OpenSubdiv/opensubdiv/osd/glslPatchLegacy.glsl

512 lines
19 KiB
Plaintext
Raw Normal View History

//
// Copyright 2013-2018 Pixar
//
// Licensed under the Apache License, Version 2.0 (the "Apache License")
// with the following modification; you may not use this file except in
// compliance with the Apache License and the following modification to it:
// Section 6. Trademarks. is deleted and replaced with:
//
// 6. Trademarks. This License does not grant permission to use the trade
// names, trademarks, service marks, or product names of the Licensor
// and its affiliates, except as required to comply with Section 4(c) of
// the License and to reproduce the content of the NOTICE file.
//
// You may obtain a copy of the Apache License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the Apache License with the above modification is
// distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
// KIND, either express or implied. See the Apache License for the specific
// language governing permissions and limitations under the Apache License.
//
// ----------------------------------------------------------------------------
// Legacy Gregory
// ----------------------------------------------------------------------------
#if defined(OSD_PATCH_GREGORY) || defined(OSD_PATCH_GREGORY_BOUNDARY)
#define M_PI 3.14159265359f
// precomputed catmark coefficient table up to valence 29
uniform float OsdCatmarkCoefficient[30] = float[](
0, 0, 0, 0.812816, 0.500000, 0.363644, 0.287514,
0.238688, 0.204544, 0.179229, 0.159657,
0.144042, 0.131276, 0.120632, 0.111614,
0.103872, 0.09715, 0.0912559, 0.0860444,
0.0814022, 0.0772401, 0.0734867, 0.0700842,
0.0669851, 0.0641504, 0.0615475, 0.0591488,
0.0569311, 0.0548745, 0.0529621
);
float
OsdComputeCatmarkCoefficient(int valence)
{
#if OSD_MAX_VALENCE < 30
return OsdCatmarkCoefficient[valence];
#else
if (valence < 30) {
return OsdCatmarkCoefficient[valence];
} else {
float t = 2.0f * float(M_PI) / float(valence);
return 1.0f / (valence * (cos(t) + 5.0f +
sqrt((cos(t) + 9) * (cos(t) + 1)))/16.0f);
}
#endif
}
float cosfn(int n, int j) {
return cos((2.0f * M_PI * j)/float(n));
}
float sinfn(int n, int j) {
return sin((2.0f * M_PI * j)/float(n));
}
#if !defined OSD_MAX_VALENCE || OSD_MAX_VALENCE < 1
#undef OSD_MAX_VALENCE
#define OSD_MAX_VALENCE 4
#endif
struct OsdPerVertexGregory {
vec3 P;
ivec3 clipFlag;
int valence;
vec3 e0;
vec3 e1;
#ifdef OSD_PATCH_GREGORY_BOUNDARY
int zerothNeighbor;
vec3 org;
#endif
vec3 r[OSD_MAX_VALENCE];
};
struct OsdPerPatchVertexGregory {
ivec3 patchParam;
vec3 P;
vec3 Ep;
vec3 Em;
vec3 Fp;
vec3 Fm;
};
#ifndef OSD_NUM_ELEMENTS
#define OSD_NUM_ELEMENTS 3
#endif
uniform samplerBuffer OsdVertexBuffer;
uniform isamplerBuffer OsdValenceBuffer;
vec3 OsdReadVertex(int vertexIndex)
{
int index = int(OSD_NUM_ELEMENTS * (vertexIndex + OsdBaseVertex()));
return vec3(texelFetch(OsdVertexBuffer, index).x,
texelFetch(OsdVertexBuffer, index+1).x,
texelFetch(OsdVertexBuffer, index+2).x);
