strange result EuclideanReconstruction about gsoc2011sfm HOT 7 OPEN

additional information below about issue 18:
compiler - MinGW,
lang - RU.
I think the problem in BundleAdjustment, right? or in the libs like blas, 
lapack, f2c or SBA.

Hi, this strange result seems to be related to BundleAdjustment, as you 
thought... But unfortunately I don't understand where the problem is :(

I will give a try to your images...

Why you do not use BundleAdjustment from the library LibMV? I`ll try to do this!

Or SSBA
URL: http://www.cs.unc.edu/~cmzach/opensource.html

I first used the LibMV bundle adjustment but, as with SSBA, it's not possible 
to have missing data (ie points which are not in all images can't be used...).

It would be really helpful if you find a way to solve this problem... I didn't 
see you proposed to try with an other bundle... Google will soon (I hope) 
release a new bundle framework, so maybe it will be easier to do that...

"Google will soon (I hope) release a new bundle framework" - you mean this 
project? http://code.google.com/p/bundle-adjusment