Changeset 279

Timestamp:

09/16/05 15:27:52 (19 years ago)

Author:

mattausch

Message:

scireport updated

Location:

trunk/VUT/doc/SciReport

Files:

• figs/viewcell_part.eps (added)
• figs/viewcell_part.pdf (added)
• sampling.tex (modified) (5 diffs)

trunk/VUT/doc/SciReport/sampling.tex

@@ -303 +303 @@
   \#input polygons & 525 & 82235 \\\hline\hline
   BSP tree generation time & 0.016s & 10.328s \\\hline\hline
-  view cell insertion time & 0.016s & 7.984s \\\hline\hline
+  %%view cell insertion time & 0.016s & 7.984s \\\hline\hline
   \#nodes & 1137 & 597933 \\\hline\hline
   \#interior nodes & 568 & 298966\\\hline\hline
@@ -320 +320 @@
 \item We use a number of input view cells given in advance. As input
 view cell  any closed mesh can be applied. The only requirement is
-that the any two view cells do not overlap.  First the view cell
-polygons are extracted, and the BSP is build from these polygons using
-some global optimizations like tree balancing or least splits. Then
-one view cell after the other is inserted into the tree to find out
-the leaves where they are contained in. The polygons of the view cell
-are filtered down the tree, guiding the insertion process. Once we
-reach a leaf and there are no more polygons left, we terminate the
-tree subdivision. If we are on the inside of the last split plane
-(i.e., the leaf represents the inside of the view cell), we associate
-the leaf with the view cell (i.e., add a pointer to the view
-cell). One input view cell can be associated with many leaves, whereas
-each leafs has only one view cell.  Some statistics about using this
+that the any two view cells do not overlap.  The view cell
+polygons are extracted, storing a pointer to the parent view cell
+with the polygon. The BSP is build from these polygons using
+some global optimizations like tree balancing or least splits. The 
+polygons guide the split process as they are filtered down the tree. 
+The subdivision terminates when there is only one polygon left, which is coincident 
+to the last split plane. Then two leaves are created and the view cell pointer 
+(stored with the polygon) is inserted into the leaf representing the inside of the view cell. 
+One input view cell can be associated with many leaves in case
+a view cell was split during the traversal. On the other hand, each leafs corresponds
+to exactly one or no view cell.  Some statistics about using this
 method on the Vienna view cells set are given in
-table~\ref{tab:viewcell_bsp}.  However, sometimes a good set of view
+table~\ref{tab:viewcell_bsp}.
+
+However, sometimes a good set of view
 cells is not available. Or the scene is changed frequently, and the
 designer does not want to create new view cells on each change.  In
@@ -345 +346 @@
 the view cell borders  along some discontinuities in the visibility
 function.
+
+\begin{figure}[htb]
+  \centerline{
+    \includegraphics[height=0.35\textwidth,draft=\DRAFTFIGS]{figs/viewcell_part}
+      }
+  \caption{A good view cell partition with respect to the sample rays piercing the scene objects
+  and the view cell minimizes the number of rays
+  piercing more than one view cell. During subdivision, this can be achieved by aligning
+  the split plane with one of the long sides of occluder $O$. }
+  \label{fig:viewcell_part}
+\end{figure}
 
 \item  The view cell generation can be guided by the sampling
@@ -369 +381 @@
 occlusion power. This criterion can be naturally combined with the
 second one.  As termination criterion we can choose the minimum PVS /
-piercing ray size or the maximal tree depth.
+piercing ray size or the maximal tree depth. An illustration of
+a good and a bad choice of a split plane is given in figure~\ref{fig:viewcell_part}.
 \end{itemize}
 
@@ -381 +394 @@
 % which intersect subregions of the given view cell.
 
-\subsection{From-object based visibility}
+\subsection{From-Object Based Visibility}
 
  Our framework is based on the idea of sampling visibility by casting