[ptx] visual studio.net compilation done

[Pablo d'Angelo]

On Fri, 28 Nov 2003, Alexandre Jenny wrote:

> > So instead of cleaning the whole pano12 source, we could just 
> > clean the transform stacks (or better, rewrite them, and 
> > provide compact, one function, "stacks" for the most 
> > important transforms ;) then hugin and nona wouldn't need 
> > pano12 dll anymore (and would probably be faster as well...)
> 
> That's a good idea. If you can provide a fast and small list of what is used
> from
> Panotools in hugin, I can cope with that quickly.

Its the coordinate transform functionality from math.c
The concept of having small transforms and stacking them is nice, but
probably slow in the current implementation. I can think of two ways:

1. also use a transform stack approach, with functions similar to pano
   tools, and provide hand written functions for the most important
   transformations, like rect -> rect, erect -> rect, cyl -> rect
   cyl -> fisheye, erect -> fisheye and so on.

2. invest some more time to implement a template based approach to
   the transformation stack and hope the the compiler will optimize well
   across all these templates.

The input for transform creatio probably doesn't need to derived from
the panorama class, a simple struct that contains all needed variables is
probably better than the std::map stuff I'm using elsewhere.

These functions are probably also needed for your image registration
preprocessor.

Yesterday I have already ported resample.c to vigra (without the gamma
adjustment and probably with a few bugs..), so all I need from pano12
currently is the transforms.

After that the next step for nona would be to add some PTStitcher
functionality :)

ciao
  Pablo
--
http://wurm.wohnheim.uni-ulm.de/~redman/
Please use PGP