}
int OsdReadVertexValence(int vertexID)
{
int index = int(vertexID * (2 * OSD_MAX_VALENCE + 1));
return texelFetch(OsdValenceBuffer, index).x;
}
int OsdReadVertexIndex(int vertexID, int valenceVertex)
{
int index = int(vertexID * (2 * OSD_MAX_VALENCE + 1) + 1 + valenceVertex);
return texelFetch(OsdValenceBuffer, index).x;
}
uniform isamplerBuffer OsdQuadOffsetBuffer;
int OsdReadQuadOffset(int primitiveID, int offsetVertex)
{
int index = int(4*primitiveID+OsdGregoryQuadOffsetBase() + offsetVertex);
return texelFetch(OsdQuadOffsetBuffer, index).x;
}
void
OsdComputePerVertexGregory(int vID, vec3 P, out OsdPerVertexGregory v)
{
v.clipFlag = ivec3(0);
int ivalence = OsdReadVertexValence(vID);
v.valence = ivalence;
int valence = abs(ivalence);
vec3 f[OSD_MAX_VALENCE];
vec3 pos = P;
vec3 opos = vec3(0);
#ifdef OSD_PATCH_GREGORY_BOUNDARY
v.org = pos;
int boundaryEdgeNeighbors[2];
int currNeighbor = 0;
int ibefore = 0;
int zerothNeighbor = 0;
#endif
for (int i=0; i<valence; ++i) {
int im = (i+valence-1)%valence;
int ip = (i+1)%valence;
int idx_neighbor = OsdReadVertexIndex(vID, 2*i);
#ifdef OSD_PATCH_GREGORY_BOUNDARY
bool isBoundaryNeighbor = false;
int valenceNeighbor = OsdReadVertexValence(idx_neighbor);
if (valenceNeighbor < 0) {
isBoundaryNeighbor = true;
if (currNeighbor<2) {
boundaryEdgeNeighbors[currNeighbor] = idx_neighbor;
}
currNeighbor++;
if (currNeighbor == 1) {
ibefore = i;
zerothNeighbor = i;
} else {
if (i-ibefore == 1) {
int tmp = boundaryEdgeNeighbors[0];
boundaryEdgeNeighbors[0] = boundaryEdgeNeighbors[1];
boundaryEdgeNeighbors[1] = tmp;
zerothNeighbor = i;
}
}
}
#endif
vec3 neighbor = OsdReadVertex(idx_neighbor);
int idx_diagonal = OsdReadVertexIndex(vID, 2*i + 1);
vec3 diagonal = OsdReadVertex(idx_diagonal);
int idx_neighbor_p = OsdReadVertexIndex(vID, 2*ip);
vec3 neighbor_p = OsdReadVertex(idx_neighbor_p);
int idx_neighbor_m = OsdReadVertexIndex(vID, 2*im);
vec3 neighbor_m = OsdReadVertex(idx_neighbor_m);
int idx_diagonal_m = OsdReadVertexIndex(vID, 2*im + 1);
vec3 diagonal_m = OsdReadVertex(idx_diagonal_m);
f[i] = (pos * float(valence) + (neighbor_p + neighbor)*2.0f + diagonal) / (float(valence)+5.0f);
opos += f[i];
v.r[i] = (neighbor_p-neighbor_m)/3.0f + (diagonal - diagonal_m)/6.0f;
}
opos /= valence;
v.P = vec4(opos, 1.0f).xyz;
vec3 e;
v.e0 = vec3(0);
v.e1 = vec3(0);
for(int i=0; i<valence; ++i) {
int im = (i + valence -1) % valence;
e = 0.5f * (f[i] + f[im]);
v.e0 += cosfn(valence, i)*e;
v.e1 += sinfn(valence, i)*e;
}
float ef = OsdComputeCatmarkCoefficient(valence);
v.e0 *= ef;
v.e1 *= ef;
#ifdef OSD_PATCH_GREGORY_BOUNDARY
v.zerothNeighbor = zerothNeighbor;
if (currNeighbor == 1) {
boundaryEdgeNeighbors[1] = boundaryEdgeNeighbors[0];
}
if (ivalence < 0) {
if (valence > 2) {
v.P = (OsdReadVertex(boundaryEdgeNeighbors[0]) +
OsdReadVertex(boundaryEdgeNeighbors[1]) +
4.0f * pos)/6.0f;
} else {
v.P = pos;
}
v.e0 = (OsdReadVertex(boundaryEdgeNeighbors[0]) -
OsdReadVertex(boundaryEdgeNeighbors[1]))/6.0;
