OpenSubdiv/examples/glStencilViewer/glStencilViewer.cpp

1110 lines
32 KiB
C++
Raw Normal View History

//
// Copyright 2013 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.
//
#if defined(__APPLE__)
#if defined(OSD_USES_GLEW)
#include <GL/glew.h>
#else
#include <OpenGL/gl3.h>
#endif
#define GLFW_INCLUDE_GL3
#define GLFW_NO_GLU
#else
#include <stdlib.h>
#include <GL/glew.h>
#if defined(WIN32)
#include <GL/wglew.h>
#endif
#endif
#include <GLFW/glfw3.h>
GLFWwindow* g_window=0;
GLFWmonitor* g_primary=0;
#include <common/vtr_utils.h>
#include "../common/stopwatch.h"
#include "../common/simple_math.h"
#include "../common/gl_common.h"
#include "../common/gl_hud.h"
#include <far/patchTablesFactory.h>
#include <far/ptexIndices.h>
#include <far/stencilTablesFactory.h>
#include <osd/cpuGLVertexBuffer.h>
#include <osd/cpuVertexBuffer.h>
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
#include <osd/cpuEvaluator.h>
#include <cfloat>
#include <list>
#include <vector>
#include <iostream>
#include <fstream>
#include <sstream>
#include <stdlib.h>
#ifdef OPENSUBDIV_HAS_OPENMP
#include <omp.h>
#endif
using namespace OpenSubdiv;
enum KernelType { kCPU = 0,
kOPENMP = 1,
kTBB = 2 };
int g_kernel = kCPU,
g_isolationLevel = 5; // max level of extraordinary feature isolation
int g_running = 1,
g_width = 1024,
g_height = 1024,
g_fullscreen = 0,
g_drawCageEdges = 1,
g_drawCageVertices = 1,
g_prev_x = 0,
g_prev_y = 0,
g_mbutton[3] = {0, 0, 0},
g_frame=0,
g_freeze=0,
g_repeatCount;
bool g_bilinear=false;
float g_rotate[2] = {0, 0},
g_dolly = 5,
g_pan[2] = {0, 0},
g_center[3] = {0, 0, 0},
g_size = 0,
g_moveScale = 0.0f;
struct Transform {
float ModelViewMatrix[16];
float ProjectionMatrix[16];
float ModelViewProjectionMatrix[16];
} g_transformData;
// performance
float g_evalTime = 0;
Stopwatch g_fpsTimer;
std::vector<float> g_orgPositions;
std::vector<int> g_coarseEdges;
std::vector<float> g_coarseEdgeSharpness;
std::vector<float> g_coarseVertexSharpness;
int g_nsamples=2000,
g_nsamplesDrawn=0;
GLuint g_cageEdgeVAO = 0,
g_cageEdgeVBO = 0,
g_cageVertexVAO = 0,
g_cageVertexVBO = 0,
g_stencilsVAO = 0;
GLhud g_hud;
//------------------------------------------------------------------------------
#include "init_shapes.h"
int g_currentShape = 0;
//------------------------------------------------------------------------------
Far::LimitStencilTables const * g_controlStencils;
// Control vertex positions (P(xyz))
Osd::CpuVertexBuffer * g_controlValues=0;
// Display VBO (collects outputs of updated stencils)
Osd::CpuGLVertexBuffer * g_stencilValues=0;
// Display 3 lines for each stencil sample (utan, vtan, normal)
// 18 elements : [ P (xyz), P+dPdu (xyz),
// P (xyz), P+dPdv (xyz),
// P (xyz), P+N (xyz) ]
Osd::VertexBufferDescriptor g_controlDesc( /*offset*/ 0, /*legnth*/ 3, /*stride*/ 3 ),
g_outputDataDesc( /*offset*/ 0, /*legnth*/ 3, /*stride*/ 18 ),
g_outputDuDesc( /*offset*/ 3, /*legnth*/ 3, /*stride*/ 18 ),
g_outputDvDesc( /*offset*/ 9, /*legnth*/ 3, /*stride*/ 18 );
#if defined(OPENSUBDIV_HAS_OPENMP)
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
#include <osd/ompEvaluator.h>
#endif
#ifdef OPENSUBDIV_HAS_TBB
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
#include <osd/tbbEvaluator.h>
#endif
//------------------------------------------------------------------------------
#define SCALE_TAN 0.02f
#define SCALE_NORM 0.02f
static void
updateGeom() {
int nverts = (int)g_orgPositions.size() / 3;
const float *p = &g_orgPositions[0];
