15 March 2016
In short, no.
PicaVoxel uses "greedy" mesh generation to group voxels of the same colour together into larger quads. So essentially, if you had a cube of voxels all of the same colour, you would get a cube mesh made up of 6 quads, 12 triangles.
But as soon as you start subdividing with voxels of a different colour, the mesh gets more complicated accordingly.
On top of that, PicaVoxel uses a chunking system like most voxel engines. In PicaVoxel, it's used mostly to get around the 65k vertex limit imposed by Unity's mesh system. By default, PicaVoxel breaks up large volumes into chunks of 16x16x16 voxels. Each chunk is a separate Unity volume.
When you enable concave mesh generation on a PicaVoxel volume, it uses the same mesh generated for the rendered mesh as the collision mesh. Technically, it would be fairly simple to add a flag to ignore colour when creating the collision mesh, but you would then have to generate two meshes. If you're using PicaVoxel to destruct voxel volumes, that's twice the overhead at runtime when destructing/constructing voxels. On top of that, you would still have "seams" between the chunks.
If you're concerned about the effect that mesh seams might have on the physics, then there are probably better ways to generate colliders for your ground than rely on mesh colliders anyway.