2011-04-27 11:58:34 +00:00
|
|
|
/*
|
|
|
|
* Copyright 2011, Blender Foundation.
|
|
|
|
*
|
|
|
|
* This program is free software; you can redistribute it and/or
|
|
|
|
* modify it under the terms of the GNU General Public License
|
|
|
|
* as published by the Free Software Foundation; either version 2
|
|
|
|
* of the License, or (at your option) any later version.
|
|
|
|
*
|
|
|
|
* This program is distributed in the hope that it will be useful,
|
|
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
|
|
* GNU General Public License for more details.
|
|
|
|
*
|
|
|
|
* You should have received a copy of the GNU General Public License
|
|
|
|
* along with this program; if not, write to the Free Software Foundation,
|
|
|
|
* Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
|
|
|
|
*/
|
|
|
|
|
|
|
|
/* Parts adapted from code in the public domain in NVidia Mesh Tools. */
|
|
|
|
|
|
|
|
#include "mesh.h"
|
|
|
|
|
|
|
|
#include "subd_patch.h"
|
|
|
|
|
|
|
|
#include "util_math.h"
|
|
|
|
#include "util_types.h"
|
|
|
|
|
|
|
|
CCL_NAMESPACE_BEGIN
|
|
|
|
|
|
|
|
/* De Casteljau Evaluation */
|
|
|
|
|
|
|
|
static float3 decasteljau_quadratic(float t, const float3 cp[3])
|
|
|
|
{
|
|
|
|
float3 d0 = cp[0] + t*(cp[1] - cp[0]);
|
|
|
|
float3 d1 = cp[1] + t*(cp[2] - cp[1]);
|
|
|
|
|
|
|
|
return d0 + t*(d1 - d0);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void decasteljau_cubic(float3 *P, float3 *dt, float t, const float3 cp[4])
|
|
|
|
{
|
|
|
|
float3 d0 = cp[0] + t*(cp[1] - cp[0]);
|
|
|
|
float3 d1 = cp[1] + t*(cp[2] - cp[1]);
|
|
|
|
float3 d2 = cp[2] + t*(cp[3] - cp[2]);
|
|
|
|
|
|
|
|
d0 += t*(d1 - d0);
|
|
|
|
d1 += t*(d2 - d1);
|
|
|
|
|
|
|
|
*P = d0 + t*(d1 - d0);
|
|
|
|
if(dt) *dt = d1 - d0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void decasteljau_bicubic(float3 *P, float3 *du, float3 *dv, const float3 cp[16], float u, float v)
|
|
|
|
{
|
|
|
|
float3 ucp[4], utn[4];
|
|
|
|
|
|
|
|
/* interpolate over u */
|
|
|
|
decasteljau_cubic(ucp+0, utn+0, u, cp);
|
|
|
|
decasteljau_cubic(ucp+1, utn+1, u, cp+4);
|
|
|
|
decasteljau_cubic(ucp+2, utn+2, u, cp+8);
|
|
|
|
decasteljau_cubic(ucp+3, utn+3, u, cp+12);
|
|
|
|
|
|
|
|
/* interpolate over v */
|
|
|
|
decasteljau_cubic(P, dv, v, ucp);
|
|
|
|
if(du) decasteljau_cubic(du, NULL, v, utn);
|
|
|
|
}
|
|
|
|
|
|
|
|
static float3 decasteljau_tangent(const float3 cp[12], float u, float v)
|
|
|
|
{
|
|
|
|
float3 ucp[3];
|
|
|
|
|
|
|
|
decasteljau_cubic(ucp+0, NULL, v, cp);
|
|
|
|
decasteljau_cubic(ucp+1, NULL, v, cp+4);
|
|
|
|
decasteljau_cubic(ucp+2, NULL, v, cp+8);
|
|
|
|
|
|
|
|
return decasteljau_quadratic(u, ucp);
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Linear Quad Patch */
|
|
|
|
|
|
|
|
void LinearQuadPatch::eval(float3 *P, float3 *dPdu, float3 *dPdv, float u, float v)
|
|
|
|
{
|
|
|
|
float3 d0 = interp(hull[0], hull[1], u);
|
|
|
|
float3 d1 = interp(hull[2], hull[3], u);
|
|
|
|
|
|
|
|
*P = interp(d0, d1, v);
|
|
|
|
|
|
|
|
if(dPdu && dPdv) {
|
|
|
|
*dPdu = interp(hull[1] - hull[0], hull[3] - hull[2], v);
|
|
|
|
*dPdv = interp(hull[2] - hull[0], hull[3] - hull[1], u);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
BoundBox LinearQuadPatch::bound()
|
|
|
|
{
|
Cycles: BVH build time optimizations:
* BVH building was multithreaded. Not all building is multithreaded, packing
and the initial bounding/splitting is still single threaded, but recursive
splitting is, which was the main bottleneck.