float r = sin(g_frame*0.001f) * g_moveScale;
float * positions = g_controlValues->BindCpuBuffer();
for (int i = 0; i < nverts; ++i) {
//float move = 0.05f*cosf(p[0]*20+g_frame*0.01f);
float ct = cos(p[2] * r);
float st = sin(p[2] * r);
positions[i*3+0] = p[0]*ct + p[1]*st;
positions[i*3+1] = -p[0]*st + p[1]*ct;
positions[i*3+2] = p[2];
p+=3;
}
Stopwatch s;
s.Start();
float * ptr = g_stencilValues->BindCpuBuffer();
memset(ptr, 0, g_controlStencils->GetNumStencils() * 18 * sizeof(float));
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
// Update random points by applying point & tangent stencils
switch (g_kernel) {
case kCPU: {
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
Osd::CpuEvaluator::EvalStencils(
g_controlValues, g_controlDesc, // input
g_stencilValues, g_outputDataDesc, // position
g_stencilValues, g_outputDuDesc, // Du
g_stencilValues, g_outputDvDesc, // Dv
// Normals will be filled afterwards
g_controlStencils);
} break;
#if defined(OPENSUBDIV_HAS_OPENMP)
case kOPENMP: {
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
// FIXME: implements OmpEvaluator
Osd::CpuEvaluator::EvalStencils(
g_controlValues, g_controlDesc, // input
g_stencilValues, g_outputDataDesc, // position
g_stencilValues, g_outputDuDesc, // Du
g_stencilValues, g_outputDvDesc, // Dv
// Normals will be filled afterwards
g_controlStencils);
} break;
#endif
#if defined(OPENSUBDIV_HAS_TBB)
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
// FIXME: implements TbbEvaluator
case kTBB: {
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
Osd::CpuEvaluator::EvalStencils(
g_controlValues, g_controlDesc, // input
g_stencilValues, g_outputDataDesc, // position
g_stencilValues, g_outputDuDesc, // Du
g_stencilValues, g_outputDvDesc, // Dv
// Normals will be filled afterwards
g_controlStencils);
} break;
#endif
default:
return;
}
s.Stop();
g_evalTime = float(s.GetElapsed() * 1000.0f);
assert(g_controlStencils);
for (int i=0; i < g_controlStencils->GetNumStencils(); ++i, ptr+=18) {
float * p = ptr,
* utan = ptr + 3,
* vtan = ptr + 9,
* normal = ptr + 15;
// copy P as starting point for each line
memcpy( ptr + 6, p, 3*sizeof(float) );
memcpy( ptr + 12, p, 3*sizeof(float) );
normalize( utan );
normalize( vtan );
cross( normal, utan, vtan );
normalize(normal);
// compute end point for each line (P + vec * scale)
for (int j=0; j<3; ++j) {
utan[j]= p[j] + utan[j]*SCALE_TAN;
vtan[j]= p[j] + vtan[j]*SCALE_TAN;
normal[j]= p[j] + normal[j]*SCALE_NORM;
}
}
}
//------------------------------------------------------------------------------
static void
createMesh(ShapeDesc const & shapeDesc, int level) {
typedef Far::ConstIndexArray IndexArray;
typedef Far::LimitStencilTablesFactory::LocationArray LocationArray;
Shape const * shape = Shape::parseObj(shapeDesc.data.c_str(), shapeDesc.scheme);
// create Vtr mesh (topology)
OpenSubdiv::Sdc::SchemeType sdctype = GetSdcType(*shape);
OpenSubdiv::Sdc::Options sdcoptions = GetSdcOptions(*shape);
OpenSubdiv::Far::TopologyRefiner * refiner =
OpenSubdiv::Far::TopologyRefinerFactory<Shape>::Create(*shape,
OpenSubdiv::Far::TopologyRefinerFactory<Shape>::Options(sdctype, sdcoptions));
// save coarse topology (used for coarse mesh drawing)
int nedges = refiner->GetNumEdges(0),
nverts = refiner->GetNumVertices(0);
g_coarseEdges.resize(nedges*2);
g_coarseEdgeSharpness.resize(nedges);
g_coarseVertexSharpness.resize(nverts);
for(int i=0; i<nedges; ++i) {
IndexArray verts = refiner->GetEdgeVertices(0, i);
