Hi Bruno, Wouldn't it make sense for cpfind to implement a few other heuristics? I'm thinking that when multirow fails or ends up with a disconnected image or set of images, that it should fall back to exhaustive cross-checking to find any matches possible to connect the image. That would make it work fast in the common case, but transparently (albeit a bit more slowly) give expected results if there's an irregular capture pattern. Another somewhat more complex heuristic would be to also query a geometric model of initial cross- matches to infer other potential overlaps and cross-check those. I suspect that many users use the defaults and are unlikely to invest the time to drill down and change command line options to cpfind.
On a slightly separate issue, it also appears that there is no easy way to specify which photos of a stack are used by cpfind when doing HDR panos with a separate stack aligner - even reordering them in Hugin doesn't seem to affect this. Depending on the auto-bracketting (or errors during capture), the first picture of each group may not be the best exposure for control point extraction. Brent On Oct 19, 10:25 am, Bruno Postle <br...@postle.net> wrote: > On Wed 19-Oct-2011 at 05:07 -0700, Sebastian wrote: > > > > >When taking pictures I follow a scheme whereby I take a picture every > >30 degrees, which gives 12 pictures for each level. The camera numbers > >the pictures chronologically and they are also entered into HUGIN in > >their chronological sequence. However, if for whatever reason once a > >picture is out of the order such that two pictures next to each other > >are not overlapping HUGIN stops and complains that there is no > >overlap. > > There are various ways of generating control points for a project, > the default method with the current release is now to use the > 'cpfind' tool with the '--multirow' option enabled - This makes > assumptions about the order of photos to speed things up, and it > works ok in most situations. > > Comparing every photo with every other photo is slower, but does > mean that ordering is irrelevant, you can switch Hugin back to using > this method here: File -> Preferences -> Control Point Detectors -> > Hugin's CPFind -> Edit... -> Arguments > > ..and remove '--multirow' from the text box. > > >However, when I look at pictures that are overlapping vertically, > >HUGIN appears to ignore these vertical overlaps in most cases. > > The --multirow option has two steps, it will fail to match rows with > each other if the rows can't be assembled first. > > -- > Bruno -- 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