Infinity objective on low-end zoom telephoto works fine

Have questions about the equipment used for macro- or micro- photography? Post those questions in this forum.

Moderators: rjlittlefield, ChrisR, Chris S., Pau

ChrisR
Site Admin
Posts: 8662
Joined: Sat Mar 14, 2009 3:58 am
Location: Near London, UK

Post by ChrisR »

Blame - the 135 was a beaten-up Nikkor f/2.8.

Pau yes you have understood correctly. I can't really explain it.
If you look in the Mitutoyo literature (as referenced by Charlie), you find
Image
(The red subscript "1" is apparently just an error.)

We know that a 200mm f/4 lens is likely to "clip" at about a 40mm image circle.
I assume that Ø2 would be 200/4 = 50mm
Ok, so you can work out what "L" must be if you believe their formula (1)
L = (Ø2-Ø1)·f2/Ø (1)

Using the Nikon 10x na0.25, and formula (2)
Ø1 = 2*20*0.25 = 10mm

If you put in (50 - 10)·200/40
to formula (1)
you get L= 200
Which means that the "tube lens" is appearing to be 200mm from the front end of the telephoto lens where the objective is placed.

I don't know if that's where the entrance pupil would be likely to be, or even if it's the entrance pupil position which matters here.

However, if you turn formula (1) around, at the limit you can say that the image circle

Ø = (Ø2-Ø1)·f2/ L
can be as big as you like, as long as you get L small enough.
So it seems to depend on the construction of the telephoto=tube lens.
If you put some numbers in, for a 135mm f2.8 with the entrance pupil in the middle, giving L =70mm , and using Ø2=135/2.8=48,

Ø =(48 -10)·135/70 =73mm.
That's a huge image circle. I'm sure the body of the lens would get in the way!

135mm lenses don't have to be optically compressed to make them compact, so perhaps the lens designer makes a more symmetrical lens, with the inlet pupil near the middle. This is speculation on my part, but seems to be plausible.

Rik will now tear this apart!
---


Don't rush off to buy short lenses - I haven't checked the IQ other than peering through the viewfinder.

rjlittlefield
Site Admin
Posts: 23543
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Post by rjlittlefield »

ChrisR wrote:Rik will now tear this apart!
I like to think of it as "clarifying", but I can certainly understand how it might feel otherwise. :)

The formula and diagram by Mitutoyo treat the tube lens as "thin", so that its entrance pupil is located 1 focal length from the sensor and coincides with the front surface of the lens. Probably neither of those is quite right, but it's close enough to work with.

When you swap in a telephoto, the computation gets more complicated. In fact it gets complicated enough that I would much rather ignore the computation and just concentrate on the concepts.

The concepts are pretty simple.

As shown in the Mitutoyo diagram, each point on the focused subject produces a bundle of parallel light rays coming out the back of the objective. All the bundles coincide with each other at the exit pupil of the objective, then diverge as you look farther back. To avoid vignetting, all of the rays in every bundle must reach the sensor. Bundles that are partially blocked by the aperture of the tube lens will darken on the sensor; bundles that are completely blocked will go black.

Measured at the specimen surface, the size of the non-vignetted field depends only on the objective and on the diameter and position of the tube lens's entrance pupil. Measured at the sensor, the size of the non-vignetted field also depends on the focal length of the tube lens.

All other things held constant, you get a larger non-vignetted field from a tube lens that has a larger entrance pupil, closer to the objective, and a longer focal length.

I think what ChrisR is telling us is that among the telephotos he has tested, the shorter lenses have entrance pupils that are enough farther forward to make up for their decreased diameter and for the shorter focal length of the lens.

Pau writes, and ChrisR confirms, that "a 100 mm tele has better coverage in full frame than a 200 mm". I have no trouble believing this, as long as we're just talking about vignetting.