g_coarseEdges[i*2 ]=verts[0];
g_coarseEdges[i*2+1]=verts[1];
g_coarseEdgeSharpness[i]=refiner->GetEdgeSharpness(0, i);
}
for(int i=0; i<nverts; ++i) {
g_coarseVertexSharpness[i]=refiner->GetVertexSharpness(0, i);
}
g_orgPositions=shape->verts;
if (g_bilinear) {
Far::TopologyRefiner::UniformOptions options(level);
options.fullTopologyInLastLevel = true;
refiner->RefineUniform(options);
} else {
Far::TopologyRefiner::AdaptiveOptions options(level);
options.useSingleCreasePatch = false;
refiner->RefineAdaptive(options);
}
Far::PtexIndices ptexIndices(*refiner);
int nfaces = ptexIndices.GetNumFaces();
float * u = new float[g_nsamples*nfaces], * uPtr = u,
* v = new float[g_nsamples*nfaces], * vPtr = v;
std::vector<LocationArray> locs(nfaces);
srand( static_cast<int>(2147483647) ); // use a large Pell prime number
for (int face=0; face<nfaces; ++face) {
LocationArray & larray = locs[face];
larray.ptexIdx = face;
larray.numLocations = g_nsamples;
larray.s = uPtr;
larray.t = vPtr;
for (int j=0; j<g_nsamples; ++j, ++uPtr, ++vPtr) {
*uPtr = (float)rand()/(float)RAND_MAX;
*vPtr = (float)rand()/(float)RAND_MAX;
}
}
delete g_controlStencils;
g_controlStencils = Far::LimitStencilTablesFactory::Create(*refiner, locs);
delete [] u;
delete [] v;
g_nsamplesDrawn = g_controlStencils->GetNumStencils();
// Create control vertex buffer (layout: [ P(xyz) ] )
delete g_controlValues;
g_controlValues = Osd::CpuVertexBuffer::Create(3, nverts);
// Create eval context & data buffers
delete g_stencilValues;
g_stencilValues = Osd::CpuGLVertexBuffer::Create(3, g_controlStencils->GetNumStencils() * 6 );
delete shape;
delete refiner;
updateGeom();
// Bind g_stencilValues as GL_LINES VAO
glBindVertexArray(g_stencilsVAO);
glBindBuffer(GL_ARRAY_BUFFER, g_stencilValues->BindVBO());
glBindVertexArray(0);
}
//------------------------------------------------------------------------------
class GLSLProgram {
public:
GLSLProgram() : _program(0), _vtxSrc(0), _frgSrc(0) { }
struct Attribute {
std::string name;
GLuint location;
GLuint size;
};
void SetVertexShaderSource( char const * src ) {
_vtxSrc = src;
}
void SetFragShaderSource( char const * src ) {
_frgSrc = src;
}
void AddAttribute( char const * attr, int size ) {
Attribute a;
a.name = attr;
a.size = size;
_attrs.push_back(a);
}
void EnableVertexAttributes( ) {
2013-09-13 16:18:13 +00:00
long int offset = 0;
for (AttrList::iterator i=_attrs.begin(); i!=_attrs.end(); ++i) {
glEnableVertexAttribArray( i->location );
glVertexAttribPointer( i->location, i->size,
GL_FLOAT, GL_FALSE, sizeof(GLfloat) * _attrStride, (void*)offset);
offset += sizeof(GLfloat) * i->size;
}
}
GLuint GetUniformModelViewProjectionMatrix() const {
return _uniformModelViewProjectionMatrix;
}
void Use( ) {
if (not _program) {
assert( _vtxSrc and _frgSrc );
_program = glCreateProgram();
GLuint vertexShader = compileShader(GL_VERTEX_SHADER, _vtxSrc);
GLuint fragmentShader = compileShader(GL_FRAGMENT_SHADER, _frgSrc);
glAttachShader(_program, vertexShader);
glAttachShader(_program, fragmentShader);
_attrStride=0;
int count=0;
for (AttrList::iterator i=_attrs.begin(); i!=_attrs.end(); ++i, ++count) {
glBindAttribLocation(_program, count, i->name.c_str());
_attrStride += i->size;
}
glBindFragDataLocation(_program, 0, "color");
glLinkProgram(_program);
GLint status;
glGetProgramiv(_program, GL_LINK_STATUS, &status);
if (status == GL_FALSE) {
GLint infoLogLength;
glGetProgramiv(_program, GL_INFO_LOG_LENGTH, &infoLogLength);