float k = float(float(valence) - 1.0f); //k is the number of faces
float c = cos(M_PI/k);
float s = sin(M_PI/k);
float gamma = -(4.0f*s)/(3.0f*k+c);
float alpha_0k = -((1.0f+2.0f*c)*sqrt(1.0f+c))/((3.0f*k+c)*sqrt(1.0f-c));
float beta_0 = s/(3.0f*k + c);
int idx_diagonal = OsdReadVertexIndex(vID, 2*zerothNeighbor + 1);
vec3 diagonal = OsdReadVertex(idx_diagonal);
v.e1 = gamma * pos +
alpha_0k * OsdReadVertex(boundaryEdgeNeighbors[0]) +
alpha_0k * OsdReadVertex(boundaryEdgeNeighbors[1]) +
beta_0 * diagonal;
for (int x=1; x<valence - 1; ++x) {
int curri = ((x + zerothNeighbor)%valence);
float alpha = (4.0f*sin((M_PI * float(x))/k))/(3.0f*k+c);
float beta = (sin((M_PI * float(x))/k) + sin((M_PI * float(x+1))/k))/(3.0f*k+c);
int idx_neighbor = OsdReadVertexIndex(vID, 2*curri);
vec3 neighbor = OsdReadVertex(idx_neighbor);
idx_diagonal = OsdReadVertexIndex(vID, 2*curri + 1);
diagonal = OsdReadVertex(idx_diagonal);
v.e1 += alpha * neighbor + beta * diagonal;
}
v.e1 /= 3.0f;
}
#endif
}
void
OsdComputePerPatchVertexGregory(ivec3 patchParam, int ID, int primitiveID,
in OsdPerVertexGregory v[4],
out OsdPerPatchVertexGregory result)
{
result.patchParam = patchParam;
result.P = v[ID].P;
int i = ID;
int ip = (i+1)%4;
int im = (i+3)%4;
int valence = abs(v[i].valence);
int n = valence;
int start = OsdReadQuadOffset(primitiveID, i) & 0xff;
int prev = (OsdReadQuadOffset(primitiveID, i) >> 8) & 0xff;
int start_m = OsdReadQuadOffset(primitiveID, im) & 0xff;
int prev_p = (OsdReadQuadOffset(primitiveID, ip) >> 8) & 0xff;
int np = abs(v[ip].valence);
int nm = abs(v[im].valence);
// Control Vertices based on :
// "Approximating Subdivision Surfaces with Gregory Patches
// for Hardware Tessellation"
// Loop, Schaefer, Ni, Castano (ACM ToG Siggraph Asia 2009)
//
// P3 e3- e2+ P2
// O--------O--------O--------O
// | | | |
// | | | |
// | | f3- | f2+ |
// | O O |
// e3+ O------O O------O e2-
// | f3+ f2- |
// | |
// | |
// | f0- f1+ |
// e0- O------O O------O e1+
// | O O |
// | | f0+ | f1- |
// | | | |
// | | | |
// O--------O--------O--------O
// P0 e0+ e1- P1
//
#ifdef OSD_PATCH_GREGORY_BOUNDARY
vec3 Em_ip;
if (v[ip].valence < -2) {
int j = (np + prev_p - v[ip].zerothNeighbor) % np;
Em_ip = v[ip].P + cos((M_PI*j)/float(np-1))*v[ip].e0 + sin((M_PI*j)/float(np-1))*v[ip].e1;
} else {
Em_ip = v[ip].P + v[ip].e0*cosfn(np, prev_p ) + v[ip].e1*sinfn(np, prev_p);
}
vec3 Ep_im;
if (v[im].valence < -2) {
int j = (nm + start_m - v[im].zerothNeighbor) % nm;
Ep_im = v[im].P + cos((M_PI*j)/float(nm-1))*v[im].e0 + sin((M_PI*j)/float(nm-1))*v[im].e1;
} else {
Ep_im = v[im].P + v[im].e0*cosfn(nm, start_m) + v[im].e1*sinfn(nm, start_m);
}
if (v[i].valence < 0) {
n = (n-1)*2;
}
if (v[im].valence < 0) {
nm = (nm-1)*2;
}
if (v[ip].valence < 0) {
np = (np-1)*2;
}
if (v[i].valence > 2) {
result.Ep = v[i].P + v[i].e0*cosfn(n, start) + v[i].e1*sinfn(n, start);
result.Em = v[i].P + v[i].e0*cosfn(n, prev ) + v[i].e1*sinfn(n, prev);
float s1=3-2*cosfn(n,1)-cosfn(np,1);