But ChrisR's last comment strikes me as really important:
Don't rush off to buy short lenses - I haven't checked the IQ other than peering through the viewfinder.
The Mitutoyo and Nikon objectives are rated for something like 24 mm or 30 mm diameter fields, when used with their rated 200 mm tube lenses. Using them with a 100 mm tube lens instead of 200 mm, while simultaneously using a full-frame sensor with diagonal = 43 mm, is asking the objectives to cover a field roughly 3 times larger than they're rated for. They might do this while retaining high quality -- and it would be wonderful if they do -- but that's definitely the sort of thing that needs to be tested. I confess, I'm not hopeful.

--Rik

ChrisR
Site Admin
Posts: 8662
Joined: Sat Mar 14, 2009 3:58 am
Location: Near London, UK

Post by ChrisR »

Thanks Rik
I'm uncomfortable. I'm thinking there's more going on.
>> Would the entrance pupil of a 200mm telephoto really be expected to be 200mm (or so) from the front end of the lens? In some cases that would be beyond the back of the lens?

>> Is it so unreasonable than these relatively low magnification objectives should cover more than the image circle required? Plenty of macro lenses ( and the 10 buck JML) seem to manage!

>> the image quality degrades, as shown here http://www.photomacrography.net/forum/v ... hp?t=11981
but it doesn't disintegrate like a finite objective's does as here 7x .(sorry, link missing, I posted it previously but I'm too tired to find it at 1:45 am..)

I need to do some better test, but this stack doesn't look too awful I don't think (4x on 200) The center to edge comparison isn't anywhere near as strong.7MB
or this single frame - nb the in-focus plane drifts into the grooves..
This is (I think) a 10x on 135mm. About 7x ( and the ruler got bent :) )
another 7MB
Edited - I saw the ruler wrong
And edited again to add the 7x link
Last edited by ChrisR on Wed Jan 12, 2011 9:52 am, edited 4 times in total.

rjlittlefield
Site Admin
Posts: 23543
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Post by rjlittlefield »

ChrisR wrote:Would the entrance pupil of a 200mm telephoto really be expected to be 200mm (or so) from the front end of the lens? In some cases that would be beyond the back of the lens?
Beats me. Stop down the lens, look into the front of it, and see where the aperture appears to be.

For the Vivitar 200 mm f/3.5 in front of me right now, it's right about level with the sensor, several cm behind the lens mount. For a Sears 135 f/3.5, it's well in front of the lens mount. In terms of distance from the front surface, about 7 cm for the 135, about 15 for the 200. These probably vary a lot between lenses.
Is it so unreasonable than these relatively low magnification objectives should cover more than the image circle required? Plenty of macro lenses ( and the 10 buck JML) seem to manage!
I don't know what's "reasonable" with these new objectives. That's something we'll learn by experience.

But it's worth noting that the Mitutoyo 10X NA 0.28 is equivalent to an "ordinary" f/1.8 lens optimized to be diffraction-limited wide open.

Naively, I would guess that the designers paid a lot of attention to performance within exactly the specified field of view, and what happens outside that is just incidental.

My undergrad degree is in numerical analysis. In that field, what often happens with curve-fitting is that by adding higher orders you can get a closer fit within any specified interval, but the fit outside the interval is likely to get worse as a result. Reasoning by analogy, I'm not surprised when I hear that some apochromat works really well to the edge of its field and then falls off a cliff. On the other hand, that could equally well be just a coincidence. More experience needed...

I agree that the last image you linked looks pretty good across the whole field (concentrating on the details in the black grooves). I'm a little puzzled by the scale since I've never seen 1/64 inch ruled off in units of 5 and 10 tick marks. :? Any chance this is actually a cm rule, 21 mm across the frame for a magnification of 1.7?

--Rik

ChrisR
Site Admin
Posts: 8662
Joined: Sat Mar 14, 2009 3:58 am
Location: Near London, UK

Post by ChrisR »

Well it's morning proper, and I got my readers out :oops: It'a foot rule all right, but marked in 100ths of an inch,
so it's about 5.2 mm, about 7x, probably a 10x on a 135


Off to work....

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

Post by Charles Krebs »

Chris S wrote:
has it really been demonstrated that the regime in which the finite/0.30 shines is more limited than that of the infinity/0.25? Or are their useful ranges just different?

