Nodal Drift. How accurate do you need to be aligned with the nodal point? How much does it really matter?

I did not actually make my bracket quite as shown here, but the fact remains that a multi-layer set of horizontal or vertical frames shot to make a panorama is going to throw the alignment seriously off from the nodal setting made when the bracket and the camera is horizontal and you tilt the camera up or down. I do not believe any commercial designs for brackets, even the most expensive ones, take care of this.

Tilting up or down throws the alignment off over the tripod so that, when you rotate the camera on the tripod, the nodal point of the lens is not aligned where it should be over the axis of rotation. Thus, some relative movement between objects in the image will occur as you rotate the camera, making the job for the stitching software more difficult when you combine shots to make a panorama.

From that I deduce that a significant deviation from perfection is quite within panorama software's capabilities to do a good job of the stitching. Furthermore, with hand held shots the same often appears to be true even when some of the control point distances are larger still because most of my panorama efforts so far have been hand held and worked out fine.

I use PTGui to do the stitching. When you start a PTGui project, the first step after telling PTGui which frames to use is to have it align the images. That is where it generates the control points although I believe the optimizer later tweaks them to improve upon that where it can. When you align, you are asked whether the frames were taken handheld or on a tripod because it wants to know whether to run an extra step evidently needed depending upon what I will call nodal drift or more appropriately, the degree of it. Who cares if it takes longer for the alignment process to run, I always use it even with tripod shots because I am then less reliant on having my camera accurately positioned on my bracket and whether it is horizontal.

From the fact that at the alignment step you are asked whether the frames were taken handheld or on a tripod, it is obvious that up to a point it can cope with errors and that probably explains why my set discussed in some detail came back in the PTGUi optimizer with a very good match even though I had tilted the camera up and down to take the two rows of shots.

My conclusion therefore is that you do not need to be aligned with the nodal point too accurately and that within a certain tolerance, some error is quite acceptable. Furthermore, regarding the set of mine to which I have just referred as one example, while I tilted the camera up and down and that does increase the errors, evidently it is not likely to be enough to matter. The figures that came back from the optimizer were pretty good which bear that out.

TIP - Aligning a camera and lens on a nodal bracket

When one aligns a camera and lens on a nodal bracket it is very much a hit and miss affair to some degree depending upon your eyesight. If you are using a camera where you rely upon the LCD screen even with a magnifier held over it, matters are even worse and that is exactlly what I had to do with my Olympus Pen E-PL3. Nonetheless I hit the jackpot. What I figured out is this and I am sure it is spot on if you use PTGui or something similar.

As soon as you think you have found the correct alignment, take a set of shots to make a test panorama in good light and of clearly defined objects like nearby houses. You only need to run the software up to and including the optimizer. If the control point numbers are very small, you are there. I was with that set of mine. The maximum distance was 4 the first time I thought I had the alignment right and checking it out in this way gave me the feedback to say that with confidence.