int
soidDoit(limnObject *obj, int look,
int gtype, float gamma, int res,
float AB[2], float ten[7]) {
int partIdx, axis;
float cl, cp, qA, qB, eval[3], evec[9], matA[16], matB[16];
if (AB) {
qA = AB[0];
qB = AB[1];
axis = 2;
} else {
tenEigensolve_f(eval, evec, ten);
ELL_SORT3(eval[0], eval[1], eval[2], cl);
cl = (eval[0] - eval[1])/(eval[0] + eval[1] + eval[2]);
cp = 2*(eval[1] - eval[2])/(eval[0] + eval[1] + eval[2]);
if (cl > cp) {
axis = 0;
qA = pow(1-cp, gamma);
qB = pow(1-cl, gamma);
} else {
axis = 2;
qA = pow(1-cl, gamma);
qB = pow(1-cp, gamma);
}
/*
fprintf(stderr, "eval = %g %g %g -> cl=%g %s cp=%g -> axis = %d\n",
eval[0], eval[1], eval[2], cl, cl > cp ? ">" : "<", cp, axis);
*/
}
if (tenGlyphTypeBox == gtype) {
partIdx = limnObjectCubeAdd(obj, look);
} else if (tenGlyphTypeSphere == gtype) {
partIdx = limnObjectPolarSphereAdd(obj, look,
0, 2*res, res);
} else {
partIdx = limnObjectPolarSuperquadAdd(obj, look,
axis, qA, qB, 2*res, res);
}
ELL_4M_IDENTITY_SET(matA);
ELL_4V_SET(matB + 0*4, eval[0], 0, 0, 0);
ELL_4V_SET(matB + 1*4, 0, eval[1], 0, 0);
ELL_4V_SET(matB + 2*4, 0, 0, eval[2], 0);
ELL_4V_SET(matB + 3*4, 0, 0, 0, 1);
ELL_4M_SCALE_SET(matB, eval[0], eval[1], eval[2]);
ell_4m_post_mul_f(matA, matB);
ELL_4V_SET(matB + 0*4, evec[0 + 0*3], evec[0 + 1*3], evec[0 + 2*3], 0);
ELL_4V_SET(matB + 1*4, evec[1 + 0*3], evec[1 + 1*3], evec[1 + 2*3], 0);
ELL_4V_SET(matB + 2*4, evec[2 + 0*3], evec[2 + 1*3], evec[2 + 2*3], 0);
ELL_4V_SET(matB + 3*4, 0, 0, 0, 1);
ell_4m_post_mul_f(matA, matB);
limnObjectPartTransform(obj, partIdx, matA);
return partIdx;
}
int
main(int argc, const char *argv[]) {
const char *me;
char *err, *outS;
float p[3], q[4], mR[9], eval[3]={0,0,0}, scale[3], len, sh, cl, cp, qA, qB;
float matA[16], matB[16], os, rad, AB[2];
hestOpt *hopt=NULL;
airArray *mop;
limnObject *obj;
limnLook *look; int lookRod, lookSoid;
int partIdx=-1; /* sssh */
int res, axis, sphere;
FILE *file;
me = argv[0];
hestOptAdd(&hopt, "sc", "scalings", airTypeFloat, 3, 3, scale, "1 1 1",
"axis-aligned scaling to do on ellipsoid");
hestOptAdd(&hopt, "AB", "A, B exponents", airTypeFloat, 2, 2, AB, "nan nan",
"Directly set the A, B parameters to the superquadric surface, "
"over-riding the default behavior of determining them from the "
"scalings \"-sc\" as superquadric tensor glyphs");
hestOptAdd(&hopt, "os", "over-all scaling", airTypeFloat, 1, 1, &os, "1",
"over-all scaling (multiplied by scalings)");
hestOptAdd(&hopt, "sh", "superquad sharpness", airTypeFloat, 1, 1, &sh, "0",
"how much to sharpen edges as a "
"function of differences between eigenvalues");
hestOptAdd(&hopt, "sphere", NULL, airTypeInt, 0, 0, &sphere, NULL,
"use a sphere instead of a superquadric");
hestOptAdd(&hopt, "p", "x y z", airTypeFloat, 3, 3, p, "0 0 0",
"location in quaternion quotient space");
hestOptAdd(&hopt, "r", "radius", airTypeFloat, 1, 1, &rad, "0.015",
"black axis cylinder radius (or 0.0 to not drawn these)");
hestOptAdd(&hopt, "res", "resolution", airTypeInt, 1, 1, &res, "25",
"tesselation resolution for both glyph and axis cylinders");
hestOptAdd(&hopt, "o", "output OFF", airTypeString, 1, 1, &outS, "out.off",
"output file to save OFF into");
hestParseOrDie(hopt, argc-1, argv+1, NULL,
me, info, AIR_TRUE, AIR_TRUE, AIR_TRUE);
mop = airMopNew();
airMopAdd(mop, hopt, (airMopper)hestOptFree, airMopAlways);
airMopAdd(mop, hopt, (airMopper)hestParseFree, airMopAlways);
obj = limnObjectNew(100, AIR_FALSE);