* Object splitting now uses binning rather than sorting of all elements, using
code from the Embree raytracer from Intel.
http://software.intel.com/en-us/articles/embree-photo-realistic-ray-tracing-kernels/
* Other small changes to avoid allocations, pack memory more tightly, avoid
some unnecessary operations, ...
These optimizations do not work yet when Spatial Splits are enabled, for that
more work is needed. There's also other optimizations still needed, in
particular for the case of many low poly objects, the packing step and node
memory allocation.
BVH raytracing time should remain about the same, but BVH build time should be
significantly reduced, test here show speedup of about 5x to 10x on a dual core
and 5x to 25x on an 8-core machine, depending on the scene.
2012-04-24 11:52:58 +00:00
|
|
|
BoundBox bbox = BoundBox::empty;
|
2011-04-27 11:58:34 +00:00
|
|
|
|
|
|
|
for(int i = 0; i < 4; i++)
|
|
|
|
bbox.grow(hull[i]);
|
|
|
|
|
|
|
|
return bbox;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Linear Triangle Patch */
|
|
|
|
|
|
|
|
void LinearTrianglePatch::eval(float3 *P, float3 *dPdu, float3 *dPdv, float u, float v)
|
|
|
|
{
|
|
|
|
*P = u*hull[0] + v*hull[1] + (1.0f - u - v)*hull[2];
|
|
|
|
|
|
|
|
if(dPdu && dPdv) {
|
|
|
|
*dPdu = hull[0] - hull[2];
|
|
|
|
*dPdv = hull[1] - hull[2];
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
BoundBox LinearTrianglePatch::bound()
|
|
|
|
{
|
Cycles: BVH build time optimizations:
* BVH building was multithreaded. Not all building is multithreaded, packing
and the initial bounding/splitting is still single threaded, but recursive
splitting is, which was the main bottleneck.
* Object splitting now uses binning rather than sorting of all elements, using
code from the Embree raytracer from Intel.
http://software.intel.com/en-us/articles/embree-photo-realistic-ray-tracing-kernels/
* Other small changes to avoid allocations, pack memory more tightly, avoid
some unnecessary operations, ...
These optimizations do not work yet when Spatial Splits are enabled, for that
more work is needed. There's also other optimizations still needed, in
particular for the case of many low poly objects, the packing step and node
memory allocation.
BVH raytracing time should remain about the same, but BVH build time should be
significantly reduced, test here show speedup of about 5x to 10x on a dual core
and 5x to 25x on an 8-core machine, depending on the scene.
2012-04-24 11:52:58 +00:00
|
|
|
BoundBox bbox = BoundBox::empty;
|
2011-04-27 11:58:34 +00:00
|
|
|
|
|
|
|
for(int i = 0; i < 3; i++)
|
|
|
|
bbox.grow(hull[i]);
|
|
|
|
|
|
|
|
return bbox;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Bicubic Patch */
|
|
|
|
|
|
|
|
void BicubicPatch::eval(float3 *P, float3 *dPdu, float3 *dPdv, float u, float v)
|
|
|
|
{
|
|
|
|
decasteljau_bicubic(P, dPdu, dPdv, hull, u, v);
|
|
|
|
}
|
|
|
|
|
|
|
|
BoundBox BicubicPatch::bound()
|
|
|
|
{
|
Cycles: BVH build time optimizations:
* BVH building was multithreaded. Not all building is multithreaded, packing
and the initial bounding/splitting is still single threaded, but recursive
splitting is, which was the main bottleneck.