char *infoLog = new char[infoLogLength];
glGetProgramInfoLog(_program, infoLogLength, NULL, infoLog);
printf("%s\n", infoLog);
delete[] infoLog;
exit(1);
}
_uniformModelViewProjectionMatrix =
glGetUniformLocation(_program, "ModelViewProjectionMatrix");
for (AttrList::iterator i=_attrs.begin(); i!=_attrs.end(); ++i) {
i->location = glGetAttribLocation(_program, i->name.c_str());
}
}
glUseProgram(_program);
}
private:
GLuint _program;
GLuint _uniformModelViewProjectionMatrix;
char const * _vtxSrc,
* _frgSrc;
typedef std::list<Attribute> AttrList;
AttrList _attrs;
int _attrStride;
};
GLSLProgram g_cageProgram,
g_samplesProgram;
//------------------------------------------------------------------------------
static bool
linkDefaultPrograms() {
#if defined(GL_ARB_tessellation_shader) || defined(GL_VERSION_4_0)
#define GLSL_VERSION_DEFINE "#version 400\n"
#else
#define GLSL_VERSION_DEFINE "#version 150\n"
#endif
{ // setup control cage program
static const char *vsSrc =
GLSL_VERSION_DEFINE
"in vec3 position;\n"
"in vec3 color;\n"
"out vec4 fragColor;\n"
"uniform mat4 ModelViewProjectionMatrix;\n"
"void main() {\n"
" fragColor = vec4(color, 1);\n"
" gl_Position = ModelViewProjectionMatrix * "
" vec4(position, 1);\n"
"}\n";
static const char *fsSrc =
GLSL_VERSION_DEFINE
"in vec4 fragColor;\n"
"out vec4 color;\n"
"void main() {\n"
" color = fragColor;\n"
"}\n";
g_cageProgram.SetVertexShaderSource(vsSrc);
g_cageProgram.SetFragShaderSource(fsSrc);
g_cageProgram.AddAttribute( "position",3 );
g_cageProgram.AddAttribute( "color",3 );
}
{ // setup samples program
static const char *vsSrc =
GLSL_VERSION_DEFINE
"in vec3 position;\n"
"uniform mat4 ModelViewProjectionMatrix;\n"
"void main() {\n"
" gl_Position = ModelViewProjectionMatrix * "
" vec4(position, 1);\n"
"}\n";
static const char *fsSrc =
GLSL_VERSION_DEFINE
"out vec4 color;\n"
"const vec4 colors[3] = vec4[3]( vec4(1.0,0.0,0.0,1.0), \n"
" vec4(0.0,1.0,0.0,1.0), \n"
" vec4(0.0,0.0,1.0,1.0) ); \n"
"void main() {\n"
" color = colors[gl_PrimitiveID % 3];\n"
"}\n";
g_samplesProgram.SetVertexShaderSource(vsSrc);
g_samplesProgram.SetFragShaderSource(fsSrc);
g_samplesProgram.AddAttribute( "position",3 );
}
return true;
}
//------------------------------------------------------------------------------
static inline void
setSharpnessColor(float s, float *r, float *g, float *b) {
// 0.0 2.0 4.0
// green --- yellow --- red
*r = std::min(1.0f, s * 0.5f);
*g = std::min(1.0f, 2.0f - s*0.5f);
*b = 0;
}
//------------------------------------------------------------------------------
static void
drawCageEdges() {
g_cageProgram.Use( );
glUniformMatrix4fv(g_cageProgram.GetUniformModelViewProjectionMatrix(),
1, GL_FALSE, g_transformData.ModelViewProjectionMatrix);
std::vector<float> vbo;
vbo.reserve(g_coarseEdges.size() * 6);
float * positions = g_controlValues->BindCpuBuffer();
float r, g, b;
for (int i = 0; i < (int)g_coarseEdges.size(); i+=2) {
setSharpnessColor(g_coarseEdgeSharpness[i/2], &r, &g, &b);
for (int j = 0; j < 2; ++j) {
vbo.push_back(positions[g_coarseEdges[i+j]*3]);
vbo.push_back(positions[g_coarseEdges[i+j]*3+1]);
vbo.push_back(positions[g_coarseEdges[i+j]*3+2]);
vbo.push_back(r);
vbo.push_back(g);
vbo.push_back(b);
}
}
glBindVertexArray(g_cageEdgeVAO);
glBindBuffer(GL_ARRAY_BUFFER, g_cageEdgeVBO);
glBufferData(GL_ARRAY_BUFFER, (int)vbo.size() * sizeof(float), &vbo[0],
GL_STATIC_DRAW);
g_cageProgram.EnableVertexAttributes();
