10x8 on location, though not in any "fields"(!) for architecture and scenic. DOF is a very significant issue, and if you want a lot, it's a definite problem. Even at small enlargements, say a double magazine spread, it shows. Having an Art Director on the phone saying it's a "pity", expecting you to spend another day taking two pictures, is ! The smaller the format, the less that would happen. While it's possible to do the maths and show that DOF shouldn't be more of a problem with bigger formats, going through the practical mechanics "in the field" doesn't teach that.
Still, the joy of digital with stitching and stacking and Photoshopping means we can overcome some of the laws of physics, and some of the discomforts of lugging plate cameras about, and produce images that were practically impossible a few years ago. It's all marvellous and I can't wait to retire and have some time!

[ChrisR]

I just had to copy an old family photo. On the original print, a relative's eyes have nice clear, round pupils in them:

I'm short of pixels. Perhaps with 8x8 pixels covering the 2x2 where the pupils should be, I could have made a good copy.
I used a 12 MP camera, but needed then a 4 x 4 x 12 = 192 MP. To be fair the corners aren't quite as clear, so maybe less, perhaps only 100MP?

The print is 12" x 10", and would have been a contact print, because that's what they did when it was taken, when Bessie was sweet 16, in 1897.