Panoramic shots with rangefinder; nodal point ?

I am trying to learn how to take panoramics with a rangefinder.

 

[With an SLR this is simple: shift the camera assembly back from the center of

the tripod mounting screw until the foreground and background do not shift

optically in the viewfinder when rotating the camera horizontally around the

tripod head: This indicates that the nodal point has been found for the lens in

use and that one can shoot a number of shots rotating the camera left to right

and will have an easy time to stitch later.]

 

How can this be done with a rangefinder camera? Of course, I could take a few

dozen shot pairs with each lens, offset a few mm more from right above the

tripod screw and watch for a pair of pictures where the fore-and background

keep still and use that mm setting to off-set the camera with that lens for

panoramics. What a waste of film, though.

 

However, the lens manufacturers do provide us with detailed optical data of

rangefinder lenses, Zeiss in particular:

 

For the Planar G f/2 45 mm for example, www.zeiss.com says (after a search for

Contax G lens on their website)) :

 

Entrance pupil position: 24.5 mm behind first lens vertex

 

Exit pupil position: 25 mm in front of last lens vertex

 

Position of principal plane H : 30.3 mm behind first lens vertex

 

Position of principal plane H': 18.4 mm in front of last lens vertex

 

Back focal distance: 28.5 mm

 

Distance between first and last lens vertex : 36.7 mm

 

Focal length: 46.9 mm

 

Flange to film distance: 29 mm

 

Somewhere in this data the distance of the film plane to the nodal point of this

lens must be hidden. How to retrieve it?

 

Can anyone help me uncover this information, please. What is the optical

definition of "nodal point" as used in panoramic shots?

 

Thank you very much!

Interesting question ....

 

... whats a nodal point ?

To Frank:

 

You can guesstimate and get it pretty close. The nodal point is going to be roughly halfway between the lens and the film plane. The fewer number of items in the foreground, the less critical finding the nodal point perfectly will be.

 

To the second poster:

 

Try Google for a better explanation, but...

 

Remember the lens diagrams from high school, that showed a lens projecting an upside-down image onto a plane? Recall that they showed the light rays criss-crossing about halfway between the lens and the film plane - ie, rays entering at the top of lens ended up bottom of the image, and vice versa.

 

The nodal point is where all the light beams criss-cross.

 

For panoramic shots, where you will splice several photos together, for best results you need to rotate the camera around the nodal point.

For non-retrofocus lenses, it should be one-half the focal length from the film. Or should the nodal point be at the focal point of the lens, which is of course the focal-length from the film plane (non-retrofocus).

 

The problem is, of course, the wide angles are now all retrofocus. Unless you go buy a Super-Angulon or Hologon. Or the cheap Jupiter 35mm.

 

A polite letter to Leica in Solms might provide the desired data. Leitz did make a panorama head for the Barnacks, although it didn't have nodal point correction.

 

Or do it experimentally without film. Open back, shutter open with cable release. Mount camera to swivel at test nodal point. Point camera straight at an object. Now sight, and see that the aerial image of the object doesn't move as you swivel camera. Leave your head (or a camera with a lens) stationary, don't swivel with the camera.

Hi Frank,

you could unhinge the back and tape a cut-down slr screen to the window and then

proceed the way you would with an slr. A dark-cloth may be necessary, though.

I *guess* that the nodal point is moving as you focus so.....

 

Cheers, Peter

> What is the optical definition of "nodal point" as used in panoramic shots?

 

It's the point where the optical axis intersects the principal plane H.

 

> Somewhere in this data the distance of the film plane to the nodal point of this lens must be hidden. How to retrieve it?

 

With above definition it's easy:

 

It's 28.5+36.7-30.3=34.9 mm in front of the film plane (or 46.9-18.4+36.7-30.3=34.9 mm, just the same).

 

Doug, you'll have to look up that diagram from school again, your explanation is awfully wrong.

 

Andreas

I always thought that lenses had two nodal points. I'd guess that the pivot should be at the front nodal point.

Actually, it's the entrance pupil that you seek:

 

http://doug.kerr.home.att.net/pumpkin/Pivot_Point.pdf

Andreas Weber wrote:

 

"Doug, you'll have to look up that diagram from school again, your explanation is awfully wrong."