glDrawArrays(GL_LINES, 0, (int)g_coarseEdges.size());
glBindVertexArray(0);
glUseProgram(0);
}
//------------------------------------------------------------------------------
static void
drawCageVertices() {
g_cageProgram.Use( );
glUniformMatrix4fv(g_cageProgram.GetUniformModelViewProjectionMatrix(),
1, GL_FALSE, g_transformData.ModelViewProjectionMatrix);
int numPoints = g_controlValues->GetNumVertices();
std::vector<float> vbo;
vbo.reserve(numPoints*6);
float * positions = g_controlValues->BindCpuBuffer();
float r, g, b;
for (int i = 0; i < numPoints; ++i) {
setSharpnessColor(g_coarseVertexSharpness[i], &r, &g, &b);
vbo.push_back(positions[i*3+0]);
vbo.push_back(positions[i*3+1]);
vbo.push_back(positions[i*3+2]);
vbo.push_back(r);
vbo.push_back(g);
vbo.push_back(b);
}
glBindVertexArray(g_cageVertexVAO);
glBindBuffer(GL_ARRAY_BUFFER, g_cageVertexVBO);
glBufferData(GL_ARRAY_BUFFER, (int)vbo.size() * sizeof(float), &vbo[0],
GL_STATIC_DRAW);
g_cageProgram.EnableVertexAttributes();
glPointSize(10.0f);
glDrawArrays(GL_POINTS, 0, numPoints);
glPointSize(1.0f);
glBindVertexArray(0);
glUseProgram(0);
}
//------------------------------------------------------------------------------
static void
drawStencils() {
g_samplesProgram.Use( );
glUniformMatrix4fv(g_cageProgram.GetUniformModelViewProjectionMatrix(),
1, GL_FALSE, g_transformData.ModelViewProjectionMatrix);
glBindVertexArray(g_stencilsVAO);
int numEdges = g_controlStencils->GetNumStencils() * 3;
g_samplesProgram.EnableVertexAttributes();
glBindBuffer(GL_ARRAY_BUFFER, g_stencilValues->BindVBO());
g_samplesProgram.EnableVertexAttributes();
Refurbish osd layer API. In OpenSubdiv 2.x, we encapsulated subdivision tables into compute context in osd layer since those tables are order-dependent and have to be applied in a certain manner. In 3.0, we adopted stencil table based refinement. It's more simple and such an encapsulation is no longer needed. Also 2.0 API has several ownership issues of GPU kernel caching, and forces unnecessary instantiation of controllers even though the cpu kernels typically don't need instances unlike GPU ones. This change completely revisit osd client facing APIs. All contexts and controllers were replaced with device-specific tables and evaluators. While we can still use consistent API across various device backends, unnecessary complexities have been removed. For example, cpu evaluator is just a set of static functions and also there's no need to replicate FarStencilTables to ComputeContext. Also the new API delegates the ownership of compiled GPU kernels to clients, for the better management of resources especially in multiple GPU environment. In addition to integrating ComputeController and EvalStencilController into a single function Evaluator::EvalStencils(), EvalLimit API is also added into Evaluator. This is working but still in progress, and we'll make a followup change for the complete implementation. -some naming convention changes: GLSLTransformFeedback to GLXFBEvaluator GLSLCompute to GLComputeEvaluator -move LimitLocation struct into examples/glEvalLimit. We're still discussing patch evaluation interface. Basically we'd like to tease all ptex-specific parametrization out of far/osd layer. TODO: -implments EvalPatches() in the right way -derivative evaluation API is still interim. -VertexBufferDescriptor needs a better API to advance its location -synchronization mechanism is not ideal (too global). -OsdMesh class is hacky. need to fix it.