airMopAdd(mop, obj, (airMopper)limnObjectNix, airMopAlways);
/* create limnLooks for ellipsoid and for rods */
lookSoid = limnObjectLookAdd(obj);
look = obj->look + lookSoid;
ELL_4V_SET(look->rgba, 1, 1, 1, 1);
ELL_3V_SET(look->kads, 0.2, 0.8, 0);
look->spow = 0;
lookRod = limnObjectLookAdd(obj);
look = obj->look + lookRod;
ELL_4V_SET(look->rgba, 0, 0, 0, 1);
ELL_3V_SET(look->kads, 1, 0, 0);
look->spow = 0;
ELL_4V_SET(q, 1, p[0], p[1], p[2]);
ELL_4V_NORM(q, q, len);
ell_q_to_3m_f(mR, q);
if (AIR_EXISTS(AB[0]) && AIR_EXISTS(AB[1])) {
qA = AB[0];
qB = AB[1];
axis = 2;
} else {
ELL_3V_SCALE(scale, os, scale);
ELL_3V_COPY(eval, scale);
ELL_SORT3(eval[0], eval[1], eval[2], cl);
cl = (eval[0] - eval[1])/(eval[0] + eval[1] + eval[2]);
cp = 2*(eval[1] - eval[2])/(eval[0] + eval[1] + eval[2]);
if (cl > cp) {
axis = ELL_MAX3_IDX(scale[0], scale[1], scale[2]);
qA = pow(1-cp, sh);
qB = pow(1-cl, sh);
} else {
axis = ELL_MIN3_IDX(scale[0], scale[1], scale[2]);
qA = pow(1-cl, sh);
qB = pow(1-cp, sh);
}
/*
fprintf(stderr, "eval = %g %g %g -> cl=%g %s cp=%g -> axis = %d\n",
eval[0], eval[1], eval[2], cl, cl > cp ? ">" : "<", cp, axis);
*/
}
if (sphere) {
partIdx = limnObjectPolarSphereAdd(obj, lookSoid,
0, 2*res, res);
} else {
partIdx = limnObjectPolarSuperquadAdd(obj, lookSoid,
axis, qA, qB, 2*res, res);
}
ELL_4M_IDENTITY_SET(matA);
ELL_4M_SCALE_SET(matB, scale[0], scale[1], scale[2]);
ell_4m_post_mul_f(matA, matB);
ELL_43M_INSET(matB, mR);
ell_4m_post_mul_f(matA, matB);
limnObjectPartTransform(obj, partIdx, matA);
if (rad) {
partIdx = limnObjectCylinderAdd(obj, lookRod, 0, res);
ELL_4M_IDENTITY_SET(matA);
ELL_4M_SCALE_SET(matB, (1-eval[0])/2, rad, rad);
ell_4m_post_mul_f(matA, matB);
ELL_4M_TRANSLATE_SET(matB, (1+eval[0])/2, 0.0, 0.0);
ell_4m_post_mul_f(matA, matB);
limnObjectPartTransform(obj, partIdx, matA);
partIdx = limnObjectCylinderAdd(obj, lookRod, 0, res);
ELL_4M_IDENTITY_SET(matA);
ELL_4M_SCALE_SET(matB, (1-eval[0])/2, rad, rad);
ell_4m_post_mul_f(matA, matB);
ELL_4M_TRANSLATE_SET(matB, -(1+eval[0])/2, 0.0, 0.0);
ell_4m_post_mul_f(matA, matB);
limnObjectPartTransform(obj, partIdx, matA);
partIdx = limnObjectCylinderAdd(obj, lookRod, 1, res);
ELL_4M_IDENTITY_SET(matA);
ELL_4M_SCALE_SET(matB, rad, (1-eval[1])/2, rad);
ell_4m_post_mul_f(matA, matB);
ELL_4M_TRANSLATE_SET(matB, 0.0, (1+eval[1])/2, 0.0);
ell_4m_post_mul_f(matA, matB);
limnObjectPartTransform(obj, partIdx, matA);
partIdx = limnObjectCylinderAdd(obj, lookRod, 1, res);
ELL_4M_IDENTITY_SET(matA);
ELL_4M_SCALE_SET(matB, rad, (1-eval[1])/2, rad);
ell_4m_post_mul_f(matA, matB);
ELL_4M_TRANSLATE_SET(matB, 0.0, -(1+eval[1])/2, 0.0);
ell_4m_post_mul_f(matA, matB);
limnObjectPartTransform(obj, partIdx, matA);
partIdx = limnObjectCylinderAdd(obj, lookRod, 2, res);
ELL_4M_IDENTITY_SET(matA);
ELL_4M_SCALE_SET(matB, rad, rad, (1-eval[2])/2);
ell_4m_post_mul_f(matA, matB);
ELL_4M_TRANSLATE_SET(matB, 0.0, 0.0, (1+eval[2])/2);
ell_4m_post_mul_f(matA, matB);
limnObjectPartTransform(obj, partIdx, matA);
partIdx = limnObjectCylinderAdd(obj, lookRod, 2, res);
ELL_4M_IDENTITY_SET(matA);
ELL_4M_SCALE_SET(matB, rad, rad, (1-eval[2])/2);
ell_4m_post_mul_f(matA, matB);
ELL_4M_TRANSLATE_SET(matB, 0.0, 0.0, -(1+eval[2])/2);
ell_4m_post_mul_f(matA, matB);
limnObjectPartTransform(obj, partIdx, matA);
}
file = airFopen(outS, stdout, "w");
airMopAdd(mop, file, (airMopper)airFclose, airMopAlways);
if (limnObjectWriteOFF(file, obj)) {