float s2=2*cosfn(n,1);
result.Fp = (cosfn(np,1)*v[i].P + s1*result.Ep + s2*Em_ip + v[i].r[start])/3.0f;
s1 = 3.0f-2.0f*cos(2.0f*M_PI/float(n))-cos(2.0f*M_PI/float(nm));
result.Fm = (cosfn(nm,1)*v[i].P + s1*result.Em + s2*Ep_im - v[i].r[prev])/3.0f;
} else if (v[i].valence < -2) {
int j = (valence + start - v[i].zerothNeighbor) % valence;
result.Ep = v[i].P + cos((M_PI*j)/float(valence-1))*v[i].e0 + sin((M_PI*j)/float(valence-1))*v[i].e1;
j = (valence + prev - v[i].zerothNeighbor) % valence;
result.Em = v[i].P + cos((M_PI*j)/float(valence-1))*v[i].e0 + sin((M_PI*j)/float(valence-1))*v[i].e1;
vec3 Rp = ((-2.0f * v[i].org - 1.0f * v[im].org) + (2.0f * v[ip].org + 1.0f * v[(i+2)%4].org))/3.0f;
vec3 Rm = ((-2.0f * v[i].org - 1.0f * v[ip].org) + (2.0f * v[im].org + 1.0f * v[(i+2)%4].org))/3.0f;
float s1 = 3-2*cosfn(n,1)-cosfn(np,1);
float s2 = 2*cosfn(n,1);
result.Fp = (cosfn(np,1)*v[i].P + s1*result.Ep + s2*Em_ip + v[i].r[start])/3.0f;
s1 = 3.0f-2.0f*cos(2.0f*M_PI/float(n))-cos(2.0f*M_PI/float(nm));
result.Fm = (cosfn(nm,1)*v[i].P + s1*result.Em + s2*Ep_im - v[i].r[prev])/3.0f;
if (v[im].valence < 0) {
s1 = 3-2*cosfn(n,1)-cosfn(np,1);
result.Fp = result.Fm = (cosfn(np,1)*v[i].P + s1*result.Ep + s2*Em_ip + v[i].r[start])/3.0f;
} else if (v[ip].valence < 0) {
s1 = 3.0f-2.0f*cos(2.0f*M_PI/n)-cos(2.0f*M_PI/nm);
result.Fm = result.Fp = (cosfn(nm,1)*v[i].P + s1*result.Em + s2*Ep_im - v[i].r[prev])/3.0f;
}
} else if (v[i].valence == -2) {
result.Ep = (2.0f * v[i].org + v[ip].org)/3.0f;
result.Em = (2.0f * v[i].org + v[im].org)/3.0f;
result.Fp = result.Fm = (4.0f * v[i].org + v[(i+2)%n].org + 2.0f * v[ip].org + 2.0f * v[im].org)/9.0f;
}
#else // not OSD_PATCH_GREGORY_BOUNDARY
result.Ep = v[i].P + v[i].e0 * cosfn(n, start) + v[i].e1*sinfn(n, start);
result.Em = v[i].P + v[i].e0 * cosfn(n, prev ) + v[i].e1*sinfn(n, prev);
vec3 Em_ip = v[ip].P + v[ip].e0 * cosfn(np, prev_p ) + v[ip].e1*sinfn(np, prev_p);
vec3 Ep_im = v[im].P + v[im].e0 * cosfn(nm, start_m) + v[im].e1*sinfn(nm, start_m);
float s1 = 3-2*cosfn(n,1)-cosfn(np,1);
float s2 = 2*cosfn(n,1);
result.Fp = (cosfn(np,1)*v[i].P + s1*result.Ep + s2*Em_ip + v[i].r[start])/3.0f;
s1 = 3.0f-2.0f*cos(2.0f*M_PI/float(n))-cos(2.0f*M_PI/float(nm));
result.Fm = (cosfn(nm,1)*v[i].P + s1*result.Em + s2*Ep_im - v[i].r[prev])/3.0f;
#endif
}
#endif // OSD_PATCH_GREGORY || OSD_PATCH_GREGORY_BOUNDARY
// ----------------------------------------------------------------------------
// Legacy Face-varying
// ----------------------------------------------------------------------------
uniform samplerBuffer OsdFVarDataBuffer;
#ifndef OSD_FVAR_WIDTH
#define OSD_FVAR_WIDTH 0
#endif
// ------ extract from quads (catmark, bilinear) ---------
// XXX: only linear interpolation is supported
#define OSD_COMPUTE_FACE_VARYING_1(result, fvarOffset, tessCoord) \
{ \
float v[4]; \
int primOffset = OsdGetPatchIndex(gl_PrimitiveID) * 4; \
for (int i = 0; i < 4; ++i) { \
int index = (primOffset+i)*OSD_FVAR_WIDTH + fvarOffset; \
