Panoramic photography with 6x6 camera, distortion

I have recently done panoramic photography with a 6x6 camera and

a 80 mm normal lens. I use a ballhead with a bubble level to get the

camera right and take two to three pictures that I after printing

put next to each other. The problem is, even with a normal lens I

get a slight distortion at the edges which means that for instance

a phone-cable going from one frame to another doesn't "meet in the

right place". This is in fact much more disturbing to the impression

than the border between the prints is. Does anyone have any experience

with this kind of

panoramic photography and can offer advice?

Right off hand, I would suggest shooting twice as many photos, and

then splicing from mid-center.

 

<p>

 

Any time you notice distortion in the viewfinder when moving the

camera, you will get distortion trying to line up consecutive

frames. I can see distortion when I am moving my P6x7 with its 90mm

lens. The 300mm Pentacon gives me no such problems.

Completely symmetrical lenses could help in your situation. They have

no distortion. If you have cross sections of your lenses in your

owners manual, that might help in determining which of your lenses

would be best to use. Distortion is differential image magnification

across the frame. SR

Well, your problem is actually that your lens has *no* appropriate

distortion.

 

<p>

 

The simplest and least expensive solution (compared to true panoramic

cameras like the Noble or the Horizon) is to use a special software

to distort and stitch the images. I'm using PhotoVista from Live

Picture (http://www.livepicture.com/) for this. Works great!

I have done some panoromic shoots using the Rolleiflex TLR F2.8 and

F3.5 with the Rollei panoramic head. These Alignment problems seem

unavoidable but something can be done to reduce it.

It is best to shoot objects have no straight lines across each shoot

i.e. natural objests. It is also hard to get truely leveled camera.

Also watch the print error at enlargements i.e. alignment of negatives

and exposures etc. Use longer focal lenses or shoot 35 mm format in

portrait mode (i.e. align at longer edges) may reduce the alignment

problems also.

The error you get from a ball head rotating around the tripod

socket is greater than the error you get with the rollei

panoramic tripod head, which swivels the camera around the "center" of

the lens. You will, nevertheless, have some lack of linearity:

I think you can expect to get one line (say, the horizon, at infinity)

all at one level and lined up.

There's a <a

href="http://db.photo.net/bboard/q-and-a-fetch-msg.tcl?msg_id=000I1o">

thread on the photo.net forum</a> on this subject at present,

describing a practical method of judging the lens centre and hence

minimising distortion. This may help.

Thank you to all who gave suggestions, on and off line. Also thanks to

Andrew, the forum-moderator, for pointing to a similar thread on

photo.net.

 

<p>

 

I have learned a lot, among other things (unfortunately!) that it

probably doesn't matter what overlap I make, the images will still

not merge perfectly with a normal-lens. The only way to make "perfect"

seams without computer help seems to be a combination of finding the

correct axis to turn the camera around and to crop from the optical

mid-center. This (the optical center) would be the only place to

crop if fore-ground is included, because the distortion is more

apparent in the fore-ground (in my belief and experience).

 

<p>

 

The conclusion (mine) must then be that it is necessary to have

a camera with shift-function if I want to use this method. The shift

camera would have to be turned on it's side for shifting sideways.

I can then use more of the lens image-circle and "place my film-area"

so that the optical center is not in the middle of the picture (that

wouldn't help I'm afraid) but as close to the side as possible. By

reversing the camera and shift it the other way (for the next

picture), I can then have some overlap that can be cut off later and

merge the pictures in the optical center.

Of course, with this method it will only be possible to merge two

pictures, not more. It would still be possible (if the shift of the

camera and the coverage of the lens are large enough) to get 1 to 2

panoramics or perhaps even 1 to 2,5.

 

<p>

 

I suppose this method would work, given that the lens is turned around

it's nodal point, but if my reasoning is off please don't hesitate

to correct me!

Wait a minute, that would just be a worse way then to use a camera

that

can handle the double film lenght (4x5 instead of 6x6 or 8x10 instead

of 4x5) and just cut out the whole part without seams!

 

<p>

 

So, I wasn't as clever as I thought!

Just a commnet for Richard' last post:

The Rollei panoramic head has the same rotating point as the camara

tripod socket which is about 1 inch before film plane. However if I

put an Rolleifix (tripod quick release) on top the paroramic head, the

rotating point will be moved to the lens rear element. I only use

panoramic head with the Rolleifix. The negatives show some

overlapping but the prints from the lab were cropping a little too

much to align them together. Overall I care less for alignment

problems but the non-linear lines across multiple frames are more

distracting to watch.

<a href="http://www.zoerk.com/">Zoerk</a> makes shift adapters for

many cameras. Their products are handled by Ken Hansen

Photographic. The Zoerk web site contains a panoramic photo composed

of three shifted images.

As an update: I asked a Hasselblad representative how they make

their panoramic slide-shows. For each panoramic image they use

three of their projectors (with built in shift) so the image is

made up of three slides. The slide mounts are special order (I

don't know which company) with some "built in vignetting" (if I remember correctly). They suggested not using

lenses wider than what corresponds to the 100 planar. I forgot to

ask about panning around the lens nodal point but I assume that

the photographer do that as well. The slides are mounted

with some kind of pin-registration device.

 

So I guess all it takes for a nice panoramic slide show is six or nine

PCP 80's, a CF100 and some special-order slide mounts. :-)

The distortion you are experiencing is either "barrel" or "pin cushion" distortion. This is typically caused by the camera being rotated on the tripod at the film plane. To limit (or avoid) this distortion you need to pivot the camera at the Optical Center of the lens. The optical center is usually where the f-stop ring is located. You can make an adapter (or have a machinist make one for you) by taking a 1/4 inch piece of flat metal, approximately 2 inches wide by 10-12 inches long, and having a grove cut into the metal so that you can attach a 1/4 20 set screw to the plate. Also have a 1/4 20 threaded hole tapped into the metal at one end. Mount the bracket to the 1/4 20 tap, and then attach the camera to the sliding set screw. Before tightening the camera down, slide the camera backwards/forwards so that the f-stop ring (optical center) is centered over the center of the pivot point on the tripod. This works no matter what lens you use. I have had very successful matchings this way. Also, if you shoot more than just a little bit of overlap you will get the best alignment for printing. I bet you have noticed that the prints you currently piece together also end up mounting in an arch formation rather than a straight line. You should see most of this disappear once you make the adapter and start shooting from the optical center of the lens. If you would like a simple drawing of the device, e-mail me at vlioce@dps.state.nm.us adn I will e-mail to you. Can't attach a bitmap to this file or I would have already.

Victor, I have used a bracket as you describe for a 35 mm camera.

Do you think a macro-focusing rail (like Novoflex) would do the

same job? And you're right about the arch-shape!

Obviously you do not intend to go the digital way which is a lot easier. Maybe you can have your frames scanned, so you use a PC for finishing off the project. Manfrotto (and some other manufacturers)have several types of panoramic heads that allow adjustment of the camera position so as to bring the nodal point (by trial and error) over the pivot of the head. Besides the head allows panning the camera by equal angles for each exposure (a variable angle "click" mechanism). A ball head is unlikey to provide any accuracy in every respect.