airMopAdd(mop, err = biffGetDone(LIMN), airFree, airMopAlways);
fprintf(stderr, "%s: trouble:\n%s\n", me, err);
airMopError(mop); return 1;
}
airMopOkay(mop);
return 0;
}
int
main(int argc, const char *argv[]) {
const char *me;
char *err, *outS;
double eval[3], matA[9], matB[9], sval[3], uu[9], vv[9], escl[5],
view[3];
float matAf[9], matBf[16];
float pp[3], qq[4], mR[9], len, gamma;
float os, vs, rad, AB[2], ten[7];
hestOpt *hopt=NULL;
airArray *mop;
limnObject *obj;
limnLook *look; int lookRod, lookSoid;
float kadsRod[3], kadsSoid[3];
int gtype, partIdx=-1; /* sssh */
int res;
FILE *file;
me = argv[0];
hestOptAdd(&hopt, "sc", "evals", airTypeDouble, 3, 3, eval, "1 1 1",
"original eigenvalues of tensor to be visualized");
hestOptAdd(&hopt, "AB", "A, B exponents", airTypeFloat, 2, 2, AB, "nan nan",
"Directly set the A, B parameters to the superquadric surface, "
"over-riding the default behavior of determining them from the "
"scalings \"-sc\" as superquadric tensor glyphs");
hestOptAdd(&hopt, "os", "over-all scaling", airTypeFloat, 1, 1, &os, "1",
"over-all scaling (multiplied by scalings)");
hestOptAdd(&hopt, "vs", "view-dir scaling", airTypeFloat, 1, 1, &vs, "1",
"scaling along view-direction (to show off bas-relief "
"ambibuity of ellipsoids versus superquads)");
hestOptAdd(&hopt, "es", "extra scaling", airTypeDouble, 5, 5, escl,
"2 1 0 0 1", "extra scaling specified with five values "
"0:tensor|1:geometry|2:none vx vy vz scaling");
hestOptAdd(&hopt, "fr", "from (eye) point", airTypeDouble, 3, 3, &view,
"4 4 4", "eye point, needed for non-unity \"-vs\"");
hestOptAdd(&hopt, "gamma", "superquad sharpness", airTypeFloat, 1, 1,
&gamma, "0",
"how much to sharpen edges as a "
"function of differences between eigenvalues");
hestOptAdd(&hopt, "g", "glyph shape", airTypeEnum, 1, 1, >ype, "sqd",
"glyph to use; not all are implemented here",
NULL, tenGlyphType);
hestOptAdd(&hopt, "pp", "x y z", airTypeFloat, 3, 3, pp, "0 0 0",
"transform: rotation identified by"
"location in quaternion quotient space");
hestOptAdd(&hopt, "r", "radius", airTypeFloat, 1, 1, &rad, "0.015",
"black axis cylinder radius (or 0.0 to not drawn these)");
hestOptAdd(&hopt, "res", "resolution", airTypeInt, 1, 1, &res, "25",
"tesselation resolution for both glyph and axis cylinders");
hestOptAdd(&hopt, "pg", "ka kd ks", airTypeFloat, 3, 3, kadsSoid,
"0.2 0.8 0.0",
"phong coefficients for glyph");
hestOptAdd(&hopt, "pr", "ka kd ks", airTypeFloat, 3, 3, kadsRod, "1 0 0",
"phong coefficients for black rods (if being drawn)");
hestOptAdd(&hopt, "o", "output OFF", airTypeString, 1, 1, &outS, "out.off",
"output file to save OFF into");
hestParseOrDie(hopt, argc-1, argv+1, NULL,
me, info, AIR_TRUE, AIR_TRUE, AIR_TRUE);
mop = airMopNew();
airMopAdd(mop, hopt, (airMopper)hestOptFree, airMopAlways);
airMopAdd(mop, hopt, (airMopper)hestParseFree, airMopAlways);
obj = limnObjectNew(1000, AIR_TRUE);
airMopAdd(mop, obj, (airMopper)limnObjectNix, airMopAlways);
if (!( 0 == escl[0] || 1 == escl[0] || 2 == escl[0] )) {
fprintf(stderr, "%s: escl[0] %g not 0, 1 or 2\n", me, escl[0]);
airMopError(mop); return 1;
}
if (!(tenGlyphTypeBox == gtype ||
tenGlyphTypeSphere == gtype ||
tenGlyphTypeSuperquad == gtype)) {
fprintf(stderr, "%s: got %s %s, but here only do %s, %s, or %s\n", me,
tenGlyphType->name,
airEnumStr(tenGlyphType, gtype),
airEnumStr(tenGlyphType, tenGlyphTypeBox),
airEnumStr(tenGlyphType, tenGlyphTypeSphere),
airEnumStr(tenGlyphType, tenGlyphTypeSuperquad));
