my bridge panoramas have improved quite a bit since I'm using my improved philopod technique to position the no-parallax-point, see:

https://groups.google.com/forum/#!topic/hugin-ptx/I6rhafCZmMY

I've been using this technique for some two years now, and I don't want to do without it any more. While I thought my panoramas done from the top of one of my Leki sticks were pretty good, I had to admit that adopting the new technique made them much better - at almost no extra weight; the device is living in an ex playing-card box, wound up around a plastic card, and I use my GPS device as the ground target, oriented to magnetic north with the built-in magnet flux compass, so I can orient the panorama before I even upload it to 360cities.

While previously I had to retouch every bridge panorama I did, do some finessing with the CPs or just live with the visible parallactic errors, I can now confidently just stitch the thing and have remaining errors which are too small to bother me, just like in this panorama. The banister does, of course, pose a challenge, since it's so close-by, and there is plenty of background, pulling the SIFT algorithm towards a population of CPs from the background. If, then, the foreground (the bridge) is off, parallactically speaking, a match of the foreground elements is quite impossible and the blending is bound to contain artifacts.

I estimate that my improved philopod technique limits my movements of the NPP to a few centimetres (if it's not too windy), which is good enough for most landscapes.