I find that the finite/0.30 works really well at 10x and over, whereas you've found that the infinity/0.25 works really well at 10x and under--or so it seems. For much of the range you describe for this lens, I'd be using my 4x/0.20.
rjlittlefield wrote:
The concept of useful range appears when we intersect what the lens can do with what the sensor needs. In the case of the finite/0.30, there's no leeway. The T1i sensor apparently captures at one time the largest available rectangle and full resolution on subject at 10X; operating higher or lower throws away some of one without a compensating gain in the other. But in the case of the infinite/0.25, there does seem to be some leeway; operating at lower magnification gains a larger field with some loss of resolution on subject, while operating at higher magnification (up to 10X) gives up field size to get more resolution....

I'd like to see some quantification of "works really well at 10x and over". When you run the 10X/0.30 at higher magnifications, do you actually pick up more detail on subject? If so, then how much, and using what sensor? When I was using a 6 megapixel camera, I'm pretty sure I got more detail by pushing beyond 10X. With 15 megapixels, I'm not.
I think there is a little "leeway" toward somewhat higher magnifications with the 10/0.30. The 10/0.30 could (theoretically) give 20% higher resolution. The pixel size of the T1i sensor is 4.7 micron. The resolution limit of a 10/0.25 is 1.34 micron. The resolution limit of a 10/0.30 is 1.12 micron. The resolution limit of the 10/0.30 will be 1.34 micron when it is used at 12X. So when the 10/0.30 is used at 12X you are getting the same number of pixels across this minimum detail as when the 10/0.25 is used at 10X (2.85 pixel per detail... which is very good, but not the 3 to 3.5 pixel per detail I've seen suggested as most desirable). Granted, there are a whole lot of "asterisks" and "other considerations" when playing with these number this way, but it does seem to match up with my actual observations on my 50D (also 15Mp). Used at 10X, The CF N 10/0.30 does show a little more detail than I see with the Mitutoyo 10/0.28, the Nikon CF M Plan 10/0.25, and the Nikon CF M Plan SLWD 10/0.23. (I don't have the CFI objective that is the subject of this thread). But we are really into pixel peeping!

Some might not consider the difference between 10X and 12X to be that big of a deal, and prefer the superior ability to go to a lower magnification that the CFI seems to present. Just depends on what you are looking for I guess.

This shows the difference between 10X and 12X (red box)

Image

Blame
Posts: 342
Joined: Fri May 14, 2010 11:56 am

Post by Blame »

Ok. I won't rush to buy a 135mm... not for a few days at least.

My hopes have been built up by a quick test. A tamron 90mm macro with a sony a350. Sensor is 16x24mm. Also we now know that the mitutoyo 10x is at least sharp to 3.7mm diameter thanks to morfa's brilliant "cheap tube lens"

It was a very limited test owing to lighting problems - I am awaiting a flash extension cable. Until I get it I have to mount the flash on the camera. The result is that the camera topples over unless I hold it in position. Not optimal.

The point is I saw no serious vignetting, and the rig is equivalent to a 135mm and my full frame camera. That would give about 5.3x3.5mm. Morfa's results show an image circle of at least 4mm, somewhere around 60% of the total area.

Unless morfa has truly squeezed the very last bit of width out of the objective, 5.3x3.5mm should be interesting. It might have blurred corners, but it will be great for portraits. All I need is to find a promising young insect, and make her a star.

Anyway, this goes back to a very old argument. A lens or objective has an image circle, limited by blur. The sensor (and before that the film) is a rectangle. Should the rectangle go inside the circle, or the circle inside the rectangle? I could find convincing arguments on both sides, economic, artistic and technical.

AndrewC
Posts: 1436
Joined: Thu Feb 14, 2008 10:05 am
Location: Belgium
Contact:

Post by AndrewC »

Blame wrote:....
Anyway, this goes back to a very old argument. A lens or objective has an image circle, limited by blur. The sensor (and before that the film) is a rectangle. Should the rectangle go inside the circle, or the circle inside the rectangle? I could find convincing arguments on both sides, economic, artistic and technical.
The strongest (IMO) argument would be if you want to tile images. Objectives are pretty much telecentric so you don't need complex nodal rotation mechanics.
rgds, Andrew

"Is that an accurate dictionary ? Charlie Eppes

Blame
Posts: 342
Joined: Fri May 14, 2010 11:56 am

Post by Blame »

Another thought.