airMopError(mop); return 1;
}
/* create limnLooks for glyph and for rods */
lookSoid = limnObjectLookAdd(obj);
look = obj->look + lookSoid;
ELL_4V_SET(look->rgba, 1, 1, 1, 1);
ELL_3V_COPY(look->kads, kadsSoid);
look->spow = 0;
lookRod = limnObjectLookAdd(obj);
look = obj->look + lookRod;
ELL_4V_SET(look->rgba, 0, 0, 0, 1);
ELL_3V_COPY(look->kads, kadsRod);
look->spow = 0;
ELL_3M_IDENTITY_SET(matA); /* A = I */
ELL_3V_SCALE(eval, os, eval);
ELL_3M_SCALE_SET(matB, eval[0], eval[1], eval[2]); /* B = diag(eval) */
ell_3m_post_mul_d(matA, matB); /* A = B*A = diag(eval) */
if (0 == escl[0]) {
scalingMatrix(matB, escl + 1, escl[4]);
ell_3m_post_mul_d(matA, matB);
}
if (1 != vs) {
if (!ELL_3V_LEN(view)) {
fprintf(stderr, "%s: need non-zero view for vs %g != 1\n", me, vs);
airMopError(mop); return 1;
}
scalingMatrix(matB, view, vs);
/* the scaling along the view direction is a symmetric matrix,
but applying that scaling to the symmetric input tensor
is not necessarily symmetric */
ell_3m_post_mul_d(matA, matB); /* A = B*A */
}
/* so we do an SVD to get rotation U and the scalings sval[] */
/* U * diag(sval) * V */
ell_3m_svd_d(uu, sval, vv, matA, AIR_TRUE);
/*
fprintf(stderr, "%s: ____________________________________\n", me);
fprintf(stderr, "%s: mat = \n", me);
ell_3m_print_d(stderr, matA);
fprintf(stderr, "%s: uu = \n", me);
ell_3m_print_d(stderr, uu);
ELL_3M_TRANSPOSE(matC, uu);
ELL_3M_MUL(matB, uu, matC);
fprintf(stderr, "%s: uu * uu^T = \n", me);
ell_3m_print_d(stderr, matB);
fprintf(stderr, "%s: sval = %g %g %g\n", me, sval[0], sval[1], sval[2]);
fprintf(stderr, "%s: vv = \n", me);
ell_3m_print_d(stderr, vv);
ELL_3M_MUL(matB, vv, vv);
fprintf(stderr, "%s: vv * vv^T = \n", me);
ELL_3M_TRANSPOSE(matC, vv);
ELL_3M_MUL(matB, vv, matC);
ell_3m_print_d(stderr, matB);
ELL_3M_IDENTITY_SET(matA);
ell_3m_pre_mul_d(matA, uu);
ELL_3M_SCALE_SET(matB, sval[0], sval[1], sval[2]);
ell_3m_pre_mul_d(matA, matB);
ell_3m_pre_mul_d(matA, vv);
fprintf(stderr, "%s: uu * diag(sval) * vv = \n", me);
ell_3m_print_d(stderr, matA);
fprintf(stderr, "%s: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n", me);
*/
/* now create symmetric matrix out of U and sval */
/* A = I */
ELL_3M_IDENTITY_SET(matA);
ell_3m_pre_mul_d(matA, uu); /* A = A*U = I*U = U */
ELL_3M_SCALE_SET(matB, sval[0], sval[1], sval[2]); /* B = diag(sval) */
ell_3m_pre_mul_d(matA, matB); /* A = U*diag(sval) */
ELL_3M_TRANSPOSE(matB, uu);
ell_3m_pre_mul_d(matA, matB); /* A = U*diag(sval)*U^T */
TEN_M2T(ten, matA);
partIdx = soidDoit(obj, lookSoid,
gtype, gamma, res,
(AIR_EXISTS(AB[0]) && AIR_EXISTS(AB[1])) ? AB : NULL,
ten);
if (1 == escl[0]) {
scalingMatrix(matB, escl + 1, escl[4]);
ELL_43M_INSET(matBf, matB);
limnObjectPartTransform(obj, partIdx, matBf);
}
/* this is a rotate on the geomtry; nothing to do with the tensor */
ELL_4V_SET(qq, 1, pp[0], pp[1], pp[2]);
ELL_4V_NORM(qq, qq, len);
ell_q_to_3m_f(mR, qq);
ELL_43M_INSET(matBf, mR);
limnObjectPartTransform(obj, partIdx, matBf);
if (rad) {
partIdx = limnObjectCylinderAdd(obj, lookRod, 0, res);
ELL_4M_IDENTITY_SET(matAf);
ELL_4M_SCALE_SET(matBf, (1-eval[0])/2, rad, rad);
ell_4m_post_mul_f(matAf, matBf);
ELL_4M_TRANSLATE_SET(matBf, (1+eval[0])/2, 0.0, 0.0);
ell_4m_post_mul_f(matAf, matBf);
limnObjectPartTransform(obj, partIdx, matAf);
partIdx = limnObjectCylinderAdd(obj, lookRod, 0, res);
ELL_4M_IDENTITY_SET(matAf);
ELL_4M_SCALE_SET(matBf, (1-eval[0])/2, rad, rad);
