Skip to content

GitLab

The expiration time for new job artifacts in CI/CD pipelines is now 30 days (GitLab default). Previously generated artifacts in already completed jobs will not be affected by the change. The latest artifacts for all jobs in the latest successful pipelines will be kept. More information: https://gitlab.lrz.de/help/user/admin_area/settings/continuous_integration.html#default-artifacts-expiration

Commit e53955a4 authored by Christian Schulte zu Berge's avatar Christian Schulte zu Berge
Browse files

IpsviRaycaster cleanup and added some documentation.

parent 6306b833
Hide whitespace changes
Inline Side-by-side
core/glsl/tools/shading.frag
View file @ e53955a4
......@@ -115,7 +115,7 @@ vec3 calculatePhongShading(in vec3 position, in LightSource light, in vec3 camer
* \param normal Normal
* \param materialColor Material color (used for all shading coefficients)
*/
vec3 calculatePhongShading(in vec3 position, in LightSource light, in vec3 camera, in vec3 normal, in vec3 materialColor) {
vec3 calculatePhongShading(in vec3 position, in vec3 ambientColorOverride, LightSource light, in vec3 camera, in vec3 normal, in vec3 materialColor) {
vec3 N = normalize(normal);
vec3 V = normalize(camera - position);
......@@ -124,7 +124,7 @@ vec3 calculatePhongShading(in vec3 position, in LightSource light, in vec3 camer
float d = length(L);
L /= d;
vec3 toReturn = materialColor * light._ambientColor; // ambient term
vec3 toReturn = materialColor * ambientColorOverride; // ambient term
toReturn += materialColor * getDiffuseTerm(light._diffuseColor, N, L);
toReturn += materialColor * getSpecularTerm(light._specularColor, N, L, V, light._shininess);
#ifdef PHONG_APPLY_ATTENUATION
......@@ -133,6 +133,19 @@ vec3 calculatePhongShading(in vec3 position, in LightSource light, in vec3 camer
return toReturn;
}
/**
* Computes the phong shading according to the given parameters.
*
* \param position sample position
* \param light LightSource
* \param camera Camera position
* \param normal Normal
* \param materialColor Material color (used for all shading coefficients)
*/
vec3 calculatePhongShading(in vec3 position, in LightSource light, in vec3 camera, in vec3 normal, in vec3 materialColor) {
return calculatePhongShading(position, light._ambientColor, light, camera, normal, materialColor);
}
/**
* Computes the phong shading intensity according to the given parameters.
......
modules/vis/glsl/ipsviraycaster.frag
View file @ e53955a4
......@@ -90,6 +90,9 @@ ivec2 calcIcSamplePosition(vec3 worldPosition) {
return ivec2(round(dot(worldProjected, _icRightVector)), round(dot(worldProjected, _icUpVector)));
}
// the composite function can be used for additional shadow ray integration
// from the current sample to the position of the IC.