I had been assuming that if you bolted the objective right up close to a camera lens, and set the aperture wide open, then any vignetting was going to be the fault of the objective.

That doesn't always seem to be true in practice, perhaps particularly with zooms. I have been very puzzled.

Maybe the answer is in the "right up close". A camera lens is a thick lens. It could be that the surface of the front lens is not the effective front, and the light from the objective has space to spread out before it meets the effective front of lens.

ChrisR
Site Admin
Posts: 8662
Joined: Sat Mar 14, 2009 3:58 am
Location: Near London, UK

Post by ChrisR »

It could be that the surface of the front lens is not the effective front
It's not, Blame. The Entrance Pupil is what matters. That can be "anywhere" including way back into, or even behind the lens.

Blame
Posts: 342
Joined: Fri May 14, 2010 11:56 am

Post by Blame »

AndrewC wrote:
The strongest (IMO) argument would be if you want to tile images. Objectives are pretty much telecentric so you don't need complex nodal rotation mechanics.

Could you explain a little? I am sure what you just said is deep and meaningful, but sadly it assumes way to much prior knowledge on my part.

Blame
Posts: 342
Joined: Fri May 14, 2010 11:56 am

Post by Blame »

ChrisR wrote:
It could be that the surface of the front lens is not the effective front
It's not, Blame. The Entrance Pupil is what matters. That can be "anywhere" including way back into, or even behind the lens.
Ouch. A vote for thin lenses - at least they are a bit more predictable.

Now I am having bad thoughts another assumption. Stop down any decent lens sufficiently and it becomes diffraction limited. A check on a lens review site showed me a variety of macros had only a few percent variation at f/22, which supported the idea.

However we are talking thick lenses again. If the light spreads out before it meets the effective front of the lens, then some or all of the blur of a wide open lens could come into play. Ouch. If so, no wonder corners have disappointed sometimes.

I am beginning to wonder if thin lenses might be the way to go for sharpness.

AndrewC
Posts: 1436
Joined: Thu Feb 14, 2008 10:05 am
Location: Belgium
Contact:

Post by AndrewC »

Blame wrote:
AndrewC wrote:
The strongest (IMO) argument would be if you want to tile images. Objectives are pretty much telecentric so you don't need complex nodal rotation mechanics.

Could you explain a little? I am sure what you just said is deep and meaningful, but sadly it assumes way to much prior knowledge on my part.
"Tiling" basically means stitching together several adjacent images to make one large one. Elf does a lot of it on this forum.

If you Google "Panoramic stitching" you can find lots of info on the net. One of the problems with tiling images is parallax errors. What this means in real life is that as you move the camera position, foreground objects appear to move relative to background objects . For "normal" lenses, this can be avoided by moving / rotating the camera around its nodal point. Telecentric setups however don't show the problem.
Search on this forum and you'll find lots of hits for "telecentric". The best core thread is probably this one:

http://www.photomacrography.net/forum/v ... elecentric

hope that helps :)
rgds, Andrew

"Is that an accurate dictionary ? Charlie Eppes

ChrisR
Site Admin
Posts: 8662
Joined: Sat Mar 14, 2009 3:58 am
Location: Near London, UK

Post by ChrisR »

However we are talking thick lenses again. If the light spreads out before it meets the effective front of the lens, then some or all of the blur of a wide open lens could come into play. Ouch. If so, no wonder corners have disappointed sometimes.
Well it prompts a question - suppose you make a hole in the middle of some black paper, say 5/16ths = 8mm diameter, and put it over the front of a telephoto. What comes out from that? Are the corners bad?

Edit _ I wrote "font" of a telephoto. That's obviously near the entrance pulpit........
Last edited by ChrisR on Mon Jan 17, 2011 9:42 am, edited 1 time in total.

Blame
Posts: 342
Joined: Fri May 14, 2010 11:56 am

Post by Blame »

ChrisR

What an easy test. Thanks.

Post Reply Previous topicNext topic