Panoramics about nodal point

[swmcl]

Hi,

I just went down to my local supplier of photo gear and was interested in the Manfrotto (Bogen) panographic tripod attachment 300N. What I got was a story about how to shoot panographic images that I want to question on the forum.

The sales guy was trying to tell me that I needed to pivot the camera about the nodal point of the lens. I am guessing this is the point at which all light rays converge. By this he meant, "Buy this new fangled 303 Virtual Reality and Pan head ..." (I then told him I wanted to use a 4x5 !!!

He reckoned that the images do not suffer distortion as they would when I just whack a 4x5 on top of a rotating head.

My plan was to use a detented head like the 300N and prior to shooting, just level the tripod using its bubble level. I thought it would be a simple matter of turning on the horizontal.

I don't think the Shen Hao tripod screw is in the vertical centre line of the lens ...

Any thoughts?

[Don Miller]

What are you trying to do?

Generally,with LF panoramics are made with a single sheet of film. 6x17, 4x10, or a crop. The first multiple film method I would try is making three exposures using shift.

The salesman is right about the nodal point. It's not always critical, depending upon the relationship between near and far objects.

[Jim Galli]

What are you trying to do?

Triptych's, quadtychs. I have been questioning this too. I do know the Cirkut camera does not rotate around the node. So what is the mathematics? Interesting question Steve. I've done lots of triptychs mindlessly rotating the camera around the top of the tripod with little thought to where the sweet spot might be.

[Leonard Evens]

I am in the process of trying to master just what you have in mind.

You need to rotate the head about the entrance pupil. For most lenses you encounter this will be the same as the nodal point, but it may not be. (There was a long discussion of this issue in this forum a while back, and if you want to understand the technical niceties, search for it. ) For most normal and wide angle lenses the entrance pupil is pretty close to the front of the lensboard. For lenses of telphoto design, it is going to be somewhere in front of the lens. Fortunately, the usual interfaces to Panorama Tools such as PTGUI or hugin have the capability to correct for small errors in the position of the lens. People can even use them with handheld cameras and no panormaic head.

A good place to start to learn about this all is http://wiki.panotools.org/

In practice, here is what you do. Mount the camera on your panormaic head on the tripod and level panormaic head and the camera. Adjust the position of the front of the lensboard (or where you think the entrance pupil may be) over the pivot point. Then choose some near and far points which are in line on the gg and rotate the camera. See if they remain in line. If they don't shift the position of the lens in one direction or the other until you eliminate this parallex error. You can also check by looking at the relative positions of two points, near and far, which don't line up, to see if the distance between them on the gg shifts as you rotate. If everything is sufficinetly far awaysso that there really aren't any near and far points that will appear in the final image, it is not going to be crucial in any case. There won't be any significant parallex error.

In picking the panoramic head to use, you want to be sure it has appropriate adjustments for positioning your large format camera. Most of them are designed for use with 35 mm SLRs or DSLRs and may not work well with a large format view camera. I use the Jasper Engineering Pano Head 2 (www.stereoscopy.com/jasper/panorama.html) which works well with my Toho FC-45X, at least for horizontal panormas. It might require some fiddling to use it for vertical panoramas.

I use hugin since I work under Linux. There are several tutorials available to help with using hugin or PTGUI. There is also a yahoo newsgroup where people will answer questions. In principle, using these interfaces with large format should be easier than with smaller formats, but in practice I found it harder. If you get stuck, I may be able to help you, but I am still learning myself. I include an example below. The original is 26,300 x 6292 pixels, but of course I've reduced it significantly for display here. I made it from two images, one rotated to the left about 22 degrees and the other to the right by about 24 degrees. The lens was a 75 mm lens. The main reason I went to panoramic photography is that I want to photograph facades like this which would be impossible to do in a single shot with an available wide angle large format lens. I would have been better off had I used three images and restricted myself to the detent positions on my head, each of which represents a rotation of 15 degrees.

One problem with the software is that it it is in some ways too powerful. It will also correct pincushion and barrel distortion, which is endemic for SLRs and DSLRs, but which is negligible for large format lenses. You would think that would make it easier, but I found just the opposite.

[Bob Salomon]

Take a look at the Novoflex Panorama Q Pro http://www.novoflex.com/Einzeldatenbl%E4tter/single_data_sheets/Dabla_Pano_Neu_E_web.pdf

[Doug Dolde]

If you have a 4x5 with enough back shift you can just shift the back right and left then stitch the two resulting images. There is no parallax this way at all.

