Focus shifting lenses

[Lindolfi]

Advertisement (gone after registration)

I was expecting a formula for the focal length and the distance h between principal pts. in terms of the two magnifications and object-to-image distances you measured: M1, M2, d1, d2.

 

 

 

#12, your expectations were correct:

 

Fill in M1, M2, d1, and d2 in formula (i) to get v2. Fill v2 and M2 in (ii) to get b2. Fill v2, b2 in (iii) to get f. Fill in d2, v2 and b2 in (iv) to get h.

 

 

I've put the Excel sheet on Google Docs, like K-H suggested : click, so you can work with it directly.

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[Guest #12]

Hey K-H

 

Here are the two books I use

 

Born and Wolf (click)

 

and

 

Hecht (click)

 

So what's on your bookshelf, K-H? I think I actually saw a picture of your bookshelf on another thread.

[k-hawinkler]

So what's on your bookshelf, K-H? I think I actually saw a picture of your bookshelf on another thread.

 

Hi #12,

 

Thanks. Ya, you did.

 

With regards to optics, off the top of my head, I have some introductory text books, like "The Feynman Lectures on Physics", volume I and in German, Optik und Atompysik by R.W. Pohl, Physik by Chr. Gerthsen, H.O. Kneser, and Optik by Arnold Sommerfeld.

 

I never got around to Max Born's Optics, but now I will.

 

Best, K-H.

[Guest #12]

Hi #12,

 

Thanks. Ya, you did.

 

With regards to optics, off the top of my head, I have some introductory text books, like "The Feynman Lectures on Physics", volume I and in German, Optik und Atompysik by R.W. Pohl, Physik by Chr. Gerthsen, H.O. Kneser, and Optik by Arnold Sommerfeld.

 

I never got around to Max Born's Optics, but now I will.

 

Best, K-H.

 

too funny,..that is really all I have, the two volumes of Feynman. Vol. 2 is a grad. text at AZ Optics and maybe Rochester. (Born is, too.) Has not really helped me with rangefinders!

 

thanks

[lars_bergquist]

No, it's not just a constant percentage—as the second graph in "#12's" post #26 clearly shows. If it was then the dashed lines were straight.

 

I was of course talking about the graph line's average slope, not its form.

 

The old man

[Guest #12]

...

 

That's interesting; thanks for your time posting the spreadsheet and the extra details.

 

I am wondering (apropos of a recent thread), how close did that 93.3 mm focal length come to the engraving on the barrel?

 

 

 

 

 

I should probably mention, the graph I posted is a little bit of a borrow from something Lindolfi posted on the rangefinder adjustment. Any mistakes are mine.

[Lindolfi]

Advertisement (gone after registration)

#12 I've played with the numbers to see what the sensitivity was of the results (focal length) as a result of the changes in the measurements (distance object to flange, object sizes and image sizes), so this is not the result from a measurement. When I find the time, I'll repeat it with the Macro Elmar, which has "98" on the barrel, so should have a focal length of 89.8 mm., to see how close it gets to that number and what the distance between the principal planes is.

[01af]

I've played with the numbers to see what the sensitivity was of the results (focal length) as a result of the changes in the measurements (distance object to flange, object sizes and image sizes) ...

I've played a little, too, and found out the formulas given in post #35 are a) correct and extremely susceptible to errors. In order to compute the principal planes' distance h with an error of less than ±10 %, you'd need to determine your distances and magnifications to at least four, better five or six significant digits. In practical terms, this is next to impossible (unless you're using sophisticated laboratory equipment, like e. g. laser interferometry to determine distances).

 

We don't even know the physical dimensions of the camera's sensor's active area, to begin with. We know, for instance, the M9's senor's size is somewhere close to 23.9 × 35.8 mm or so ... but that's by far not accurate enough to compute magnifications to a sufficient precision. We'd need to know the size with micrometer accuracy—do you know? I don't. Furthermore, we'd also need to include the lens' distortion into consideration when trying to compute magnification. Otherwise, we'd end up with different results for the focal length, depending on which area of the frame we've used to determine magnification.

 

So the formulas given in post #35 are very interesting academically, yet hardly useful for the hobbyist's practical intents and purposes. If you still want to try using them, be aware that they are extremely sensitive to the slightest variations of the input parameters, so don't expect to arrive at too accurate results.

[jaapv]

:rolleyes:It would not surprise me if a university professor of neuromechanics had access to highly sophisticated measuring equipment...

I've played a little, too, and found out the formulas given in post #35 are a) correct and extremely susceptible to errors. In order to compute the principal planes' distance h with an error of less than ±10 %, you'd need to determine your distances and magnifications to at least four, better five or six significant digits. In practical terms, this is next to impossible (unless you're using sophisticated laboratory equipment, like e. g. laser interferometry to determine distances).

 

We don't even know the physical dimensions of the camera's sensor's active area, to begin with. We know, for instance, the M9's senor's size is somewhere close to 23.9 × 35.8 mm or so ... but that's by far not accurate enough to compute magnifications to a sufficient precision. We'd need to know the size with micrometer accuracy—do you know? I don't. Furthermore, we'd also need to include the lens' distortion into consideration when trying to compute magnification. Otherwise, we'd end up with different results for the focal length, depending on which area of the frame we've used to determine magnification.

