[EMAIL PROTECTED]:
 |Randy,  a multigrid will do it for you.  Here's how you can  create the
 |overlap.   Slab the 0->m-1 field in dimension 0, all positions, thickness
 |0.  That'll create a series of m 1-D vertical slices.  Now slab the m->2m-1
 |field in dimension 0, position 0, thickness 0.  That'll create a  field
 |containing only the first vertical slice of the second grid.  Now append
 |the second result onto the first, with id = m.  That will create a series
 |with m+1 vertical slices, with series positions 0 through m.  Now Stack
 |that, and you'll get a regular grid with X from 0 to m.  CollectMultigrid
 |that with the original second grid, and you're done.

Thanks, I'll work with this further.  I noticed that it converts positions
from implicit to explicit in the Slab->Append step.  This is going to be a
problem since (pre-Reduce, pre-Rubbersheet), this is over 98MB of
coordinate data per tile.

The app may need to focus on the area around a tile boundary (which is why
I'm even working with tile merging), so the merged output would have to be
pre-Reduce -- for that area at least.

I might combine my simple LOD technique with your Slab-Append approach, or
write a custom loader.  I'll have to think about it.

Thanks,

Randy

-- 
Randall Hopper (mailto:[EMAIL PROTECTED])
EPA Scientific Visualization Center
US EPA MD/24 ERC-1A; RTP, NC 27711

Reply via email to