[Daniel_Buck]

you don't NEED a panorama head setup to take panoramas. for a full 360/360 panorama, yes you would probably want a pano head, but for simple one or two row panoramas (like a 6x2 grid, or something) you don't need a panorama head. any differences in paralax aren't going to be that big, unless you are shooting something that is very close to the camera.

[Helen Bach]

Doug wrote: "If you have a 4x5 with enough back shift you can just shift the back right and left then stitch the two resulting images. There is no parallax this way at all."

...Or if you use a slide bar (Jasper Engineering makes an excellent heavy duty version) you can use front shift as well as back shift.

Swing lens cameras usually rotate the lens about the rear nodal point, not the entrance pupil, by the way. It's a different situation.

Best,
Helen

[Leonard Evens]

A couple of comments.

First, I mispoke about checking the separation on the gg of a near and far point. What should remain the same in the absence of parallax shift is the angle between the rays from the entrance pupil to the the points. But if this angle is small, and if you don't rotate too far, the horizontal distance on the gg should stay pretty much the same. In the example I posted, note the right hand side of the marquee which is almost, but not quite perpenducular to the plane of the facade. Also, look at the strut just above it. I examined both when setting up my camera to be sure I didn't have any significant parallax shift. It is best, of course, if you have a near and far point which should be in line with the entrance pupil, or at worst would be in line if they were at the same height.

Second, it is certainly true that you can avoid the whole issue just by shifting the back in both directions. But that will only allow a relatively minor extension of the angle of view. Thus, I would never have been able to encompass the whole facade in the example I posted that way.

[Walter Calahan]

Panos made simple. Go to 'Really Right Stuff' web site. They spell out exactly what you have to do, and sell excellent equipment to do it.

http://www.reallyrightstuff.com/pano/index.html

[Greg Lockrey]

This may seem to be a dumb question but here it goes, how do you know for sure where the nodal point really is? Some lenses have a given length, but the actual nodal point is either in front or behind the midpoint of the lens where you would think it is.

[Greg Lockrey]

Nevermind, I found the answer.

Nodal point of lens

--------------------------------------------------------------------------------

For a "panoramic" camera you need to determine the length of the conical tube needed. Do this by experiment: Arrange the lens to be looking at a distant object and move a ground glass until an imaage focuses clearly. Measure the distance between the frosty side of the glass and the place on the lens where it will bear against the tube you are designing. Some make such a tube using a threaded insert which can be used to "tweak" the length and thus avoid critical measurement.

The "nodal point" is only useful in the case of a "panning" (not panorama) camera. In a panning camera the film is arranged inside a cylinder and the lens is placed so that it is at a distance to focus on the film. The nodal point is the axle around which the lens is turned so as to pan around and make a travelling image painted onto the film, usually thru a vertical slot.

This point (nodal point) can also be determined by simple experiment: Arrange the lens so that it focuses at infinity on a ground glass. The point at which the lens can be rotated (swuing about a vertical axis parralel to the film) and the image on the ground glass will not displace either left or right is the nodal point. Finding this point makes the trick of the panning slot work properly. Its usually right near the iris.

SKG

[swmcl]

Hi guys,

I guess what I'm trying to do is not waste money again and again. I don't want to buy something to find it not work.

Jim, I too am mindless most of the time! I reckon a little more mind might make me more pleased when I get the results back from the lab!

Leonard, you are a champ. I'll sit down now and try to digest your good words. Cheers.

Bob these are similar to the Manfrotto product. I would hope to buy only the revolving product but this local sales guy wanted to push a much more elaborate solution.

Doug, my wife will need to approve the next cap ex to get a body that allows back shift! My TFC45-II from Shen Hao is rather fixed on the back except for rise/fall.

Daniel you're the winner! Here in Australia, my subject are usually (unfortunately) a LONG way away. Even if I get closer I may not be worried by it. I don't plan on taking too many 6x12s in succession!

Helen, I've bookmarked http://www.stereoscopy.com/jasper/index.html. Thanks!! They also do a "panorama tripod attachment" but it seems to be limited to MF or smaller.

Walter, I see the gear and I like it in my mind but not in my wallet! I appreciate it, I really do.

Thanks Greg I'll learn more in the next few days I'm sure.

Cheers, all.

[swmcl]

Hey Leonard,

Is there a software product on 'Linux' that can handle an image with 16bits per pixel including an infrared channel ??

(When you say Linux I hear Debian. Much of the software is not available on Debian unfortunately.)