2015-05-09 00:31:26 +00:00
glDrawArrays(GL_POINTS, 0, numEdges*2);
glDrawArrays(GL_LINES, 0, numEdges*2);
glBindVertexArray(0);
glUseProgram(0);
}
//------------------------------------------------------------------------------
static void
display() {
g_hud.GetFrameBuffer()->Bind();
Stopwatch s;
s.Start();
glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT);
glViewport(0, 0, g_width, g_height);
double aspect = g_width/(double)g_height;
identity(g_transformData.ModelViewMatrix);
translate(g_transformData.ModelViewMatrix, -g_pan[0], -g_pan[1], -g_dolly);
rotate(g_transformData.ModelViewMatrix, g_rotate[1], 1, 0, 0);
rotate(g_transformData.ModelViewMatrix, g_rotate[0], 0, 1, 0);
rotate(g_transformData.ModelViewMatrix, -90, 1, 0, 0);
translate(g_transformData.ModelViewMatrix,
-g_center[0], -g_center[1], -g_center[2]);
perspective(g_transformData.ProjectionMatrix,
45.0f, (float)aspect, 0.01f, 500.0f);
multMatrix(g_transformData.ModelViewProjectionMatrix,
g_transformData.ModelViewMatrix,
g_transformData.ProjectionMatrix);
2014-09-18 00:52:37 +00:00
glEnable(GL_DEPTH_TEST);
if (g_drawCageEdges)
drawCageEdges();
if (g_drawCageVertices)
drawCageVertices();
drawStencils();
s.Stop();
float drawCpuTime = float(s.GetElapsed() * 1000.0f);
s.Start();
glFinish();
s.Stop();
float drawGpuTime = float(s.GetElapsed() * 1000.0f);
g_hud.GetFrameBuffer()->ApplyImageShader();
if (g_hud.IsVisible()) {
g_fpsTimer.Stop();
double fps = 1.0/g_fpsTimer.GetElapsed();
g_fpsTimer.Start();
g_hud.DrawString(10, -100, "# stencils : %d", g_nsamplesDrawn);
g_hud.DrawString(10, -80, "EvalStencils : %.3f ms", g_evalTime);
g_hud.DrawString(10, -60, "GPU Draw : %.3f ms", drawGpuTime);
g_hud.DrawString(10, -40, "CPU Draw : %.3f ms", drawCpuTime);
g_hud.DrawString(10, -20, "FPS : %3.1f", fps);
g_hud.Flush();
}
glFinish();
//checkGLErrors("display leave");
}
//------------------------------------------------------------------------------
static void
idle() {
if (not g_freeze)
g_frame++;
updateGeom();
if (g_repeatCount != 0 and g_frame >= g_repeatCount)
g_running = 0;
}
//------------------------------------------------------------------------------
static void
motion(GLFWwindow *, double dx, double dy) {
int x=(int)dx, y=(int)dy;
if (g_mbutton[0] && !g_mbutton[1] && !g_mbutton[2]) {
// orbit
g_rotate[0] += x - g_prev_x;
g_rotate[1] += y - g_prev_y;
} else if (!g_mbutton[0] && !g_mbutton[1] && g_mbutton[2]) {
// pan
g_pan[0] -= g_dolly*(x - g_prev_x)/g_width;
g_pan[1] += g_dolly*(y - g_prev_y)/g_height;
} else if ((g_mbutton[0] && !g_mbutton[1] && g_mbutton[2]) or
(!g_mbutton[0] && g_mbutton[1] && !g_mbutton[2])) {
// dolly
g_dolly -= g_dolly*0.01f*(x - g_prev_x);
if(g_dolly <= 0.01) g_dolly = 0.01f;
}
g_prev_x = x;
g_prev_y = y;
}
//------------------------------------------------------------------------------
static void
mouse(GLFWwindow *, int button, int state, int /* mods */) {
if (button == 0 && state == GLFW_PRESS && g_hud.MouseClick(g_prev_x, g_prev_y))
return;
if (button < 3) {
g_mbutton[button] = (state == GLFW_PRESS);
}
}
//------------------------------------------------------------------------------
static void
reshape(GLFWwindow *, int width, int height) {
g_width = width;
g_height = height;
int windowWidth = g_width, windowHeight = g_height;
// window size might not match framebuffer size on a high DPI display
glfwGetWindowSize(g_window, &windowWidth, &windowHeight);
g_hud.Rebuild(windowWidth, windowHeight, width, height);
}
//------------------------------------------------------------------------------
void windowClose(GLFWwindow*) {
g_running = false;
}
//------------------------------------------------------------------------------
static void
rebuildMesh() {
createMesh( g_defaultShapes[g_currentShape], g_isolationLevel );
}
