/*========================================================================= Program: Visualization Toolkit Module: vtkSobelGradientMagnitudePass.h Copyright (c) Ken Martin, Will Schroeder, Bill Lorensen All rights reserved. See Copyright.txt or http://www.kitware.com/Copyright.htm for details. This software is distributed WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the above copyright notice for more information. =========================================================================*/ /** * @class vtkSobelGradientMagnitudePass * @brief Implement a post-processing edge * detection with a Sobel gradient magnitude render pass. * * Detect the edges of the image renderered by its delegate. Edge-detection * uses a Sobel high-pass filter (3x3 kernel). * * This pass expects an initialized depth buffer and color buffer. * Initialized buffers means they have been cleared with farest z-value and * background color/gradient/transparent color. * An opaque pass may have been performed right after the initialization. * * The delegate is used once. * * Its delegate is usually set to a vtkCameraPass or to a post-processing pass. * * This pass requires a OpenGL context that supports texture objects (TO), * framebuffer objects (FBO) and GLSL. If not, it will emit an error message * and will render its delegate and return. * * @par Implementation: * To compute the gradient magnitude, the x and y components of the gradient * (Gx and Gy) have to be computed first. Each computation of Gx and Gy uses * a separable filter. * The first pass takes the image from the delegate as the single input * texture. * The first pass has two outputs, one for the first part of Gx, Gx1, result of * a convolution with (-1 0 1), one for the first part of Gy, Gy1, result of a * convolution with (1 2 1). * The second pass has two inputs, Gx1 and Gy1. Kernel (1 2 1)^T is applied * to Gx1 and kernel (-1 0 1)^T is applied to Gx2. It gives the values for * Gx and Gy. Those values are then used to compute the magnitude of the * gradient which is stored in the render target. * The gradient computation happens per component (R,G,B). A is arbitrarily set * to 1 (full opacity). * * @par Implementation: * \image html vtkSobelGradientMagnitudePassFigure.png * \image latex vtkSobelGradientMagnitudePassFigure.eps * * @sa * vtkRenderPass */ #ifndef vtkSobelGradientMagnitudePass_h #define vtkSobelGradientMagnitudePass_h #include "vtkImageProcessingPass.h" #include "vtkRenderingOpenGL2Module.h" // For export macro class vtkDepthPeelingPassLayerList; // Pimpl class vtkOpenGLFramebufferObject; class vtkOpenGLHelper; class vtkOpenGLRenderWindow; class vtkTextureObject; class VTKRENDERINGOPENGL2_EXPORT vtkSobelGradientMagnitudePass : public vtkImageProcessingPass { public: static vtkSobelGradientMagnitudePass* New(); vtkTypeMacro(vtkSobelGradientMagnitudePass, vtkImageProcessingPass); void PrintSelf(ostream& os, vtkIndent indent) override; /** * Perform rendering according to a render state \p s. * \pre s_exists: s!=0 */ void Render(const vtkRenderState* s) override; /** * Release graphics resources and ask components to release their own * resources. * \pre w_exists: w!=0 */ void ReleaseGraphicsResources(vtkWindow* w) override; protected: /** * Default constructor. DelegatePass is set to NULL. */ vtkSobelGradientMagnitudePass(); /** * Destructor. */ ~vtkSobelGradientMagnitudePass() override; /** * Graphics resources. */ vtkOpenGLFramebufferObject* FrameBufferObject; vtkTextureObject* Pass1; // render target for the scene vtkTextureObject* Gx1; // render target 0 for the first shader vtkTextureObject* Gy1; // render target 1 for the first shader // Structures for the various cell types we render. vtkOpenGLHelper* Program1; // shader to compute Gx1 and Gy1 vtkOpenGLHelper* Program2; // shader to compute |G| from Gx1 and Gy1 private: vtkSobelGradientMagnitudePass(const vtkSobelGradientMagnitudePass&) = delete; void operator=(const vtkSobelGradientMagnitudePass&) = delete; }; #endif