ell_4m_post_mul_f(matAf, matBf);
ELL_4M_TRANSLATE_SET(matBf, -(1+eval[0])/2, 0.0, 0.0);
ell_4m_post_mul_f(matAf, matBf);
limnObjectPartTransform(obj, partIdx, matAf);
partIdx = limnObjectCylinderAdd(obj, lookRod, 1, res);
ELL_4M_IDENTITY_SET(matAf);
ELL_4M_SCALE_SET(matBf, rad, (1-eval[1])/2, rad);
ell_4m_post_mul_f(matAf, matBf);
ELL_4M_TRANSLATE_SET(matBf, 0.0, (1+eval[1])/2, 0.0);
ell_4m_post_mul_f(matAf, matBf);
limnObjectPartTransform(obj, partIdx, matAf);
partIdx = limnObjectCylinderAdd(obj, lookRod, 1, res);
ELL_4M_IDENTITY_SET(matAf);
ELL_4M_SCALE_SET(matBf, rad, (1-eval[1])/2, rad);
ell_4m_post_mul_f(matAf, matBf);
ELL_4M_TRANSLATE_SET(matBf, 0.0, -(1+eval[1])/2, 0.0);
ell_4m_post_mul_f(matAf, matBf);
limnObjectPartTransform(obj, partIdx, matAf);
partIdx = limnObjectCylinderAdd(obj, lookRod, 2, res);
ELL_4M_IDENTITY_SET(matAf);
ELL_4M_SCALE_SET(matBf, rad, rad, (1-eval[2])/2);
ell_4m_post_mul_f(matAf, matBf);
ELL_4M_TRANSLATE_SET(matBf, 0.0, 0.0, (1+eval[2])/2);
ell_4m_post_mul_f(matAf, matBf);
limnObjectPartTransform(obj, partIdx, matAf);
partIdx = limnObjectCylinderAdd(obj, lookRod, 2, res);
ELL_4M_IDENTITY_SET(matAf);
ELL_4M_SCALE_SET(matBf, rad, rad, (1-eval[2])/2);
ell_4m_post_mul_f(matAf, matBf);
ELL_4M_TRANSLATE_SET(matBf, 0.0, 0.0, -(1+eval[2])/2);
ell_4m_post_mul_f(matAf, matBf);
limnObjectPartTransform(obj, partIdx, matAf);
}
file = airFopen(outS, stdout, "w");
airMopAdd(mop, file, (airMopper)airFclose, airMopAlways);
if (limnObjectWriteOFF(file, obj)) {
airMopAdd(mop, err = biffGetDone(LIMN), airFree, airMopAlways);
fprintf(stderr, "%s: trouble:\n%s\n", me, err);
airMopError(mop); return 1;
}
airMopOkay(mop);
return 0;
}
int
main(int argc, const char *argv[]) {
const char *me;
char *err, *outS;
double scale[3], matA[9], matB[9], matC[9], sval[3], uu[9], vv[9];
float matAf[9], matBf[16];
float p[3], q[4], mR[9], len, gamma;
float os, vs, rad, AB[2], ten[7], view[3];
hestOpt *hopt=NULL;
airArray *mop;
limnObject *obj;
limnLook *look; int lookRod, lookSoid;
int partIdx=-1; /* sssh */
int res, sphere;
FILE *file;
me = argv[0];
hestOptAdd(&hopt, "sc", "scalings", airTypeDouble, 3, 3, scale, "1 1 1",
"axis-aligned scaling to do on ellipsoid");
hestOptAdd(&hopt, "AB", "A, B exponents", airTypeFloat, 2, 2, AB, "nan nan",
"Directly set the A, B parameters to the superquadric surface, "
"over-riding the default behavior of determining them from the "
"scalings \"-sc\" as superquadric tensor glyphs");
hestOptAdd(&hopt, "os", "over-all scaling", airTypeFloat, 1, 1, &os, "1",
"over-all scaling (multiplied by scalings)");
hestOptAdd(&hopt, "vs", "over-all scaling", airTypeFloat, 1, 1, &vs, "1",
"scaling along view-direction (to show off bas-relief "
"ambibuity of ellipsoids versus superquads)");
hestOptAdd(&hopt, "fr", "from (eye) point", airTypeFloat, 3, 3, &view,
"4 4 4", "eye point, needed for non-unity \"-vs\"");
hestOptAdd(&hopt, "gamma", "superquad sharpness", airTypeFloat, 1, 1,
&gamma, "0",
"how much to sharpen edges as a "
"function of differences between eigenvalues");
hestOptAdd(&hopt, "sphere", NULL, airTypeInt, 0, 0, &sphere, NULL,
"use a sphere instead of a superquadric");
hestOptAdd(&hopt, "p", "x y z", airTypeFloat, 3, 3, p, "0 0 0",
"location in quaternion quotient space");
hestOptAdd(&hopt, "r", "radius", airTypeFloat, 1, 1, &rad, "0.015",
"black axis cylinder radius (or 0.0 to not drawn these)");
hestOptAdd(&hopt, "res", "resolution", airTypeInt, 1, 1, &res, "25",
