Shooting panos from a high vantage point

E.J.
Please try it with your camera exactly as I described above and you will see that the frames match up. Your illustration shows what happens when you are not using the pano base of the camera ballhead to pan, e.g. as when the pano is done at the top of the head.
Edited as noted in red. Meant ballhead.
Joe

OK, just did it, camera mounted on rail at nodal point, in other words no background to foreground shift as you pan, camera pointed down 20 degrees, panning with only the pano base above the ball (which of course is what you always do in a pano situation) not only do the frames move, horizontal lines also tilt, just as I have been saying.  Somehow, we are not doing the same thing or interpreting what the OP is trying to do differently.  Now if I compensate for that by releveling the camera between every shot and aligning a horizontal line in the scene to a grid line in the camera, then of course it works just fine, but that requires a lot of intervention between every frame and could cause problems in aligning clouds that are moving due to the amount of time it takes to do that.  If you are doing it that way, then you could just as easily hand hold the entire panorama, which is what some people do and I have done when in a hurry.  Any slight error is easily handled by decent stitching software.

Joe, please mount a lens on the camera in the example you gave and try to pan on a horizontal line and watch what happens!

E.J. Peiker wrote:OK, just did it, camera mounted on rail at nodal point, in other words no background to foreground shift as you pan, camera pointed down 20 degrees, panning with only the pano base above the ball (which of course is what you always do in a pano situation) [...]

E.J., we are saying you pan with the rotating base BELOW the ball, not above the ball. I have made literally hundreds of pano's without a full 2-axis pano rig, and I almost always de-center my horizon line by tilting the camera up or down. (Or both.) If you use a leveling base as I do, or otherwise level the ballhead, and then rotate with the panning base below the ball as Joe & I are describing, you will not have any problems maintaining frame-to-frame alignment.

I don't even have a panning base as part of my BH-55 above the ball, because it's not particularly useful. I have a standard non-panning clamp.

Royce, that causes even more angular shift then above the ball.  I just did that.  We are clearly talking about something completely different because rotating it below the ball causes extreme angles on anything close to the camera.  It results in a completely unusable sequence of images. If I have time tomorrow, I'll photograph a sequence using the below ball rotation, above ball rotation, and multi-row setup.

Martin 095 wrote:I sometimes shoot panoramas, and I have a RRS extension bar that allows me to move my camera so the entrance pupil of the lens is essentially over the axis of rotation of my tripod head (http://www.reallyrightstuff.com/Pano-El ... Screw-Knob). 

I am not sure how to handle a pano from a vantage point of elevation, where I am looking down on the scene such as when atop a mountain overlooking a valley. 

Martin
One thing we did not directly address in your post, although referred to in our posts, is the importance of learning the nodal point of your lens. It is not the same as the entrance pupil of the lens, as referred to by you. Please refer to the video link posted by Mark a few replies above this one. Sorry for not picking up on this earlier.

E.J.
Martin mentioned shooting from a mountaintop overlooking a valley.
You said:
We are clearly talking about something completely different because rotating it below the ball causes extreme angle on anything close to the camera.

I don't know of anyone who shoots that kind of close-up panoramas, but I'm sure that would have its' own complexities, e.g. keystoning. At a distance, things are different. This image is the of angle I had the head at when I did the images posted earlier. If I had used the pano head atop the ballhead, then the level would have shown things to be very changeable from shot to shot and arcing, as you suggested. Instead, the pano at the bottom of the ballhead was used. This might be a little better of the bubbles "bottoming out" at the top edges of the pano head (not used for pano) atop the ballhead and the nodal slide at that angle. This is to demonstrate how much of an angle I had the head at:
I'm sorry I can't show you more now because I've spent most of the day at the hospital with my sister and will probably be busy with that tomorrow. I hope you can test this out in the real world. I wish I could so I could show you, but that has to wait.
Joe

