PUBLIC

t_casteljau::t_casteljau(int) t_casteljau::~t_casteljau(void) t_casteljau::spatch_load_control(t_vector *,t_vector *) t_casteljau::do_casteljau(t_vector *,double *) t_casteljau::matrix_sab_bez(t_vector *,t_vector *) t_casteljau::matrix_uM(double *,double *) t_casteljau::matrix_make(void) t_casteljau::spatch_load_contr_look(void) t_casteljau::enumerate_indices(int)

t_casteljau
include	"amma/sl_cast.hh"
User Level:	Default
Library:	Slime
Example:	cri.cc
Section:	default.Andrew Stoddart
In Scope:	std

Comments:
A Casteljau set of tables for some nsimplex. nsimplex = 2=tri, 4=pent. 5=hex alldepths from depth_elev_min to depth_elev_max are accomodated. Note that the multiindices, Sabin net, etc are all 0 indexed, like Shoemake, not like Loop & DeRose

int n_simplex;

the simplex of this set of tables

int no_points[];

the number of points fro each depth

int * look_shoe;

the shoemake look up table, to store multindexed info in 1D and facilitate the castejau routine

int * look_derose;

a lookup table used to save having to recompute the offsets when converting the sabins to bezier control points

int * indices[];

all the possible indices enumerated, size no_points(d,n)*(n+1) for each depth.

int * elevate[];

tables containing all the offsets necessary for depth elevation

double matrix[];

M matrix; fixed size is a little wasteful.

t_casteljau(int n)

~t_casteljau()

the direct scheme uses these 2 functions ----------------------------------------

void spatch_load_control(t_vector * bezcont,t_vector * sabin)

Given the control points in Sabin, this constructs the sabin net p,q,v and then from these computes the bezier control points which are returned in bezcont

t_vector do_casteljau(t_vector * work,double * bary)

Given a working array containing the bezier control points, a barycentric coord, using the internal lookup tables output the point. The working array is overwritten. matrix based schemes --------------------

void matrix_sab_bez(t_vector * bezcont,t_vector * sabin)

does the same as spatch_load_control using the matrix

void matrix_uM(double * answer,double * bary)

does similar to do_casteljau, does dot product of u vector with matrix returns uM ready to be dotted with a sabin vector

void matrix_make()

make up the M matrix

void spatch_load_contr_look()

build up the lookup table look_derose

void enumerate_indices(int d)

use a local recursive function to enumerate the indices