* Object splitting now uses binning rather than sorting of all elements, using
code from the Embree raytracer from Intel.
http://software.intel.com/en-us/articles/embree-photo-realistic-ray-tracing-kernels/
* Other small changes to avoid allocations, pack memory more tightly, avoid
some unnecessary operations, ...
These optimizations do not work yet when Spatial Splits are enabled, for that
more work is needed. There's also other optimizations still needed, in
particular for the case of many low poly objects, the packing step and node
memory allocation.
BVH raytracing time should remain about the same, but BVH build time should be
significantly reduced, test here show speedup of about 5x to 10x on a dual core
and 5x to 25x on an 8-core machine, depending on the scene.
2012-04-24 11:52:58 +00:00
|
|
|
BoundBox bbox = BoundBox::empty;
|
2011-04-27 11:58:34 +00:00
|
|
|
|
|
|
|
for(int i = 0; i < 16; i++)
|
|
|
|
bbox.grow(hull[i]);
|
|
|
|
|
|
|
|
return bbox;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Bicubic Patch with Tangent Fields */
|
|
|
|
|
|
|
|
void BicubicTangentPatch::eval(float3 *P, float3 *dPdu, float3 *dPdv, float u, float v)
|
|
|
|
{
|
|
|
|
decasteljau_bicubic(P, NULL, NULL, hull, u, v);
|
|
|
|
|
|
|
|
if(dPdu) *dPdu = decasteljau_tangent(utan, u, v);
|
|
|
|
if(dPdv) *dPdv = decasteljau_tangent(vtan, v, u);
|
|
|
|
}
|
|
|
|
|
|
|
|
BoundBox BicubicTangentPatch::bound()
|
|
|
|
{
|
Cycles: BVH build time optimizations:
* BVH building was multithreaded. Not all building is multithreaded, packing
and the initial bounding/splitting is still single threaded, but recursive
splitting is, which was the main bottleneck.
* Object splitting now uses binning rather than sorting of all elements, using
code from the Embree raytracer from Intel.
http://software.intel.com/en-us/articles/embree-photo-realistic-ray-tracing-kernels/
* Other small changes to avoid allocations, pack memory more tightly, avoid
some unnecessary operations, ...
These optimizations do not work yet when Spatial Splits are enabled, for that
more work is needed. There's also other optimizations still needed, in
particular for the case of many low poly objects, the packing step and node
memory allocation.
BVH raytracing time should remain about the same, but BVH build time should be
significantly reduced, test here show speedup of about 5x to 10x on a dual core
and 5x to 25x on an 8-core machine, depending on the scene.
2012-04-24 11:52:58 +00:00
|
|
|
BoundBox bbox = BoundBox::empty;
|
2011-04-27 11:58:34 +00:00
|
|
|
|
|
|
|
for(int i = 0; i < 16; i++)
|
|
|
|
bbox.grow(hull[i]);
|
|
|
|
|
|
|
|
return bbox;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Gregory Patch */
|
|
|
|
|
|
|
|
static float no_zero_div(float f)
|
|
|
|
{
|
|
|
|
if(f == 0.0f) return 1.0f;
|
|
|
|
return f;
|
|
|
|
}
|
|
|
|
|
|
|
|
void GregoryQuadPatch::eval(float3 *P, float3 *dPdu, float3 *dPdv, float u, float v)
|
|
|
|
{
|
|
|
|
float3 bicubic[16];
|
|
|
|
|
|
|
|
float U = 1 - u;
|
|
|
|
float V = 1 - v;
|
|
|
|
|
|
|
|
/* 8 9 10 11
|
|
|
|
* 12 0\1 2/3 13
|
|
|
|
* 14 4/5 6\7 15
|
|
|
|
* 16 17 18 19
|
|
|
|
*/
|
|
|
|
|
|
|
|
bicubic[5] = (u*hull[1] + v*hull[0])/no_zero_div(u + v);
|
|
|
|
bicubic[6] = (U*hull[2] + v*hull[3])/no_zero_div(U + v);
|
|
|
|
bicubic[9] = (u*hull[5] + V*hull[4])/no_zero_div(u + V);
|
|
|
|
bicubic[10] = (U*hull[6] + V*hull[7])/no_zero_div(U + V);
|
|
|
|
|
|
|
|
// Map gregory control points to bezier control points.