I'd love to think one day I'd be free of MS but that day will be some years from now!

I wonder if Photoshop can do it all with less fuss?

Rgds,

Steve

[Helen Bach]

Steve,

I should have mentioned that you can ask Jasper Engineering (Tim Chervenak) for 3/8" tapped holes in the bottom of the bar instead of the standard 1/4" ones.


Nevermind, I found the answer.

Nodal point of lens

--------------------------------------------------------------------------------
...

The "nodal point" is only useful in the case of a "panning" (not panorama) camera. In a panning camera the film is arranged inside a cylinder and the lens is placed so that it is at a distance to focus on the film. The nodal point is the axle around which the lens is turned so as to pan around and make a travelling image painted onto the film, usually thru a vertical slot.

This point (nodal point) can also be determined by simple experiment: Arrange the lens so that it focuses at infinity on a ground glass. The point at which the lens can be rotated (swuing about a vertical axis parralel to the film) and the image on the ground glass will not displace either left or right is the nodal point. Finding this point makes the trick of the panning slot work properly. Its usually right near the iris.


Just to avoid any confusion it might be worth pointing out that that article refers to swing-lens cameras, not stitched panoramas. The correct point of rotation for a stitched panorama is the entrance pupil (see posts above) not the front or rear nodal points. The correct point of rotation for a swing-lens camera is the rear nodal point.

Best,
Helen

[Leonard Evens]

Hugin is available as a tarball from sourceforge. It should compile under debian, but of course you may find you need also to get other packages. . I just did a google search on "hugin debian" and came up with
packages.debian.org/unstable/graphics/hugin
so apparently it is available as a debian package also.

Hugin does handle 16 bits per channel. I don't believe it can handle an infrared channel, but I'm not really sure.

I found learning to use hugin somewhat frustrating. Asking questions at the yahoo groups website helped, but I ended up having to do a lot of experimentation. The manual at
wiki.panotools.org/Hugin
which I eventually found was very helpful, but it was designed for the latest version of hugin, which I didn't have.

[Greg Miller]

Hi guys,

I guess what I'm trying to do is not waste money again and again. I don't want to buy something to find it not work.



I would suggest trying a panorama with the gear you already have. Parallax is not a serious issue in many cases. I have a book that will be published in 2008 with 200 pages of (mostly) panoramas. It was not necessary to be concerned with parallax (in a hardware sense) in the field for any of the images that I expect to be in the book.

It will depend, however, on the focal length of the lens and the subject matter. Parallax will be a concern for architectural images but rarely for nature landscapes. For normal to telephoto focal lengths parallax will rarely be an issue. You can deal with parallax with shorter focal lengths largely by overlapping the panning more (as much as 50% overlap on very short lenses). For the situations where I do want to deal with parallax in the field, I use an Wimberley long platet that also has an Arca-Swiss type clamp - this allows me to move the camera forward and backwards until the nodal point of the lens is over the pivot point. It is easy to find the nodal point by observing the relationship of near/far points as I pan the camera.

If you have Photoshop CS3, you should try Photomerge. It is vastly improved over previous versions of Photoshop. I used to manually stitch my panoramas but now I frequently use Photomerge for all but the most fussy images.

So experiment with panoramas before you invest in more hardware. You may find that you do not need to buy anything at all.

[Emmanuel BIGLER]

Since the question of entrance pupils for parallax-free stitching with a conventional camera re-surfaces again, I hope that the honourable raders of this forum will forgive me to re-post two diagrams explaining the question, as we had discussed before (Leonard and Helen see what I mean ;-)
previous discussions :
http://www.largeformatphotography.info/forum/showthread.php?t=13167
http://www.largeformatphotography.info/forum/showthread.php?t=17775 including some practical tests with a 360mm Tele-Arton

Diagrams :
principles http://farm2.static.flickr.com/1156/827305231_37bc942399_o_d.png

diagram for an (hypothetical) extreme telephoto with a pupillar magnification of 0.5 http://farm2.static.flickr.com/1122/827305245_19fe318edf_o_d.png

Modern telephoto are less extreme than this case ; I think I remember reading here that one of the old German telephotos like the Voigtländer Telomar had a pupillar magnification factor close to 0.5. I have found experimentally something like 0.57 for the 360mm tele-arton.
A friend of mine has found that the entrance pupil of his 200mm telephoto lens (for 35mm photography, off-topic) is located close to the film plane... A rare case where the tripod socket under the camera body suffices ;)