//------------------------------------------------------------------------------
static void
setSamples(bool add) {
g_nsamples += add ? 1000 : -1000;
g_nsamples = std::max(1000, g_nsamples);
rebuildMesh();
}
//------------------------------------------------------------------------------
static void
keyboard(GLFWwindow *, int key, int /* scancode */, int event, int /* mods */) {
if (event == GLFW_RELEASE) return;
if (g_hud.KeyDown(tolower(key))) return;
switch (key) {
case 'Q': g_running = 0; break;
case '=': setSamples(true); break;
case '-': setSamples(false); break;
case GLFW_KEY_ESCAPE: g_hud.SetVisible(!g_hud.IsVisible()); break;
}
}
//------------------------------------------------------------------------------
static void
callbackKernel(int k) {
g_kernel = k;
}
static void
callbackLevel(int l) {
g_isolationLevel = l;
rebuildMesh();
}
//------------------------------------------------------------------------------
static void
callbackAnimate(bool checked, int /* m */) {
g_moveScale = checked;
}
//------------------------------------------------------------------------------
static void
callbackFreeze(bool checked, int /* f */) {
g_freeze = checked;
}
//------------------------------------------------------------------------------
static void
callbackDisplayCageVertices(bool checked, int /* d */) {
g_drawCageVertices = checked;
}
//------------------------------------------------------------------------------
static void
callbackDisplayCageEdges(bool checked, int /* d */) {
g_drawCageEdges = checked;
}
static void
callbackBilinear(bool checked, int /* a */) {
g_bilinear = checked;
rebuildMesh();
}
//------------------------------------------------------------------------------
static void
callbackModel(int m) {
if (m < 0)
m = 0;
if (m >= (int)g_defaultShapes.size())
m = (int)g_defaultShapes.size() - 1;
g_currentShape = m;
rebuildMesh();
}
//------------------------------------------------------------------------------
static void
initHUD() {
int windowWidth = g_width, windowHeight = g_height,
frameBufferWidth = g_width, frameBufferHeight = g_height;
// window size might not match framebuffer size on a high DPI display
glfwGetWindowSize(g_window, &windowWidth, &windowHeight);
glfwGetFramebufferSize(g_window, &frameBufferWidth, &frameBufferHeight);
g_hud.Init(windowWidth, windowHeight, frameBufferWidth, frameBufferHeight);
g_hud.SetFrameBuffer(new GLFrameBuffer);
g_hud.AddCheckBox("Cage Edges (H)", true, 10, 10, callbackDisplayCageEdges, 0, 'h');
g_hud.AddCheckBox("Cage Verts (J)", true, 10, 30, callbackDisplayCageVertices, 0, 'j');
g_hud.AddCheckBox("Animate vertices (M)", g_moveScale != 0, 10, 50, callbackAnimate, 0, 'm');
g_hud.AddCheckBox("Freeze (spc)", false, 10, 70, callbackFreeze, 0, ' ');
g_hud.AddCheckBox("Bilinear Stencils (`)", g_bilinear!=0, 10, 190, callbackBilinear, 0, '`');
int compute_pulldown = g_hud.AddPullDown("Compute (K)", 250, 10, 300, callbackKernel, 'k');
g_hud.AddPullDownButton(compute_pulldown, "CPU", kCPU);
#ifdef OPENSUBDIV_HAS_OPENMP
g_hud.AddPullDownButton(compute_pulldown, "OpenMP", kOPENMP);
#endif
#ifdef OPENSUBDIV_HAS_TBB
g_hud.AddPullDownButton(compute_pulldown, "TBB", kTBB);
#endif
for (int i = 1; i < 11; ++i) {
char level[16];
sprintf(level, "Lv. %d", i);
g_hud.AddRadioButton(3, level, i==g_isolationLevel, 10, 210+i*20, callbackLevel, i, '0'+(i%10));
}
int pulldown_handle = g_hud.AddPullDown("Shape (N)", -300, 10, 300, callbackModel, 'n');
for (int i = 0; i < (int)g_defaultShapes.size(); ++i) {
g_hud.AddPullDownButton(pulldown_handle, g_defaultShapes[i].name.c_str(),i);
}
}
//------------------------------------------------------------------------------
static void
initGL() {
glClearColor(0.1f, 0.1f, 0.1f, 0.0f);
glEnable(GL_DEPTH_TEST);
glDepthFunc(GL_LEQUAL);
glCullFace(GL_BACK);