|
|
|
|
bicubic[0] = hull[8];
|
|
|
|
bicubic[1] = hull[9];
|
|
|
|
bicubic[2] = hull[10];
|
|
|
|
bicubic[3] = hull[11];
|
|
|
|
bicubic[4] = hull[12];
|
|
|
|
bicubic[7] = hull[13];
|
|
|
|
bicubic[8] = hull[14];
|
|
|
|
bicubic[11] = hull[15];
|
|
|
|
bicubic[12] = hull[16];
|
|
|
|
bicubic[13] = hull[17];
|
|
|
|
bicubic[14] = hull[18];
|
|
|
|
bicubic[15] = hull[19];
|
|
|
|
|
|
|
|
decasteljau_bicubic(P, dPdu, dPdv, bicubic, u, v);
|
|
|
|
}
|
|
|
|
|
|
|
|
BoundBox GregoryQuadPatch::bound()
|
|
|
|
{
|
Cycles: BVH build time optimizations:
* BVH building was multithreaded. Not all building is multithreaded, packing
and the initial bounding/splitting is still single threaded, but recursive
splitting is, which was the main bottleneck.
* Object splitting now uses binning rather than sorting of all elements, using
code from the Embree raytracer from Intel.
http://software.intel.com/en-us/articles/embree-photo-realistic-ray-tracing-kernels/
* Other small changes to avoid allocations, pack memory more tightly, avoid
some unnecessary operations, ...
These optimizations do not work yet when Spatial Splits are enabled, for that
more work is needed. There's also other optimizations still needed, in
particular for the case of many low poly objects, the packing step and node
memory allocation.
BVH raytracing time should remain about the same, but BVH build time should be
significantly reduced, test here show speedup of about 5x to 10x on a dual core
and 5x to 25x on an 8-core machine, depending on the scene.
2012-04-24 11:52:58 +00:00
|
|
|
BoundBox bbox = BoundBox::empty;
|
2011-04-27 11:58:34 +00:00
|
|
|
|
|
|
|
for(int i = 0; i < 20; i++)
|
|
|
|
bbox.grow(hull[i]);
|
|
|
|
|
|
|
|
return bbox;
|
|
|
|
}
|
|
|
|
|
|
|
|
void GregoryTrianglePatch::eval(float3 *P, float3 *dPdu, float3 *dPdv, float u, float v)
|
|
|
|
{
|
|
|
|
/* 6
|
|
|
|
*
|
|
|
|
* 14 0/1 7
|
|
|
|
*
|
|
|
|
* 13 5/4 3\2 8
|
|
|
|
*
|
|
|
|
* 12 11 10 9
|
|
|
|
*/
|
|
|
|
|
|
|
|
float w = 1 - u - v;
|
|
|
|
float uu = u * u;
|
|
|
|
float vv = v * v;
|
|
|
|
float ww = w * w;
|
|
|
|
float uuu = uu * u;
|
|
|
|
float vvv = vv * v;
|
|
|
|
float www = ww * w;
|
|
|
|
|
|
|
|
float U = 1 - u;
|
|
|
|
float V = 1 - v;
|
|
|
|
float W = 1 - w;
|
|
|
|
|
|
|
|
float3 C0 = ( v*U * hull[5] + u*V * hull[4] ) / no_zero_div(v*U + u*V);
|
|
|
|
float3 C1 = ( w*V * hull[3] + v*W * hull[2] ) / no_zero_div(w*V + v*W);
|
|
|
|
float3 C2 = ( u*W * hull[1] + w*U * hull[0] ) / no_zero_div(u*W + w*U);
|
|
|
|
|
|
|
|
*P =
|
|
|
|
(hull[12] * www + 3*hull[11] * ww*u + 3*hull[10] * w*uu + hull[ 9]*uuu) * (w + u) +
|
|
|
|
(hull[ 9] * uuu + 3*hull[ 8] * uu*v + 3*hull[ 7] * u*vv + hull[ 6]*vvv) * (u + v) +
|
|
|
|
(hull[ 6] * vvv + 3*hull[14] * vv*w + 3*hull[13] * v*ww + hull[12]*www) * (v + w) -