Hey E.J.
I just thought of a simple test you can perform that demonstrates what the difference is between what you do and what we do. Mount a lens like a 200mm on your tripod with the angle pointed down and nodal point over the center of the ballhead. Rotate the base pano of the ballhead and measure the distance from floor to front edge of lens around the 360 degree rotation. Now lock the base pano and rotate only the pano head attached to the top of the head while starting out with the same downward angle. You will see that, as you now rotate the pano, the measurement will change from floor to front of lens around the 360 degree circle. Using our method, the measurement stays the same with no arcing. Using your method, you will be pointing up at the sky at 180 degrees rotation.
Joe

OK, now we are getting somewhere   In your case you have no foreground, basically everything is at infinity.  If you are doing that, you may not even need to worry about the nodal point.  Since the OP was making a nodal point adjustment to the position of the camera, the assumption for me was that there is foreground that he wanted to get properly registered which simply can not be done with the camera pointing downward using just a ball head without the foreground ending up curved.  Certainly if you only have the other side of the valley and the valley floor then of course your method will work, any method that keeps the horizon level, including hand holding and aligning the horizon on a viewfinder or LCD grid line, will work.

Question for you E.J. What do you feel the difference is between using the ballhead and a dedicated pano rig like the RRS? Both rotate around the base at the tripod or leveling base level. Why would the dedicated rig not distort close-up objects in exactly the same manner? In reality, they both function in the same manner. I really hope you can shoot some images today and include the foregrounds. I'm curious about how much noticeable curvature and distortion you will result with.

IRONIC DISCLOSURE: FYI, I feel like the confirmed bachelor who does marriage counseling. I have never shot a stitched pano other than with the automatic programs and apps in an iPod and S7 phone. That's why I'm curious! LOL!

Joe

It has the problem too but a little less due to where the camera is mounted relative to the pivot point.  That's why I said that maybe a slider would be better but that too has problems with registration between the foreground and background.  I would have to try it but I think the real way to do what I am referring to is to use a shift lens and then shift the camera side to side while keeping the lens in a fixed position - that way you keep foreground and background level and you do not change the relationship between foreground and background.  Of course that limits the width of the panorama to what you can achieve within the confines of the shift capability of the lens which is directly related to the size of the image circle.

I think in the end, the confusion in this thread is largely around making a different set of assumptions about what the OP was trying to do.  If we had clarified that up front, we'd all be on the same page I think

Thanks for all the helpful discussion – I am sorry I wasn't here to fill in the gaps as this was going down.  The issue I was observing was similar to the diagram EJ posted earlier where there was decent horizontal overlap but (sometimes very) poor vertical alignment.  This is the reason I  thought only a gimbal system would solve the problem for me.  But perhaps I was making it too difficult … and maybe I need to reacquaint myself with understanding parallax issues.  I will check out all the links provided – thanks again.  But it might be worth just grabbing the camera bag and giving it a go, eh?

SantaFeJoe wrote:Another thought. As wildlife photographers, we seem to have tunnel vision about using ballheads. A simple pan and tilt head leveled as above is very likely even easier to use, since it can be leveled side to side when vertical. It will maintain that side to side level when tilted, so then it is mostly a matter of adjusting nodal point (in this case, it must be the central part of the lens at nodal point, not the top or bottom of the lens body at nodal point) to be above the axis of rotation to match the tilt. The leveling base is still critical to having the vertical axis of rotation of the head plumb vertically.

Joe

So now that we have established (I think, hopefully, maybe?) that using the pano base at the bottom of the head will allow you to match up the horizontal and vertical edges of the frame nicely, I want to address the other point I was trying to express (in red above). I think that the images will be self explanatory, since verbalizing it is not as easy. The chain is at what I like to call the nodal plane, while the arrow represents the nodal point, since a camera utilizes the central part of the lens. The first image shows how the nodal plane/point align with the center of the ball (the vertical pivot we talked about) when lens is horizontal. The second image shows how the alignment of the nodal point at the center of the lens no longer aligns with the pivot point when tilted downward (or upward if you shoot uphill). The third image shows how much you have to move the lens backwards (look at foot) to keep the nodal point in line with the center of the ball (pivot is center of the ball vertically from top to bottom). How critical this is, I don't know, but the geometry says it probably does make a difference. The lens used is a Nikon 200-400f4 EDIF manual focus because the height of the foot is pretty extreme. A camera turned sideways with an L bracket would also put the camera at center of lens up pretty high above the ball, as well. 

