great interest. The need for compact representations is fueled by the
increasing popularity of 3D content on the Internet, where transmission
bandwidth is a scarce resource.
In general encoding schemes for maximum mesh compression encode the mesh
topology seperately from the mesh geometry. Our work focuses on encoding
the topology. I will present a new edge-based approach. Our scheme gives
compact encodings and is very simple to implement. However, its main
advantage over other schemes are its 'natural' extensions:
- Triangle Strip Compression
Decomposing a triangle mesh into triangle strips is desireable for fast
rendering. ``Good'' stripifications reduce the number of vertices to be sent
through the graphics pipeline by a factor of 2 ~ 3. Finding an optimal
stripification seems to be NP-hard. Current mesh compression schemes
loose the stripification, which needs to be recomputed each time after
decoding the mesh. We present a scheme that stores both, topology and
stripification in a very efficient manner.
- Face Fixer
The majority of 3D models are not purely triangular. Typical polygon
meshes contain quadrangles, pentagons, hexagons, etc ... Current mesh
compression schemes triangulate such meshes prior to compression. The
original polygon information is either lost, or 'edge bits' are needed
to mark the edges added in the triangulation step.
The first has the disadvantage that it increases the number of faces
and corners of the mesh. Those can have properties associated (normals,
texture, colour) which then need to be specified multiple times. The
latter has the disadvantage that it makes the topology encoding more
%c\losinive.
Face Fixer is the first scheme to compress polygon meshes directly in
their polygonal representation. Furthermore structural information that
groups faces of the mesh into logical units can be easily incorporated
into the scheme.