|
|
|
|
(hull[12] * www*w + hull[ 9] * uuu*u + hull[ 6] * vvv*v) +
|
|
|
|
12*(C0 * u*v*ww + C1 * uu*v*w + C2 * u*vv*w);
|
|
|
|
|
|
|
|
if(dPdu || dPdv) {
|
|
|
|
float3 E1 = (hull[12]*www + 3*hull[11]*ww*u + 3*hull[10]*w*uu + hull[ 9]*uuu);
|
|
|
|
float3 E2 = (hull[ 9]*uuu + 3*hull[ 8]*uu*v + 3*hull[ 7]*u*vv + hull[ 6]*vvv);
|
|
|
|
float3 E3 = (hull[ 6]*vvv + 3*hull[14]*vv*w + 3*hull[13]*v*ww + hull[12]*www);
|
|
|
|
|
|
|
|
if(dPdu) {
|
|
|
|
float3 E1u = 3*( - hull[12]*ww + hull[11]*(ww-2*u*w) + hull[10]*(2*u*w-uu) + hull[ 9]*uu);
|
|
|
|
float3 E2u = 3*( hull[ 9]*uu + 2*hull[ 8]*u*v + hull[ 7]*vv );
|
|
|
|
float3 E3u = 3*( - hull[14]*vv - 2*hull[13]*v*w - hull[12]*ww);
|
|
|
|
float3 Su = 4*( -hull[12]*www + hull[9]*uuu);
|
|
|
|
float3 Cu = 12*( C0*(ww*v-2*u*v*w) + C1*(2*u*v*w-uu*v) + C2*vv*(w-u) );
|
|
|
|
|
|
|
|
*dPdu = E1u*(w+u) + (E2+E2u*(u+v)) + (E3u*(v+w)-E3) - Su + Cu;
|
|
|
|
}
|
|
|
|
|
|
|
|
if(dPdv) {
|
|
|
|
float3 E1v = 3*(-hull[12]*ww - 2*hull[11]*w*u - hull[10]*uu );
|
|
|
|
float3 E2v = 3*( hull[ 8]*uu + 2*hull[ 7]*u*v + hull[ 6]*vv);
|
|
|
|
float3 E3v = 3*( hull[ 6]*vv + hull[14]*(2*w*v-vv) + hull[13]*(ww-2*w*v) - hull[12]*ww);
|
|
|
|
float3 Sv = 4*(-hull[12]*www + hull[ 6]*vvv);
|
|
|
|
float3 Cv = 12*(C0*(u*ww-2*u*v*w) + C1*uu*(w-v) + C2*(2*u*v*w-u*vv));
|
|
|
|
|
|
|
|
*dPdv = ((E1v*(w+u)-E1) + (E2+E2v*(u+v)) + E3v*(v+w) - Sv + Cv );
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
BoundBox GregoryTrianglePatch::bound()
|
|
|
|
{
|
Cycles: BVH build time optimizations:
* BVH building was multithreaded. Not all building is multithreaded, packing
and the initial bounding/splitting is still single threaded, but recursive
splitting is, which was the main bottleneck.
* Object splitting now uses binning rather than sorting of all elements, using
code from the Embree raytracer from Intel.
http://software.intel.com/en-us/articles/embree-photo-realistic-ray-tracing-kernels/
* Other small changes to avoid allocations, pack memory more tightly, avoid
some unnecessary operations, ...
These optimizations do not work yet when Spatial Splits are enabled, for that
more work is needed. There's also other optimizations still needed, in
particular for the case of many low poly objects, the packing step and node
memory allocation.
BVH raytracing time should remain about the same, but BVH build time should be
significantly reduced, test here show speedup of about 5x to 10x on a dual core
and 5x to 25x on an 8-core machine, depending on the scene.
2012-04-24 11:52:58 +00:00
|
|
|
BoundBox bbox = BoundBox::empty;
|
2011-04-27 11:58:34 +00:00
|
|
|
|
|
|
|
for(int i = 0; i < 20; i++)
|
|
|
|
bbox.grow(hull[i]);
|
|
|
|
|
|
|
|
return bbox;
|
|
|
|
}
|
|
|
|
|
|
|
|
CCL_NAMESPACE_END
|
|
|
|
|