Joe

Joe,

So I'd think your setup works fine for as a single-row pano setup. Though it would be exceedingly hard and certainly not quick, to use this setup for multi-row panos. So were back with using a multi-row pano setup, which is essentially a gimbal, when multi-row panos are needed. Also, the multi-row setup is quicker to change the vertical angle on for single-row images since you don't have to deal with sliding the lens to compensate for the nodal point for each change in angle as you do with your setup.

Anthony

The OP only asked for a system to align the borders of the frames vertically and horizontally without the arc when pointing downhill. No mention was made of multi-row panos. A pan and tilt head, as mentioned above, would work easily to do multi-row images, though. If you try the dedicated  multi-row pano setup, you will find the nodal point will shift exactly as the ball or pan/tilt head because you are physically moving the nodal point backwards or forwards from the vertical axis each time you change rows by swinging the arm. If you hold your closed fist up with your index finger pointed up and bend the middle knuckle, you will see what pivoting the pan/tilt head is like. If you do the same pointing down and move your whole finger back and forth, you will see how the gimbal shifts. The only way I can see where this wouldn't be true is if the horizontal axis is exactly lined up with the nodal point and the vertical axis is, as well. Then the swing arm is basically not being used at all. It probably only matters at a steep angle and with vertical elements in the scene. If I have time tomorrow, I will post images using the Kaidan multi-row head with the chain to show you how it works the same way as above. The only thing is, the angle must be severe enough to notice the change in the final images. A multi-row generally starts out horizontally level or close, so no major change in angles from horizontal row to row. I doubt anyone would notice the change. A vertical single or multi-row stitch would be more likely to show the change because of the overall distance moved from level. My preferred head is a Manfrotto fluid video head. It is stable and easy to adjust while performing like a gimbal IMO and no ballhead flop. A ballhead is miserable for multi-rows, no doubt.

Joe

So I guess it comes down to what it usually comes down to in photography. Money versus ease of use.

Anthony

Here are the images with the Kaidan dedicated multi-row head. The first shows the nodal point lined up with the plumb vertical axis back/front and side to side, lens horizontal and head leveled, but with lens low on swing arm. The second shows how much the nodal point (arrow) moves when tilted. The third shows the lens centered on horizontal and vertical axis with lens level (lens mounted high). The fourth shows no real change because both axis are centered with nodal point. The last shows one of the main advantages of a dedicated pano head, IMO, which is the ability to center the lens horizontally with the vertical axis side to side by utilizing the lower slide bracket. The other highly valuable advantage is the indexing for different angles of view. In reality, I don't see much use for the swing arm. A QR mounted on the horizontal axis is all you need on the pano head upright bar.

I don't think any of this matters much regarding nodal point and which head you use, as long as you line up the nodal point to match the angle starting with the first image, or middle row if shooting more than two rows. But, you can use whatever method works best for you, obviously.

Joe

The lens should be aligned with both the vertical pivot point as well as the horizontal one. In your image, that means you need to raise the lens. Once you do that, the center of the lens will not move as you pivot left, right, up or down.

Anthony Medici wrote:The lens should be aligned with both the vertical pivot point as well as the horizontal one. In your image, that means you need to raise the lens. Once you do that, the center of the lens will not move as you pivot left, right, up or down.

Anthony
That's exactly what the third and fourth images show and why I stated that the swing arm is really of no use. A QR mounted at the horizontal pivot is all you need. The swing arm changes the relationship of the horizontal pivot and the nodal point if the lens is anywhere besides centered on the horizontal pivot. If you re-read my post you will see that I stated all this.

Joe