Re: [Meshlab-devel] texture atlas - please comment

[Paolo C. <pao...@is...>]

I like your proposal, but I strongly suggest you to proceed gradually 
and start and complete and test a functionality at a time.

First of all start to just add a simple rectangular packer,

- then try to use the rect_packer provided code

- then try to look at the code in the trimesh/textcoord_optimization.h file.

Before facing LSCM we should solve the issue of what numerical solver 
use for it and for other issues that we have around in the vcg.
I hope we could take some decision in the next weeks about the numeric 
solver in the vcg.

I noticed that some notes on the basic types were missing so i added it 
in http://meshlab.sourceforge.net/wiki/index.php/Writing_a_Plugin
Feel also free to contribute to the wiki in the part about the writing 
of the plugins when you see points needing some clarification or a 
deeper discussion.


cheers
p.







skunkwerk wrote:
> thanks for the hints paolo,
>    i wanted to use this algorithm: 
> http://alice.loria.fr/php/article.php?pub=../publications/papers/2002/lscm
> but it seemed like it needed information on the actual models, and was 
> taking in arbitrarily-shaped textures.
>
> the only other algorithm i've found for texture packing is a recursive 
> tree-based one:
> http://www.blackpawn.com/texts/lightmaps/default.html
>
> i tried this line (along with everything else i could think of):
> MeshModel::FaceIterator fit;
> but it didn't like the 'MeshModel' part - is 'mm' not an object of 
> type MeshModel?
>
> if someone could suggest the correct syntax i can get started on 
> testing the texture atlas
>
> thanks,
> imran
>
> On Tue, Apr 29, 2008 at 2:13 AM, Paolo Cignoni 
> <pao...@is... <mailto:pao...@is...>> wrote:
>
>     skunkwerk wrote:
>     > i've attached a texture atlas function.
>     >
>     > it compiles except for a ConstFaceIterator error - I've tried using
>     > the template<typename MESHTYPE> but that didn't work either.  the
>     > algorithm is VERY basic (textures are added one after another in a
>     > row), but should work.  i couldn't find any algorithms for more
>     > closely-packed positioning, but i could look at the code in
>     > openscenegraph's texture atlas generator.
>     Nice Job!
>
>     The right place for this stuff is the filter_texture plugin that i
>     have
>     just added,
>     Now it is just a placeholder filter you should add your code there.
>
>     Important:
>     In the plugin dir you can find a rect_packer that should make good
>     job.
>     please use it to build a good packing of the texture size
>
>     > questions:
>     > 1)  if someone could tell me how to iterate through faces from a
>     > function outside vcg, that'd be great so i could test this out
>     We have started to improve the developer documentation: two new
>     pages in
>     the wiki should help you
>
>     http://meshlab.sourceforge.net/wiki/index.php/Accessing_the_mesh
>     http://meshlab.sourceforge.net/wiki/index.php/Algorithms
>
>     > 2)  paolo, what's the best location of this function?  i'm currently
>     > putting it in mainwindow_Runtime.cpp to test it out
>     The right location is inside the filter_texture plugin
>     possible parameter of the plugin:
>     - format of the saved texture (as a enum)
>     - max size of the texture
>
>     > 3)  how do I access the data from already-loaded textures?  so i can
>     > prevent having to load all the textures again...
>     Do not worry about that. Feel free to use QT to load again the images
>     and process them, it should not create a significant overhead.
>
>     Once your processing is ended (and you have saved the new texture)
>      you
>     should update the texture names stored inside the mesh sending a
>     signal
>     to the  updateTexture() slot of the mainwindow (the parent of the
>     action)
>
>     cheers
>     p.
>
>     --
>     Paolo Cignoni  -- Senior Researcher
>     Visual Computing Laboratory - ISTI - CNR
>     http://vcg.isti.cnr.it/~cignoni <http://vcg.isti.cnr.it/%7Ecignoni>
>
>     ISTI - CNR
>     Via Moruzzi 1,
>     56124 Pisa
>     ITALY
>
>
>     -------------------------------------------------------------------------
>     This SF.net email is sponsored by the 2008 JavaOne(SM) Conference
>     Don't miss this year's exciting event. There's still time to save
>     $100.
>     Use priority code J8TL2D2.
>     http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
>     _______________________________________________
>     Meshlab-devel mailing list
>     Mes...@li...
>     <mailto:Mes...@li...>
>     https://lists.sourceforge.net/lists/listinfo/meshlab-devel
>
>
> ------------------------------------------------------------------------
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by the 2008 JavaOne(SM) Conference 
> Don't miss this year's exciting event. There's still time to save $100. 
> Use priority code J8TL2D2. 
> http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
>   
> ------------------------------------------------------------------------
>
> _______________________________________________
> Meshlab-devel mailing list
> Mes...@li...
> https://lists.sourceforge.net/lists/listinfo/meshlab-devel
>   


-- 
Paolo Cignoni  -- Senior Researcher
Visual Computing Laboratory - ISTI - CNR 
http://vcg.isti.cnr.it/~cignoni

ISTI - CNR 
Via Moruzzi 1,
56124 Pisa
ITALY