[ptx] Separate Optimizer/Stitcher Package?

alexandre jenny alexandre.jenny at le-geo.com
Tue Dec 16 18:27:32 GMT 2003


Hi,

My vision for the future of hugin, ptoptimizer and stitcher is deeply
bounded with the structure of the code.
I hope to get a small core of essential tools for making panorama in a
package (which actually is called 'panorama').
In this package should reside many of the basis code necessary to create
panorama but not all of them.
Derived from this package will come the many other feature we can imagine :
 - nona (the stitcher)
 - optimizer
 - hugin
 - auto-hugin (something using sift for automatic control point creation)
 - etc (I have a bunch of other ideas to develop ;-) )
Anyway. I trully believe that the underground layer to allow all of that is
to have a good core package for panorama.
And after having this revamp of the code, we could think about that.
Having another team working on the stitcher during our work for example on
hugin is definitely a good idea. But we have to rely on the same core
package which should be well designed and tested (It's a nightmare trying to
get a good UML design out of them, but anyway, it has to be done).

Bye,
  Alexandre

> -----Message d'origine-----
> De : ptx-bounces at email-lists.org 
> [mailto:ptx-bounces at email-lists.org] De la part de Pablo d'Angelo
> Envoyé : mardi 16 décembre 2003 18:25
> À : JD Smith
> Objet : Re: [ptx] Separate Optimizer/Stitcher Package?
> 
> 
> On Tue, 16 Dec 2003, JD Smith wrote:
> 
> > 
> > The creation of an optimizer and stitcher completely separate from 
> > PTOptimizer/PTStitcher is a real achievement, which I think 
> would have 
> > great impact outside of the hugin project.  Since these 
> have so many 
> > fewer dependencies than hugin, does it make sense to split them off 
> > into their own package?  Assuming they are both drop-in 
> replacements 
> > (except faster and with source available), I know many users of 
> > PanoTools would benefit from their wide availability, e.g. 
> those with 
> > a large collection of control script files already in hand. 
>  It would 
> > also be good
> 
> Well, the script parser is only able to parse hugin, 
> PTAssembler and PTGui project files, at least right now.
> 
> > advertisement for the capabilities hugin, though I suspect 
> the other 
> > front-ends would catch on to the speed-up techniques and implement 
> > them themselves.
> 
> Yes, I see the possibilities. Currently the stitcher is 
> integrated into the Panorama Data model of hugin, with 
> support for interactive changes and some other stuff that was 
> modelled after the panorama tools.
> 
> Otherwise it is definately possible to break them apart. The 
> question is whether this should be done now, or after our 
> stitcher has gotten a new architecture (Alexandre had some 
> good ideas).
> 
> ciao
>   Pablo
> --
> http://wurm.wohnheim.uni-ulm.de/~redman/
> Please use PGP
> 



More information about the ptX mailing list