// However, it's currently not used for performance reasons
void composite(vec3 startPosition, vec3 endPosition, inout float opacity) {
vec3 direction = endPosition - startPosition;
float t = _samplingStepSize;
......@@ -145,11 +148,6 @@ vec4 performRaycasting(in vec3 entryPoint, in vec3 exitPoint, in vec2 texCoords)
// optimization: Only store/load when the icPosition has changed
// otherwise we can reuse the variables from the previous sample
if (icPositionPrev != icPosition) {
// if there is no updated illumination information to be saved,
// carry over the old pixel
//if (! toBeSaved)
// icOut = imageLoad(_icImageIn, icPositionPrev);
// write illumination information
if (toBeSaved)
imageStore(_icImageOut, icPositionPrev, vec4(icOut));
......@@ -171,8 +169,7 @@ vec4 performRaycasting(in vec3 entryPoint, in vec3 exitPoint, in vec2 texCoords)
// perform compositing
if (color.a > 0.0) {
// compute gradient (needed for shading and normals)
vec3 gradient = computeGradient(_volume, _volumeTextureParams, samplePosition);
color.rgb = calculatePhongShading(worldPos, _lightSource, _cameraPosition, gradient, color.rgb);
vec3 gradient = computeGradient(_volume, _volumeTextureParams, samplePosition);
// accomodate for variable sampling rates
color.a = 1.0 - pow(1.0 - color.a, _samplingStepSize * SAMPLING_BASE_INTERVAL_RCP);
......@@ -186,7 +183,8 @@ vec4 performRaycasting(in vec3 entryPoint, in vec3 exitPoint, in vec2 texCoords)
toBeSaved = true;
// apply shadowing
color.rgb *= (1.0 - icIn * _shadowIntensity);
const vec3 ambientColorOverride = _lightSource._ambientColor * (1.0 - icIn * _shadowIntensity);
color.rgb = calculatePhongShading(worldPos, ambientColorOverride, _lightSource, _cameraPosition, gradient, color.rgb);
// front-to-back compositing along view direction
result.rgb = result.rgb + color.rgb * color.a * (1.0 - result.a);
......@@ -212,8 +210,6 @@ vec4 performRaycasting(in vec3 entryPoint, in vec3 exitPoint, in vec2 texCoords)
t += _samplingStepSize / len;
}
//if (! toBeSaved)
// icOut = imageLoad(_icImageIn, icPositionPrev);
if (toBeSaved)
imageStore(_icImageOut, icPositionPrev, vec4(icOut));
......
modules/vis/processors/ipsviraycaster.cpp
View file @ e53955a4
......@@ -43,8 +43,7 @@ namespace campvis {
, p_lightId("LightId", "Input Light Source", "lightsource", DataNameProperty::READ)
, p_sweepLineWidth("SweepLineWidth", "Sweep Line Width", 2, 1, 32)
, p_icTextureSize("IcTextureSize", "Illumination Cache Texture Size", cgt::ivec2(512), cgt::ivec2(32), cgt::ivec2(2048))
, p_shadowIntensity("ShadowIntensity", "Shadow Intensity", .9f, 0.f, 1.f)
, p_numLines("NumLines", "Max Number of Lines", 2000, 1, 2000)
, p_shadowIntensity("ShadowIntensity", "Shadow Intensity", .75f, 0.f, 1.f)
, _vhm(nullptr)
{
_icTextures[0] = nullptr;
......@@ -54,7 +53,6 @@ namespace campvis {
addProperty(p_sweepLineWidth);
addProperty(p_icTextureSize, INVALID_RESULT | INVALID_IC_TEXTURES);
addProperty(p_shadowIntensity);
addProperty(p_numLines);
setPropertyInvalidationLevel(p_transferFunction, INVALID_BBV | INVALID_RESULT);
setPropertyInvalidationLevel(p_sourceImageID, INVALID_BBV | INVALID_RESULT);
......@@ -82,13 +80,13 @@ namespace campvis {
}
void IpsviRaycaster::processImpl(DataContainer& data, ImageRepresentationGL::ScopedRepresentation& image) {
// (re)create Illumination Cache (IC) textures if needed
if (getInvalidationLevel() & INVALID_IC_TEXTURES) {
delete _icTextures[0];
delete _icTextures[1];
cgt::ivec3 icSize(p_icTextureSize.getValue(), 1);
cgt::TextureUnit icUnit;