glEnable(GL_CULL_FACE);
glGenVertexArrays(1, &g_cageVertexVAO);
glGenVertexArrays(1, &g_cageEdgeVAO);
glGenVertexArrays(1, &g_stencilsVAO);
glGenBuffers(1, &g_cageVertexVBO);
glGenBuffers(1, &g_cageEdgeVBO);
}
//------------------------------------------------------------------------------
static void
uninitGL() {
glDeleteBuffers(1, &g_cageVertexVBO);
glDeleteBuffers(1, &g_cageEdgeVBO);
glDeleteVertexArrays(1, &g_cageVertexVAO);
glDeleteVertexArrays(1, &g_cageEdgeVAO);
glDeleteVertexArrays(1, &g_stencilsVAO);
}
//------------------------------------------------------------------------------
static void
callbackErrorGLFW(int error, const char* description) {
fprintf(stderr, "GLFW Error (%d) : %s\n", error, description);
}
//------------------------------------------------------------------------------
static void
setGLCoreProfile() {
#define glfwOpenWindowHint glfwWindowHint
#define GLFW_OPENGL_VERSION_MAJOR GLFW_CONTEXT_VERSION_MAJOR
#define GLFW_OPENGL_VERSION_MINOR GLFW_CONTEXT_VERSION_MINOR
glfwOpenWindowHint(GLFW_OPENGL_PROFILE, GLFW_OPENGL_CORE_PROFILE);
#if not defined(__APPLE__)
glfwOpenWindowHint(GLFW_OPENGL_VERSION_MAJOR, 4);
glfwOpenWindowHint(GLFW_OPENGL_VERSION_MINOR, 2);
#else
glfwOpenWindowHint(GLFW_OPENGL_VERSION_MAJOR, 3);
glfwOpenWindowHint(GLFW_OPENGL_VERSION_MINOR, 2);
#endif
glfwOpenWindowHint(GLFW_OPENGL_FORWARD_COMPAT, GL_TRUE);
}
//------------------------------------------------------------------------------
int main(int argc, char **argv) {
bool fullscreen = false;
std::string str;
for (int i = 1; i < argc; ++i) {
if (!strcmp(argv[i], "-d")) {
g_isolationLevel = atoi(argv[++i]);
} else if (!strcmp(argv[i], "-f")) {
fullscreen = true;
} else {
std::ifstream ifs(argv[1]);
if (ifs) {
std::stringstream ss;
ss << ifs.rdbuf();
ifs.close();
str = ss.str();
g_defaultShapes.push_back(ShapeDesc(argv[1], str.c_str(), kCatmark));
}
}
}
initShapes();
glfwSetErrorCallback(callbackErrorGLFW);
if (not glfwInit()) {
printf("Failed to initialize GLFW\n");
return 1;
}
static const char windowTitle[] = "OpenSubdiv glStencilViewer " OPENSUBDIV_VERSION_STRING;
#define CORE_PROFILE
#ifdef CORE_PROFILE
setGLCoreProfile();
#endif
if (fullscreen) {
g_primary = glfwGetPrimaryMonitor();
// apparently glfwGetPrimaryMonitor fails under linux : if no primary,
// settle for the first one in the list
if (not g_primary) {
int count=0;
GLFWmonitor ** monitors = glfwGetMonitors(&count);
if (count)
g_primary = monitors[0];
}
if (g_primary) {
GLFWvidmode const * vidmode = glfwGetVideoMode(g_primary);
g_width = vidmode->width;
g_height = vidmode->height;
}
}
if (not (g_window=glfwCreateWindow(g_width, g_height, windowTitle,
fullscreen and g_primary ? g_primary : NULL, NULL))) {
printf("Failed to open window.\n");
glfwTerminate();
return 1;
}
glfwMakeContextCurrent(g_window);
// accommodate high DPI displays (e.g. mac retina displays)
glfwGetFramebufferSize(g_window, &g_width, &g_height);
glfwSetFramebufferSizeCallback(g_window, reshape);
glfwSetKeyCallback(g_window, keyboard);
glfwSetCursorPosCallback(g_window, motion);
glfwSetMouseButtonCallback(g_window, mouse);
glfwSetWindowCloseCallback(g_window, windowClose);
#if defined(OSD_USES_GLEW)
#ifdef CORE_PROFILE
// this is the only way to initialize glew correctly under core profile context.
glewExperimental = true;
#endif
if (GLenum r = glewInit() != GLEW_OK) {
printf("Failed to initialize glew. Error = %s\n", glewGetErrorString(r));
exit(1);
}
#ifdef CORE_PROFILE
// clear GL errors which was generated during glewInit()
glGetError();
#endif
#endif
initGL();
linkDefaultPrograms();
glfwSwapInterval(0);
initHUD();
callbackModel(g_currentShape);
while (g_running) {
idle();
display();
glfwPollEvents();
glfwSwapBuffers(g_window);
glFinish();
}
uninitGL();
glfwTerminate();
}