Hello,
I just compiled Hugin two days ago, and love that there is an
open-source panorama stitching tool (and that it works on Linux).
I've read through a few tutorials (and watched a couple on YouTube),
but there are still a few things I don't have a good handle on.
Plus I have a handful of basic questions.
I thought I'd make my first post to the list one of the bigger
issues that I'd like to work through -- stitching an MxN array
of images from a robotically controlled camera, which may include
a featureless sky and/or limited/changing features in water.
I use a GigaPan EpicPro device for the image sequence capture, and
I get considerable overlap in the images, so I should have a decent
set of images for my panoramic captures. The GigaPan unit comes
with software (not for Linux) that will automatically do the
stitching. This software does a decent job, but there are clear
seeming issues, and also it doesn't do any context-aware analysis
to determine from which image to take a possibly moving object. All
the stitching parameters are determined by the fact that it knows
it's taking an MxN array of images, so it doesn't need feature-based
control points to do the work.
But because of the fact that this stitching isn't perfect (or really
sufficiently good), I've been looking for other solutions, and I
recently found Hugin. Hugin is also nice because I'll be able to
do the work on my desktop (though I suppose when it's sucking all
the resources from my machine I won't appreciate it so much).
So after taking a couple hours to get Hugin and it's dependencies
compiled, and reading through/watching a tutorial or two, I began
some practice panorams. It happens that my first trial is a
night image of the Washington Monument, and so there are some
featureless sky segments. I managed to get a panoram if I removed
those images, but I would like to widen the image, which requires
the inclusion of pure sky (or ripply water).
So for the areas where there are good features, Hugin worked well
in finding control points. But since the relationship between all
my images is known from the mechanized capture process, I would
like to be able to resort to this knowledge when no feature matches
can be found.
Has anyone successfully created a workflow that adds control points
based on geometry rather than feature matches?
If one were to somehow calculate these control points, how best to
get them into the system, and then not have them altered by the
optimizer stage, etc.?
I have placed some sample images on the web for clarification of
what I'd like to accomplish:
- http://www.avl.iu.edu/~shermanw/HuginTest/
There are 16 images in an 8x2 array shot with 400mm focal length
in that directory.
Oh, and I'm using Hugin version 2010.4.0.854952d82c8f (released
01/01/11).
Finally, if you go to my base URL, you can see some of the panorams
I've made using the GigaPan software (and some of the other issues
I hope Hugin will help with, like correcting my use of auto-exposure):
- http://www.avl.iu.edu/~shermanw
Thank you for your anticipated help,
Bill
--
Bill Sherman
Sr. Technology Advisor
Advanced Visualization Lab
Pervasive Technology Inst
Indiana University
[email protected]
--
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 [email protected]
To unsubscribe from this group, send email to
[email protected]
For more options, visit this group at http://groups.google.com/group/hugin-ptx
