Some other comments about this whole thing: 1) Panoramas with extreme wide angle lenses do work. The 360 of Provincetown and the shot down Crawford Notch (with the road and railroad running off into the distance) were both taken with my 8-16 mm lens at 8 mm. There is some distortion at that setting, which didn't matter for the Crawford Notch shot but which did for Provincetown (with the horizon visible just about everywhere). In the latter case, I had to use a lot of horizontal control points and other linear control points to correct it, but when I did, things snapped into place. For Crawford Notch, I simply didn't worry about it and had no problem.
2) Even with an extreme wide angle lens, photos should be taken as closely spaced as practicable, preferably not more than 45 degrees apart. Remapping to cylindrical (which I use) or equirectangular causes the images to severely barrel; if there isn't a lot of overlap, there will be a lot of foreground that isn't usable after cropping. At Provincetown, I had no choice but to space them 90 degrees apart due to the construction of the monument I was shooting from (and it's very obvious that there's no other place from where I could take this shot). If the viewing platform were octagonal rather than square, I could have gotten a lot more foreground. 3) Except for the sunsets (on a tripod) and the 360 (on a monopod, but from different locations), everything was hand held. That should have caused problems in the foreground from perspective errors...but in practice didn't. The reason is that these shots emphasize background more than foreground, and I simply tossed control points in the foreground that were correct but which had large errors. Enblend managed to do a good job; there aren't a lot of visible seams even if you do know where to look. 4) Blend stacks or fuse layers: that is the question. Blending from stacks caused some colorimetric problems in the 360; the areas near the edges of the original photos turned out noticeably lighter. However, fusing layers caused ghosting problems where enblend picked different seam lines for each exposure set (quite apart from ghosting caused by subject motion between the three shots in each stack). In the end, I went with blending stacks. I guess one way around this would be to use enblend with one layer, generate the masks from it, and use those masks to blend the other layers. Perhaps Hugin could automate that? It would probably yield the best results of all when blending a panorama consisting of exposure-bracketed stacks. 5) Related to point (2), if you don't take enough extra shots at both ends of what you're interested in, the barreling effect of the remapping will result in either the left and right ends not being usable or having to crop a lot vertically. I found an easy workaround for this problem. After creating the initial panorama, run Hugin again on the panorama (just the one image). Specify the lens as a wide angle cylindrical lens, and then specify the output projection as rectilinear. This has the opposite effect of remapping rectilinear to cylindrical: it pincushions the result. Experimenting with different angles (I used 20 mm for Crawford Notch and 30-35 for most of the others, which were taken with the longer end of the 8-16 lens) eventually gave me something good in each case. Until I came up with that trick, I was faced with a very unpleasant cropping decision on the sunset panorama, but this trick expanded the height at the left and right enough that I got everything I wanted. Taking one more shot (or bracket sequence) at each end, and setting the lens to somewhat wider than I really wanted, also would have solved this problem. This obviously won't work for the 360. 6) Autocrop isn't all that useful. Sometimes the cropping decisions it made were rather strange. But there's another kind of autocrop that would have been very useful to me: crop to the outer envelope of the panorama (rather than something approximating an inner envelope or maximum fully covered area). (Sorry, I don't have time to get involved here; I have a big backlog of stuff as it is with Gutenprint that I need to get to, so perhaps Yuv will forgive me for suggesting this enhancement without providing code to go along with it!) 7) I'm still not entirely happy with what enfuse does, particularly with sunsets and very bright sky near the horizon (see the sunset panorama -- the dark exposures have plenty of detail in the clouds near the horizon that enfuse lost, and there's no way to get it back). The exposure-mu and exposure-sigma parameters (and, for that matter, the others) are not at all intuitive. It's a great tool and a lot easier than tone mapping with qtpfsgui aka luminance, but for a photographer who thinks in terms of dynamic range (or at least contrast) and exposure, it doesn't make any sense to me at all. Interestingly, when I played around with luminance, the very best results were looking at the original .exr with a gamma of 2.2 or 2.6, but I couldn't find a way to save that. It would be very handy indeed if luminance provided a way to choose the gamma and simply save the result with no further tone mapping. But this isn't the pfstools list... 8) For control points with exposure stacks, I found the following strategy to work well: connect the middle exposures to the other middle exposures, and connect the low and high exposures only to the middle exposure of the same stack (and maybe to each other, but only within the same stack). The CP finders that I've tried don't do too good of a job of matching features between images that are exposed very differently (or in general for areas that are badly exposed, and I'm intentionally badly exposing to get selective good exposure in the highlights and shadows), and it's important to have a really good match between the exposures in each stack. So I simply knock out all the control points connecting high and low exposures with any shot in a different stack. -- You received this message because you are subscribed to the Google Groups "Hugin and other free panoramic software" group. A list of frequently asked questions is available at: http://wiki.panotools.org/Hugin_FAQ To post to this group, send email to hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