"tesselation resolution for both glyph and axis cylinders");
hestOptAdd(&hopt, "o", "output OFF", airTypeString, 1, 1, &outS, "out.off",
"output file to save OFF into");
hestParseOrDie(hopt, argc-1, argv+1, NULL,
me, info, AIR_TRUE, AIR_TRUE, AIR_TRUE);
mop = airMopNew();
airMopAdd(mop, hopt, (airMopper)hestOptFree, airMopAlways);
airMopAdd(mop, hopt, (airMopper)hestParseFree, airMopAlways);
obj = limnObjectNew(1000, AIR_TRUE);
airMopAdd(mop, obj, (airMopper)limnObjectNix, airMopAlways);
/* create limnLooks for ellipsoid and for rods */
lookSoid = limnObjectLookAdd(obj);
look = obj->look + lookSoid;
ELL_4V_SET(look->rgba, 1, 1, 1, 1);
ELL_3V_SET(look->kads, 0.2, 0.8, 0);
look->spow = 0;
lookRod = limnObjectLookAdd(obj);
look = obj->look + lookRod;
ELL_4V_SET(look->rgba, 0, 0, 0, 1);
ELL_3V_SET(look->kads, 1, 0, 0);
look->spow = 0;
ELL_3M_IDENTITY_SET(matA);
ELL_3V_SCALE(scale, os, scale);
ELL_3M_SCALE_SET(matB, scale[0], scale[1], scale[2]);
ell_3m_post_mul_d(matA, matB);
if (1 != vs) {
ELL_3V_NORM(view, view, len);
if (!len) {
/* HEY: perhaps do more diplomatic error message here */
fprintf(stderr, "%s: stupido!\n", me);
exit(1);
}
ELL_3MV_OUTER(matB, view, view);
ELL_3M_SCALE(matB, vs-1, matB);
ELL_3M_IDENTITY_SET(matC);
ELL_3M_ADD2(matB, matC, matB);
ell_3m_post_mul_d(matA, matB);
}
ell_3m_svd_d(uu, sval, vv, matA, AIR_TRUE);
/*
fprintf(stderr, "%s: ____________________________________\n", me);
fprintf(stderr, "%s: mat = \n", me);
ell_3m_print_d(stderr, matA);
fprintf(stderr, "%s: uu = \n", me);
ell_3m_print_d(stderr, uu);
ELL_3M_TRANSPOSE(matC, uu);
ELL_3M_MUL(matB, uu, matC);
fprintf(stderr, "%s: uu * uu^T = \n", me);
ell_3m_print_d(stderr, matB);
fprintf(stderr, "%s: sval = %g %g %g\n", me, sval[0], sval[1], sval[2]);
fprintf(stderr, "%s: vv = \n", me);
ell_3m_print_d(stderr, vv);
ELL_3M_MUL(matB, vv, vv);
fprintf(stderr, "%s: vv * vv^T = \n", me);
ELL_3M_TRANSPOSE(matC, vv);
ELL_3M_MUL(matB, vv, matC);
ell_3m_print_d(stderr, matB);
ELL_3M_IDENTITY_SET(matA);
ell_3m_pre_mul_d(matA, uu);
ELL_3M_SCALE_SET(matB, sval[0], sval[1], sval[2]);
ell_3m_pre_mul_d(matA, matB);
ell_3m_pre_mul_d(matA, vv);
fprintf(stderr, "%s: uu * diag(sval) * vv = \n", me);
ell_3m_print_d(stderr, matA);
fprintf(stderr, "%s: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n", me);
*/
ELL_3M_IDENTITY_SET(matA);
ell_3m_pre_mul_d(matA, uu);
ELL_3M_SCALE_SET(matB, sval[0], sval[1], sval[2]);
ell_3m_pre_mul_d(matA, matB);
ELL_3M_TRANSPOSE(matB, uu);
ell_3m_pre_mul_d(matA, matB);
TEN_M2T(ten, matA);
partIdx = soidDoit(obj, lookSoid,
sphere, gamma, res,
(AIR_EXISTS(AB[0]) && AIR_EXISTS(AB[1])) ? AB : NULL,
ten);
ELL_4V_SET(q, 1, p[0], p[1], p[2]);
ELL_4V_NORM(q, q, len);
ell_q_to_3m_f(mR, q);
ELL_43M_INSET(matBf, mR);
limnObjectPartTransform(obj, partIdx, matBf);
if (rad) {
partIdx = limnObjectCylinderAdd(obj, lookRod, 0, res);
ELL_4M_IDENTITY_SET(matAf);
ELL_4M_SCALE_SET(matBf, (1-scale[0])/2, rad, rad);
ell_4m_post_mul_f(matAf, matBf);
ELL_4M_TRANSLATE_SET(matBf, (1+scale[0])/2, 0.0, 0.0);
ell_4m_post_mul_f(matAf, matBf);
limnObjectPartTransform(obj, partIdx, matAf);
partIdx = limnObjectCylinderAdd(obj, lookRod, 0, res);
ELL_4M_IDENTITY_SET(matAf);
ELL_4M_SCALE_SET(matBf, (1-scale[0])/2, rad, rad);
ell_4m_post_mul_f(matAf, matBf);
ELL_4M_TRANSLATE_SET(matBf, -(1+scale[0])/2, 0.0, 0.0);
ell_4m_post_mul_f(matAf, matBf);
limnObjectPartTransform(obj, partIdx, matAf);
partIdx = limnObjectCylinderAdd(obj, lookRod, 1, res);