v[i] = texelFetch(OsdFVarDataBuffer, index).s \
} \
result = mix(mix(v[0], v[1], tessCoord.s), \
mix(v[3], v[2], tessCoord.s), \
tessCoord.t); \
}
#define OSD_COMPUTE_FACE_VARYING_2(result, fvarOffset, tessCoord) \
{ \
vec2 v[4]; \
int primOffset = OsdGetPatchIndex(gl_PrimitiveID) * 4; \
for (int i = 0; i < 4; ++i) { \
int index = (primOffset+i)*OSD_FVAR_WIDTH + fvarOffset; \
v[i] = vec2(texelFetch(OsdFVarDataBuffer, index).s, \
texelFetch(OsdFVarDataBuffer, index + 1).s); \
} \
result = mix(mix(v[0], v[1], tessCoord.s), \
mix(v[3], v[2], tessCoord.s), \
tessCoord.t); \
}
#define OSD_COMPUTE_FACE_VARYING_3(result, fvarOffset, tessCoord) \
{ \
vec3 v[4]; \
int primOffset = OsdGetPatchIndex(gl_PrimitiveID) * 4; \
for (int i = 0; i < 4; ++i) { \
int index = (primOffset+i)*OSD_FVAR_WIDTH + fvarOffset; \
v[i] = vec3(texelFetch(OsdFVarDataBuffer, index).s, \
texelFetch(OsdFVarDataBuffer, index + 1).s, \
texelFetch(OsdFVarDataBuffer, index + 2).s); \
} \
result = mix(mix(v[0], v[1], tessCoord.s), \
mix(v[3], v[2], tessCoord.s), \
tessCoord.t); \
}
#define OSD_COMPUTE_FACE_VARYING_4(result, fvarOffset, tessCoord) \
{ \
vec4 v[4]; \
int primOffset = OsdGetPatchIndex(gl_PrimitiveID) * 4; \
for (int i = 0; i < 4; ++i) { \
int index = (primOffset+i)*OSD_FVAR_WIDTH + fvarOffset; \
v[i] = vec4(texelFetch(OsdFVarDataBuffer, index).s, \
texelFetch(OsdFVarDataBuffer, index + 1).s, \
texelFetch(OsdFVarDataBuffer, index + 2).s, \
texelFetch(OsdFVarDataBuffer, index + 3).s); \
} \
result = mix(mix(v[0], v[1], tessCoord.s), \
mix(v[3], v[2], tessCoord.s), \
tessCoord.t); \
}
// ------ extract from triangles barycentric (loop) ---------
// XXX: no interpolation supported
#define OSD_COMPUTE_FACE_VARYING_TRI_1(result, fvarOffset, triVert) \
{ \
int primOffset = OsdGetPatchIndex(gl_PrimitiveID) * 3; \
int index = (primOffset+triVert)*OSD_FVAR_WIDTH + fvarOffset; \
result = texelFetch(OsdFVarDataBuffer, index).s; \
}
#define OSD_COMPUTE_FACE_VARYING_TRI_2(result, fvarOffset, triVert) \
{ \
int primOffset = OsdGetPatchIndex(gl_PrimitiveID) * 3; \
int index = (primOffset+triVert)*OSD_FVAR_WIDTH + fvarOffset; \
result = vec2(texelFetch(OsdFVarDataBuffer, index).s, \
texelFetch(OsdFVarDataBuffer, index + 1).s); \
}
#define OSD_COMPUTE_FACE_VARYING_TRI_3(result, fvarOffset, triVert) \
{ \
int primOffset = OsdGetPatchIndex(gl_PrimitiveID) * 3; \
int index = (primOffset+triVert)*OSD_FVAR_WIDTH + fvarOffset; \
result = vec3(texelFetch(OsdFVarDataBuffer, index).s, \
texelFetch(OsdFVarDataBuffer, index + 1).s, \
texelFetch(OsdFVarDataBuffer, index + 2).s); \
}
#define OSD_COMPUTE_FACE_VARYING_TRI_4(result, fvarOffset, triVert) \
{ \
int primOffset = OsdGetPatchIndex(gl_PrimitiveID) * 3; \
int index = (primOffset+triVert)*OSD_FVAR_WIDTH + fvarOffset; \
result = vec4(texelFetch(OsdFVarDataBuffer, index).s, \
texelFetch(OsdFVarDataBuffer, index + 1).s, \
texelFetch(OsdFVarDataBuffer, index + 2).s, \
texelFetch(OsdFVarDataBuffer, index + 3).s); \
}