icUnit.activate();
_icTextures[0] = new cgt::Texture(GL_TEXTURE_2D, icSize, GL_R32F);
_icTextures[1] = new cgt::Texture(GL_TEXTURE_2D, icSize, GL_R32F);
......@@ -96,6 +94,7 @@ namespace campvis {
validate(INVALID_IC_TEXTURES);
}
// update VoxelHierarchyMapper's hierarchy
if (getInvalidationLevel() & INVALID_BBV) {
_shader->deactivate();
_vhm->createHierarchy(image, p_transferFunction.getTF());
......@@ -116,23 +115,22 @@ namespace campvis {
processDirectional(data, image, *camera, *light);
}
else {
LDEBUG("Could not load light source from DataContainer.");
LWARNING("Could not load all the needed data from the DataContainer.");
}
}
void IpsviRaycaster::processDirectional(DataContainer& data, ImageRepresentationGL::ScopedRepresentation& image, const CameraData& camera, const LightSourceData& light) {
const cgt::vec3& lightSink = camera.getCamera().getFocus();
const cgt::vec3 lightSource = camera.getCamera().getFocus() + light.getLightPosition();
const cgt::vec3& lightDirection = light.getLightPosition();
const cgt::vec3 lightSource = camera.getCamera().getFocus() - light.getLightPosition();
const cgt::vec3& lightDirection = -light.getLightPosition();
// TODO: This should be a world to NDC space conversion, but it does not work... :(
// transformation matrices for world -> NDC -> viewport
const auto V = camera.getCamera().getViewMatrix();
const auto P = camera.getCamera().getProjectionMatrix();
// calculate viewport matrix for NDC -> viewport conversion
const cgt::vec2 halfViewport = cgt::vec2(getEffectiveViewportSize()) / 2.f;
const cgt::mat4 viewportMatrix = cgt::mat4::createTranslation(cgt::vec3(halfViewport, 0.f)) * cgt::mat4::createScale(cgt::vec3(halfViewport, 1.f));
// project light source and direction into pixel space
const cgt::vec4 projectedLight = viewportMatrix*P*V* cgt::vec4(lightSource, 1.f);
const cgt::vec4 projectedOrigin = viewportMatrix*P*V* cgt::vec4(lightSink, 1.f);
cgt::vec2 projectedLightDirection = projectedOrigin.xy()/projectedOrigin.w - projectedLight.xy()/projectedLight.w;
......@@ -155,26 +153,26 @@ namespace campvis {
sweepDir = TopToBottom;
}
// START: compute illumination cache (IC) plane/texture
// compute illumination cache (IC) plane
// the plane is defined by the light direction
cgt::vec3 icNormal = cgt::normalize(lightDirection);
cgt::vec3 icUpVector = (std::abs(cgt::dot(icNormal, cgt::vec3(0.f, 0.f, 1.f))) < 0.99) ? cgt::vec3(0.f, 0.f, 1.f) : cgt::vec3(0.f, 1.f, 0.f);
cgt::vec3 icRightVector = cgt::normalize(cgt::cross(icNormal, icUpVector));
icUpVector = cgt::normalize(cgt::cross(icRightVector, icNormal));
// project all 8 corners of the volume onto the IC plane
// project all 8 corners of the volume onto the IC plane to get the volume bounds in viewport space
cgt::Bounds worldBounds = image->getParent()->getWorldBounds();
cgt::Bounds viewportBounds;
cgt::vec3 minPixel(0.f), maxPixel(0.f);
std::vector<cgt::vec3> corners;
corners.push_back(cgt::vec3(worldBounds.getLLF().x, worldBounds.getLLF().y, worldBounds.getLLF().z));
corners.push_back(cgt::vec3(worldBounds.getLLF().x, worldBounds.getLLF().y, worldBounds.getURB().z));
corners.push_back(cgt::vec3(worldBounds.getLLF().x, worldBounds.getURB().y, worldBounds.getLLF().z));
corners.push_back(cgt::vec3(worldBounds.getLLF().x, worldBounds.getURB().y, worldBounds.getURB().z));
corners.push_back(cgt::vec3(worldBounds.getURB().x, worldBounds.getLLF().y, worldBounds.getLLF().z));
corners.push_back(cgt::vec3(worldBounds.getURB().x, worldBounds.getLLF().y, worldBounds.getURB().z));