ELL_4M_IDENTITY_SET(matAf);
ELL_4M_SCALE_SET(matBf, rad, (1-scale[1])/2, rad);
ell_4m_post_mul_f(matAf, matBf);
ELL_4M_TRANSLATE_SET(matBf, 0.0, (1+scale[1])/2, 0.0);
ell_4m_post_mul_f(matAf, matBf);
limnObjectPartTransform(obj, partIdx, matAf);
partIdx = limnObjectCylinderAdd(obj, lookRod, 1, res);
ELL_4M_IDENTITY_SET(matAf);
ELL_4M_SCALE_SET(matBf, rad, (1-scale[1])/2, rad);
ell_4m_post_mul_f(matAf, matBf);
ELL_4M_TRANSLATE_SET(matBf, 0.0, -(1+scale[1])/2, 0.0);
ell_4m_post_mul_f(matAf, matBf);
limnObjectPartTransform(obj, partIdx, matAf);
partIdx = limnObjectCylinderAdd(obj, lookRod, 2, res);
ELL_4M_IDENTITY_SET(matAf);
ELL_4M_SCALE_SET(matBf, rad, rad, (1-scale[2])/2);
ell_4m_post_mul_f(matAf, matBf);
ELL_4M_TRANSLATE_SET(matBf, 0.0, 0.0, (1+scale[2])/2);
ell_4m_post_mul_f(matAf, matBf);
limnObjectPartTransform(obj, partIdx, matAf);
partIdx = limnObjectCylinderAdd(obj, lookRod, 2, res);
ELL_4M_IDENTITY_SET(matAf);
ELL_4M_SCALE_SET(matBf, rad, rad, (1-scale[2])/2);
ell_4m_post_mul_f(matAf, matBf);
ELL_4M_TRANSLATE_SET(matBf, 0.0, 0.0, -(1+scale[2])/2);
ell_4m_post_mul_f(matAf, matBf);
limnObjectPartTransform(obj, partIdx, matAf);
}
file = airFopen(outS, stdout, "w");
airMopAdd(mop, file, (airMopper)airFclose, airMopAlways);
if (limnObjectWriteOFF(file, obj)) {
airMopAdd(mop, err = biffGetDone(LIMN), airFree, airMopAlways);
fprintf(stderr, "%s: trouble:\n%s\n", me, err);
airMopError(mop); return 1;
}
airMopOkay(mop);
return 0;
}
int
main(int argc, char *argv[]) {
char *me, *err, *outS;
limnCamera *cam;
float matA[16], matB[16], winscale, edgeWidth[5];
hestOpt *hopt=NULL;
airArray *mop;
limnObject *obj;
limnLook *look;
int lookIdx;
limnWindow *win;
int partIdx, wire, concave;
Nrrd *nmap;
mop = airMopNew();
cam = limnCameraNew();
airMopAdd(mop, cam, (airMopper)limnCameraNix, airMopAlways);
me = argv[0];
hestOptAdd(&hopt, "fr", "from point", airTypeDouble, 3, 3, cam->from,"4 4 4",
"position of camera, used to determine view vector");
hestOptAdd(&hopt, "at", "at point", airTypeDouble, 3, 3, cam->at, "0 0 0",
"camera look-at point, used to determine view vector");
hestOptAdd(&hopt, "up", "up vector", airTypeDouble, 3, 3, cam->up, "0 0 1",
"camera pseudo-up vector, used to determine view coordinates");
hestOptAdd(&hopt, "rh", NULL, airTypeInt, 0, 0, &(cam->rightHanded), NULL,
"use a right-handed UVN frame (V points down)");
hestOptAdd(&hopt, "or", NULL, airTypeInt, 0, 0, &(cam->orthographic), NULL,
"use orthogonal projection");
hestOptAdd(&hopt, "ur", "uMin uMax", airTypeDouble, 2, 2, cam->uRange,
"-1 1", "range in U direction of image plane");
hestOptAdd(&hopt, "vr", "vMin vMax", airTypeDouble, 2, 2, cam->vRange,
"-1 1", "range in V direction of image plane");
hestOptAdd(&hopt, "e", "envmap", airTypeOther, 1, 1, &nmap, NULL,
"16checker-based environment map",
NULL, NULL, nrrdHestNrrd);
hestOptAdd(&hopt, "ws", "winscale", airTypeFloat, 1, 1, &winscale,
"200", "world to points (PostScript) scaling");
hestOptAdd(&hopt, "wire", NULL, airTypeInt, 0, 0, &wire, NULL,
"just do wire-frame rendering");
hestOptAdd(&hopt, "concave", NULL, airTypeInt, 0, 0, &concave, NULL,
"use slightly buggy rendering method suitable for "
"concave or self-occluding objects");
hestOptAdd(&hopt, "wd", "5 widths", airTypeFloat, 5, 5, edgeWidth,
"0.0 0.0 3.0 2.0 0.0",
"width of edges drawn for five kinds of "
"edges: back non-crease, back crease, "
"silohuette, front crease, front non-crease");