 

So the formulas given in post #35 are very interesting academically, yet hardly useful for the hobbyist's practical intents and purposes. If you still want to try using them, be aware that they are extremely sensitive to the slightest variations of the input parameters, so don't expect to arrive at too accurate results.

[01af]

When I come to think of it ... I don't even know if the lenses' focal lengths (i. e. those indicated by the tiny correction values on the lenses' barrels, next to the infinity symbol) are supposed to refer to the magnification at the frame's center or to the angle of view across the frame's diagonal. For zero-distortion lenses, the two options would be identical ... but our lenses are not distortion-free.

 

Does anyone know?

[jaapv]

Afaik the focal length is always measured in the optical axis.

[k-hawinkler]

Very interesting - but no surprise (post #49).

 

K-H.

[Xmas]

Hi

 

I note that for film Ms the actual frame size is variable dependent both on the lens and shutter frame separation from the film plane, this effect may also apply to M9, or does the sensor always limit the size of the photo?

 

The effect means that some wide angle lenses produce larger photos, with film cameras.

 

Noel

[Lindolfi]

Thanks 01af for your critical remarks, which are in line with what I found concerning the distance between the principal planes. There are more accurate laboratory methods to measure that distance. (click)

 

However, for the calculation of the focal length it is a very adequate method. I've run some sensitivity calculations and you can get the focal length of a 90 mm lens with an accuracy of 0.2% with household methods and equipment, and very careful alignment.

 

Pixelpitch is definitely no big issue, since the effective sensor size is known to the 0.1 mm and number of pixels is known to the pixel.

 

An excellent source of verification of focal length is to set the lens to infinity and take a photograph of the night sky with some star landmarks between which the visual angle can be found to a high degree of accuracy by using their sky coordinates. (You can use Stellarium (great software!) : click or an official star atlas)

 

Another method for measuring focal length of a lens is the Bessel Method. You take a picture of an object. Measure the distance between the sensor and the object (d). Now move the lens towards the object until you get a sharp image on the sensor again. Measure the displacement you need and call it a. The focal length is now f=(d - (a^2)/d)/4. With a Leica M, this is only practical with a Visoflex and bellows.

[Lindolfi]

A remark about the Bessel method: only applies to thin lenses the way it is normally posed, but can be extended with a second measurement to solve for the distance between the principal planes. That second measurement involves shifting the camera towards the object until the enlarged and reduced images coincide. The distance between object and sensor/film is then equal to four times the focal length plus the distance between the principal planes.

[Guest #12]

.... Unfortunately, a lens' principal planes' distance usually is not included in the tech specs provided to regular mortals.

 

 

 

...

 

Can you just ask them for this design information?

 

 

 

 

P.S. So what about you--what's on your optics bookshelf? Like the other poster, I am looking for some reading suggestions to get up to speed.

Thanks

[ho_co]

... The effect means that some wide angle lenses produce larger photos, with film cameras....

Good observation, Noel!

 

But it only applies to film:

• The film gate lies in front of the film, so a lens with its exit pupil further back can "look over" the film gate, just as one can see a wider angle of view by moving closer to a window.
  
• The sensitized emulsion extends over the entire surface of the film, so when you get that extra image spillover, there's something there that can use it. Extreme examples can even show the edges of the sprocket holes. (Very dramatic, if nothing else. )
  

 

With the digital sensor, Leica uses all but the very edge pixels. (Other manufacturers use quite a bit less.) And all the pixels are always in the same place, so there's no extra image area outside the sensor area, as there was with film.

 

Ah, for the good old days, when men were men and cats were black.

[pico]

[...] With the digital sensor, Leica uses all but the very edge pixels. (Other manufacturers use quite a bit less.) And all the pixels are always in the same place, so there's no extra image area outside the sensor area, as there was with film.

 

There is or was extra border/edge image information in some cameras beyond what is normally returned with DNG. I think the program to view them (DNG Recover Edges) is long gone, perhaps no longer needed.

 

Found it! Silly me. I never throw away anything.

 

Mac version

Wiin version (I have not used this one)

 

It is a drag-n-drop app. Drag your DNG image onto the file icon.

[Advertisement]

Hallo,

Look here: Interesting Leica gear

[Lindolfi]

The issue of pixel pitch and effective sensor size is interesting. According to most sources, the pitch of the M9 should be 6.8 µm and if you calculate the sensor length from that with the 5212 pixels that come out of a file, the effective sensor size should be 35.44 mm. However, that 6.8 µm may be a rounded number. 6.85 µm yields a sensor size of 35.70 mm.

 

There is an extra factor involved compared to film, and that is the cover glass of 0.8 mm thick over the sensor, which bends light rays inward at the edges.

 

So for measurements of focal length it would be very interesting to know the effective length of our 5212 pixels.

 

If we use the extra digits on a lens and use the indicated focal length as being correct, it is possible to measure pixel pitch from that.

 

Anyone who has more accurate data on the Kodak sensor?

[Lindolfi]

Thanks Pico. Just applied it and the long side of 5212 becomes 5216, so you gain about 0.027 mm sensor length with an M9 file.