corners.push_back(cgt::vec3(worldBounds.getURB().x, worldBounds.getURB().y, worldBounds.getLLF().z));
corners.push_back(cgt::vec3(worldBounds.getURB().x, worldBounds.getURB().y, worldBounds.getURB().z));
std::vector<cgt::vec3> corners(8, cgt::vec3(0.f));
corners[0] = cgt::vec3(worldBounds.getLLF().x, worldBounds.getLLF().y, worldBounds.getLLF().z);
corners[1] = cgt::vec3(worldBounds.getLLF().x, worldBounds.getLLF().y, worldBounds.getURB().z);
corners[2] = cgt::vec3(worldBounds.getLLF().x, worldBounds.getURB().y, worldBounds.getLLF().z);
corners[3] = cgt::vec3(worldBounds.getLLF().x, worldBounds.getURB().y, worldBounds.getURB().z);
corners[4] = cgt::vec3(worldBounds.getURB().x, worldBounds.getLLF().y, worldBounds.getLLF().z);
corners[5] = cgt::vec3(worldBounds.getURB().x, worldBounds.getLLF().y, worldBounds.getURB().z);
corners[6] = cgt::vec3(worldBounds.getURB().x, worldBounds.getURB().y, worldBounds.getLLF().z);
corners[7] = cgt::vec3(worldBounds.getURB().x, worldBounds.getURB().y, worldBounds.getURB().z);
for (auto i = 0; i < corners.size(); ++i) {
const cgt::vec3 diag = corners[i];
......@@ -195,6 +193,9 @@ namespace campvis {
icRightVector *= float(icSize.x - 1) / (std::ceil(maxPixel.x) - std::floor(minPixel.x)) ;
icUpVector *= float(icSize.y - 1) / (std::ceil(maxPixel.y) - std::floor(minPixel.y));
// * all preparations done, let's to the rendering *
// bind voxel hierarchy to shader
cgt::TextureUnit xorUnit, bbvUnit;
{
......@@ -239,8 +240,10 @@ namespace campvis {
glEnable(GL_DEPTH_TEST);
glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT);
auto viewportSize = getEffectiveViewportSize();
const auto viewportSize = getEffectiveViewportSize();
// setup the helper variables for rendering the sweep lines (according to the sweep direction)
// as additionall optimization, restrict the rendered lines to the volume extent in viewport space
cgt::mat4 projection;
cgt::mat4 viewScale;
cgt::vec3 viewTranslationBase;
......@@ -248,54 +251,54 @@ namespace campvis {
int lineMax;
float scale = 1.f;
float bias = 0.f;
switch (sweepDir) {
case LeftToRight:
scale = float(viewportSize.y) / viewportBounds.diagonal().y;
bias = viewportBounds.getLLF().y / float(viewportSize.y) * scale;
projection = cgt::mat4::createOrtho(0, viewportSize.x, scale-bias, -bias, -1, 1);
viewScale = cgt::mat4::createScale(cgt::vec3(p_sweepLineWidth.getValue(), 1.f, 1.f));
viewTranslationBase = cgt::vec3(1.f, 0.f, 0.f);
line += std::max(0, int(viewportBounds.getLLF().x));
lineMax = std::min(viewportSize.x, int(viewportBounds.getURB().x));
break;
case RightToLeft:
scale = float(viewportSize.y) / viewportBounds.diagonal().y;
bias = viewportBounds.getLLF().y / float(viewportSize.y) * scale;
projection = cgt::mat4::createOrtho(viewportSize.x, 0, scale-bias, -bias, -1, 1);
viewScale = cgt::mat4::createScale(cgt::vec3(p_sweepLineWidth.getValue(), 1.f, 1.f));
viewTranslationBase = cgt::vec3(1.f, 0.f, 0.f);
line += std::max(0, viewportSize.x - int(viewportBounds.getURB().x));
lineMax = std::min(viewportSize.x, viewportSize.x - int(viewportBounds.getLLF().x));
break;
case BottomToTop:
scale = float(viewportSize.x) / viewportBounds.diagonal().x;
bias = viewportBounds.getLLF().x / float(viewportSize.x) * scale;
projection = cgt::mat4::createOrtho(-bias, scale-bias, viewportSize.y, 0, -1, 1);
viewScale = cgt::mat4::createScale(cgt::vec3(1.f, p_sweepLineWidth.getValue(), 1.f));
viewTranslationBase = cgt::vec3(0.f, 1.f, 0.f);
line += std::max(0, int(viewportBounds.getLLF().y));
lineMax = std::min(viewportSize.y, int(viewportBounds.getURB().y));
