[Plib-devel] fast smooth normals in 3DS loader

[Paolo L. <p.l...@ci...>]

I'd be ready to contribute the subject functionality, but I need to know
how.
I wrote a new class, presently named ssgaSmooth, stored in a .h and .cxx
files, whose methods are called in the new smooth_normals(). But
implementing it as an SSGAux class implies SSG to depend on SSGA.
Alternatively it could also be one more SSG class, or, forgetting someone
else access, an utility only accessed by the 3DS loader - either within the
loader's source file or in separate ones (better).

Public methods are:

	/********* public procs ************/
	void	addTriangles( int start, int end, sgVec3 *vertex_list, unsigned short
*vertex_index, sgVec3 *face_normals );
	void	includePolygon(int numVerts, Point3 *verts, float face_normal[], void
*user);
	void	makeVertexNormals( void );
	void	getVertexNormals( int num_faces, sgVec3 vertex_normals[] );

	/********* public option control procs ************/
	void	setFuzzFraction(float fuzzFraction);
	void	enableEdgePreservation(float minDot);
	void	disableEdgePreservation( void );

The original C code I adapted in a class is:

 * ANSI C code from the article
 * "Building Vertex Normals from an Unstructured Polygon List"
 * by Andrew Glassner, gla...@pa...
 * in "Graphics Gems IV", Academic Press, 1994

This is a fairly general tool for n-sided polygon, yet very fast, keeping
into account a "crease angle" for sharp EdgePreservation, that doesn't need
connectivity info. It hasn't presently the elegant interface of other SSG
classes, but it does the work very well. What I did additionally in the 3DS
loader is the management of an hash table for the smoothing groups.

Waiting for master opinion ...

greetings -

----------------------------------------------------------------------------
-
Paolo Leoncini						phone:  +39 (0823) 623134
Scientific Visualization & Virtual Reality	fax:    +39 (0823) 623126
CIRA - Italian Center for Aerospace Researches	mailto:p.l...@ci...
Via Maiorise - 81043 Capua (CE) Italy		http://www.cira.it/research/vis