Alex wrote:
As i say, "Holes aren't in the intersection of two mesh."

Sorry, I obviously didn't read very carefully

Alex wrote:
I use the recommended release "PolyVox-GIT-2e4bf67.zip (11th December 2011 – Recommended)"
Is it possible that a earlier revision will correct my problem ?
No, I don't think an earlier version will help.
Alex wrote:
I have just tried the Ogre::ProgressiveMesh to use the LOD system to create decimated mesh.
Polyvox decimator generate better mesh (ProgressiveMesh = odd faces).
Meshes generated with ProgressiveMesh don't fit together perfectly...
The problem with using the Ogre ProgressiveMesh is that it won't understand the input data properly. For example, I would expect that Ogre decimates the mesh based on certain properties such as the position of the vertices and their texture coordinates, but with a PolyVox mesh you also need to consider the material identifier so that you don't combine two vertices with different materials (though I realise you haven't started playing with multiple materials yet).
Ogre::ProgressiveMesh also won't understand that the meshes are meant to fit together, so it won't know that it has to exactly preserve the boundaries of the mesh (or at least, any changes it makes also have to be made to neighbouring meshes). This is probably why they don't fit together.
If you use an external decimation library then it will need to be configurable, so that you can express the the kind of properties described above. I don't know how well OpenMesh will fulfill this criteria but it will be investigated in the future.
Unfortunatly the decimation of voxel terrain meshes is really an open research area so I don't think you'll find a solution that just works 100%. It might be better to downsample the volume data (like the SmoothLODExample) but this hasn't been well tested, and will have it's own set of problems with multiple materials.
You might also find these blog posts interesting as they describe some of the reasons why this problem is difficult:
http://0fps.wordpress.com/2012/07/21/simplifying-isosurfaces-part-1/http://0fps.wordpress.com/2012/08/20/simplifying-isosurfaces-part-2/