break;
case TopToBottom:
scale = float(viewportSize.x) / viewportBounds.diagonal().x;
bias = viewportBounds.getLLF().x / float(viewportSize.x) * scale;
projection = cgt::mat4::createOrtho(-bias, scale-bias, 0, viewportSize.y, -1, 1);
viewScale = cgt::mat4::createScale(cgt::vec3(1.f, p_sweepLineWidth.getValue(), 1.f));
viewTranslationBase = cgt::vec3(0.f, 1.f, 0.f);
line += std::max(0, viewportSize.y - int(viewportBounds.getURB().y));
lineMax = std::min(viewportSize.y, viewportSize.y - int(viewportBounds.getLLF().y));
break;
case LeftToRight:
scale = float(viewportSize.y) / viewportBounds.diagonal().y;
bias = viewportBounds.getLLF().y / float(viewportSize.y) * scale;
projection = cgt::mat4::createOrtho(0, float(viewportSize.x), scale-bias, -bias, -1, 1);
viewScale = cgt::mat4::createScale(cgt::vec3(float(p_sweepLineWidth.getValue()), 1.f, 1.f));
viewTranslationBase = cgt::vec3(1.f, 0.f, 0.f);
line += std::max(0, int(viewportBounds.getLLF().x));
lineMax = std::min(viewportSize.x, int(viewportBounds.getURB().x));
break;
case RightToLeft:
scale = float(viewportSize.y) / viewportBounds.diagonal().y;
bias = viewportBounds.getLLF().y / float(viewportSize.y) * scale;
projection = cgt::mat4::createOrtho(float(viewportSize.x), 0, scale-bias, -bias, -1, 1);
viewScale = cgt::mat4::createScale(cgt::vec3(float(p_sweepLineWidth.getValue()), 1.f, 1.f));
viewTranslationBase = cgt::vec3(1.f, 0.f, 0.f);
line += std::max(0, viewportSize.x - int(viewportBounds.getURB().x));
lineMax = std::min(viewportSize.x, viewportSize.x - int(viewportBounds.getLLF().x));
break;
case BottomToTop:
scale = float(viewportSize.x) / viewportBounds.diagonal().x;
bias = viewportBounds.getLLF().x / float(viewportSize.x) * scale;
projection = cgt::mat4::createOrtho(-bias, scale-bias, float(viewportSize.y), 0, -1, 1);
viewScale = cgt::mat4::createScale(cgt::vec3(1.f, float(p_sweepLineWidth.getValue()), 1.f));
viewTranslationBase = cgt::vec3(0.f, 1.f, 0.f);
line += std::max(0, int(viewportBounds.getLLF().y));
lineMax = std::min(viewportSize.y, int(viewportBounds.getURB().y));
break;
case TopToBottom:
scale = float(viewportSize.x) / viewportBounds.diagonal().x;
bias = viewportBounds.getLLF().x / float(viewportSize.x) * scale;
projection = cgt::mat4::createOrtho(-bias, scale-bias, 0, float(viewportSize.y), -1, 1);
viewScale = cgt::mat4::createScale(cgt::vec3(1.f, float(p_sweepLineWidth.getValue()), 1.f));
viewTranslationBase = cgt::vec3(0.f, 1.f, 0.f);
line += std::max(0, viewportSize.y - int(viewportBounds.getURB().y));
lineMax = std::min(viewportSize.y, viewportSize.y - int(viewportBounds.getLLF().y));
break;
}
int evenOdd = 0;
// some more shader setup
_shader->setUniform("_projectionMatrix", projection);
GLint uIcImageIn = _shader->getUniformLocation("_icImageIn");
GLint uIcImageOut = _shader->getUniformLocation("_icImageOut");
GLint uViewMatrix = _shader->getUniformLocation("_viewMatrix");
// finally we can start rendering :)
int evenOdd = 0;
while (line < lineMax) {
// ping-pong buffering to avoid concurrent read-writes
if (evenOdd % 2 == 0) {
......@@ -313,9 +316,6 @@ namespace campvis {
line += p_sweepLineWidth.getValue();
++evenOdd;
glMemoryBarrier(GL_SHADER_IMAGE_ACCESS_BARRIER_BIT);
if (evenOdd > p_numLines.getValue())
break;
}
// restore state
......@@ -339,6 +339,11 @@ namespace campvis {
cgt::vec4 projectedLight = viewportMatrix * P * V * cgt::vec4(light.getLightPosition(), 1.f);
projectedLight /= projectedLight.w;
// The processing of point light sources is not yet implemented.