 

I stand corrected.

 

I Googled around after your post, and you're right and I'm wrong. Thanks for pushing me to look up the correct version.

 

Best,

 

Doug Grosjean

Mark is right. The desired location is the intersection of the entrance pupil and the optical access. When the lens pivots arround the nodal point the image does not move in real space. This is used to find the nodal point on a lens. However, given the lens is connected to the camera one cares where the image moves to in image space. It turns out the right place to pivot to have no parallax is the intersection of the nodal point and the optical access.

 

> Somewhere in this data the distance of the film plane to the nodal point of this lens must be hidden. How to retrieve it?

 

It is 28.5+36.7-24.5 = 40.7 mm in front of the film plane. However, please check by ground glass. Also this value is for a lens focus at infinity. The point move forward by the amount f^2/o where f=46.9mm and o is the distance focused on in milimeters. This is a small correction.

 

One can calculate the location of the enterance pupil for other lenses if one has the lens formula. Refectractive index, radius of curviture, thickness and spacings for all elements. Indeed, get me this data and I will calculate it for you.

I am glad several people knew how to sort this data well and easily for my use. THANKS!

Actually, Andreas is incorrect too. Here's a fuller discussion with all the principal planes, entrance and exit pupil discussed in relation to real lenses:

 

http://www.janrik.net/PanoPostings/NoParallaxPoint/TheoryOfTheNoParallaxPoint.pdf

Thanks to everyone for a genuinely useful discussion.

> Actually, Andreas is incorrect too.

 

You're right, too ;-) I had a special case in my mind, a lens with it's aperture in the principal plane. In general, Miles calculation is the right one, the wanted location is indeed the center of the entrance pupil, as you said.

 

Andreas

Take the back off, replace the bottom, attach some matte acetate or ground glass substitute, put a cloth over your head, swing the camera with appropiate targets. You can see on the gg when everything is right.

 

A macro focusing guide works fine to vary the offset from tripod center.

 

One I did last week if I can find it.<div></div>

Miles said:

 

"The point move forward by the amount f^2/o where f=46.9mm and o is the distance focused on in milimeters.", referring to focussing closer than infinity.

 

I am maybe baffled by the formula (f^2)/subject dist in mm:

Is it correct to thus take an extra 50^2/5000 mm distance, or .5mm extra shift of the pivot point only for a 50mm lens focused at 5m = 5000 mm?

 

Just checking, Miles, please confirm or correct. Thanks!

Generally good answers, but speaking as a QTVR professional with ten years experience, they are far too theoretical and technical.<p>

 

In practice the middle of the lens will do for the nodal point. You don't have to be super-precise because you can easily fix things in post when authoring the panoramas.<p>

 

FWIW the current panorama I am working on is a linear-pano of a street, with 50x shots taken at 5m intervals. The end result will depict @ 300m of a city street with the buildings depicted front-on (similar to the San Francisco "Seamlesscity" project by Michael Koller a few years back).<p>

 

Now because the camera was moved 5m between shots, there are a-m-a-z-i-n-g parallax errors! Yet you can fix them adequately in Photoshop to get seamless joins.<p>

 

So fussing about +/- 0.001mm accuracy for placing the lens nodal point is... unnecessary :?)<p>

 

BTW, using a rangefinder camera for panorama/ QTVR shooting is much more convenient than using an SLR, especially for low-light work. For samples of my M6TTL + 16mm Leica R fisheye work, see mywebsite at:<p>

 

<a href="http://www.4020.net/vr/">vr.4020.net</a>

"I am maybe baffled by the formula (f^2)/subject dist in mm: Is it correct to thus take an extra 50^2/5000 mm distance, or .5mm extra shift of the pivot point only for a 50mm lens focused at 5m = 5000 mm?" Yes. However, spend more time reflecting on Andrew's comments than my formula.

As Andrew said, getting the exact spot does not really matter. However, the entrance pupil is easy to find. Look into the front of the lens. The entrance pupil is where the iris appears to be, not where it actually is. Put your finger on the lens barrel and move it until it looks like it is in the same plane as the iris. That is the entrance pupil - the centre of perspective of the lens and the point about which the lens should be rotated for this type of stitched panorama.

 

Best, Helen