hestOptAdd(&hopt, "o", "output PS", airTypeString, 1, 1, &outS, "out.ps",
"output file to render postscript into");
hestParseOrDie(hopt, argc-1, argv+1, NULL,
me, info, AIR_TRUE, AIR_TRUE, AIR_TRUE);
airMopAdd(mop, hopt, (airMopper)hestOptFree, airMopAlways);
airMopAdd(mop, hopt, (airMopper)hestParseFree, airMopAlways);
cam->neer = -0.000000001;
cam->dist = 0;
cam->faar = 0.0000000001;
cam->atRelative = AIR_TRUE;
if (limnCameraUpdate(cam)) {
fprintf(stderr, "%s: trouble:\n%s\n", me, err = biffGet(LIMN));
free(err);
return 1;
}
obj = limnObjectNew(10, AIR_TRUE);
airMopAdd(mop, obj, (airMopper)limnObjectNix, airMopAlways);
/* create limnLooks for diffuse (#0) and flat (#1) shading */
lookIdx = airArrayLenIncr(obj->lookArr, 2);
look = obj->look + lookIdx + 0;
ELL_4V_SET(look->rgba, 1, 1, 1, 1);
ELL_3V_SET(look->kads, 0, 1, 0);
look->spow = 0;
look = obj->look + lookIdx + 1;
ELL_4V_SET(look->rgba, 1, 1, 1, 1);
ELL_3V_SET(look->kads, 1, 0, 0);
look->spow = 0;
/* X axis: rod */
partIdx = limnObjectCylinderAdd(obj, 0, 0, 16);
ELL_4M_IDENTITY_SET(matA);
ELL_4M_SCALE_SET(matB, 1, 0.2, 0.2);
ell_4m_post_mul_f(matA, matB);
ELL_4M_TRANSLATE_SET(matB, 1.3, 0.0, 0.0);
ell_4m_post_mul_f(matA, matB);
limnObjectPartTransform(obj, partIdx, matA);
/* Y axis: rod + ball */
partIdx = limnObjectCylinderAdd(obj, 0, 1, 16);
ELL_4M_IDENTITY_SET(matA);
ELL_4M_SCALE_SET(matB, 0.2, 1, 0.2);
ell_4m_post_mul_f(matA, matB);
ELL_4M_TRANSLATE_SET(matB, 0.0, 1.3, 0.0);
ell_4m_post_mul_f(matA, matB);
limnObjectPartTransform(obj, partIdx, matA);
partIdx = limnObjectPolarSphereAdd(obj, 0, 0, 32, 16);
ELL_4M_IDENTITY_SET(matA);
ELL_4M_SCALE_SET(matB, 0.28, 0.28, 0.28);
ell_4m_post_mul_f(matA, matB);
ELL_4M_TRANSLATE_SET(matB, 0.0, 2.6, 0.0);
ell_4m_post_mul_f(matA, matB);
limnObjectPartTransform(obj, partIdx, matA);
/* Z axis: rod + ball + ball */
partIdx = limnObjectCylinderAdd(obj, 0, 2, 16);
ELL_4M_IDENTITY_SET(matA);
ELL_4M_SCALE_SET(matB, 0.2, 0.2, 1);
ell_4m_post_mul_f(matA, matB);
ELL_4M_TRANSLATE_SET(matB, 0.0, 0.0, 1.3);
ell_4m_post_mul_f(matA, matB);
limnObjectPartTransform(obj, partIdx, matA);
partIdx = limnObjectPolarSphereAdd(obj, 0, 1, 32, 16);
ELL_4M_IDENTITY_SET(matA);
ELL_4M_SCALE_SET(matB, 0.28, 0.28, 0.28);
ell_4m_post_mul_f(matA, matB);
ELL_4M_TRANSLATE_SET(matB, 0.0, 0.0, 2.6);
ell_4m_post_mul_f(matA, matB);
limnObjectPartTransform(obj, partIdx, matA);
partIdx = limnObjectPolarSphereAdd(obj, 0, 2, 32, 16);
ELL_4M_IDENTITY_SET(matA);
ELL_4M_SCALE_SET(matB, 0.28, 0.28, 0.28);
ell_4m_post_mul_f(matA, matB);
ELL_4M_TRANSLATE_SET(matB, 0.0, 0.0, 3.2);
ell_4m_post_mul_f(matA, matB);
limnObjectPartTransform(obj, partIdx, matA);
win = limnWindowNew(limnDevicePS);
win->scale = winscale;
win->ps.wireFrame = wire;
win->ps.lineWidth[limnEdgeTypeBackFacet] = edgeWidth[0];
win->ps.lineWidth[limnEdgeTypeBackCrease] = edgeWidth[1];
win->ps.lineWidth[limnEdgeTypeContour] = edgeWidth[2];
win->ps.lineWidth[limnEdgeTypeFrontCrease] = edgeWidth[3];
win->ps.lineWidth[limnEdgeTypeFrontFacet] = edgeWidth[4];
win->file = fopen(outS, "w");
airMopAdd(mop, win, (airMopper)limnWindowNix, airMopAlways);
if (limnObjectRender(obj, cam, win)
|| (concave
? limnObjectPSDrawConcave(obj, cam, nmap, win)
: limnObjectPSDraw(obj, cam, nmap, win))) {
airMopAdd(mop, err = biffGetDone(LIMN), airFree, airMopAlways);
fprintf(stderr, "%s: trouble:\n%s\n", me, err);
airMopError(mop);
return 1;
}
fclose(win->file);
airMopOkay(mop);
return 0;
}