// It mostly works light four passes with directional lights. However, the projection
// of the illumination cache is no longer orthogonal. That complicates the it's definition.
// So far, I was too lazy to do so - ambient occlusion might be cooler anyway.
cgtAssert(false, "Processing of point light sources not yet implemented.");
}
}
modules/vis/processors/ipsviraycaster.h
View file @ e53955a4
......@@ -45,7 +45,11 @@ namespace campvis {
class VoxelHierarchyMapper;
/**
* Performs a simple volume ray casting.
* Raycaster that implements the Image Plane Sweep Volume Illumation algorithm of Sundn et al.
* This raycasting processor supports real-time directional light shadowing and additionally
* uses the VoxelHierarchyMapper for optimized entry-exit points.
*
* Requires OpenGL 4.4!
*/
class CAMPVIS_MODULES_API IpsviRaycaster : public RaycastingProcessor {
public:
......@@ -71,7 +75,7 @@ namespace campvis {
/// \see AbstractProcessor::getName()
virtual const std::string getName() const { return getId(); };
/// \see AbstractProcessor::getDescription()
virtual const std::string getDescription() const { return "Performs a simple volume ray casting."; };
virtual const std::string getDescription() const { return "Raycaster that implements the Image Plane Sweep Volume Illumation algorithm of Sundn et al."; };
/// \see AbstractProcessor::getAuthor()
virtual const std::string getAuthor() const { return "Christian Schulte zu Berge <christian.szb@in.tum.de>"; };
/// \see AbstractProcessor::getProcessorState()
......@@ -83,22 +87,19 @@ namespace campvis {
virtual void deinit();
DataNameProperty p_lightId; ///< Name/ID for the LightSource to use
IntProperty p_sweepLineWidth;
IVec2Property p_icTextureSize;
FloatProperty p_shadowIntensity;
IntProperty p_numLines;
IntProperty p_sweepLineWidth; ///< Width of the sweep line in pixels
IVec2Property p_icTextureSize; ///< Size of the Illumination Cache texture
FloatProperty p_shadowIntensity; ///< Intensity of the shadowing effect
protected:
/// \see RaycastingProcessor::processImpl()
virtual void processImpl(DataContainer& data, ImageRepresentationGL::ScopedRepresentation& image);
void processDirectional(DataContainer& data, ImageRepresentationGL::ScopedRepresentation& image, const CameraData& camera, const LightSourceData& light);
void processPointLight(DataContainer& data, ImageRepresentationGL::ScopedRepresentation& image, const CameraData& camera, const LightSourceData& light);
VoxelHierarchyMapper* _vhm;
cgt::Texture* _icTextures[2];
VoxelHierarchyMapper* _vhm; ///< for optimizing entry/exit points
cgt::Texture* _icTextures[2]; ///< Illumination cache textures
static const std::string loggerCat_;
};
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment