This filter is used to add blur to an image.

1  2  1
2  4  2
1  2  1

The demo associated with this filter is DEMO_gaussian_filter.xml
.
[Vertex_Shader] void main(void) { gl_TexCoord[0] = gl_MultiTexCoord0; gl_Position = ftransform(); } [Pixel_Shader] #define KERNEL_SIZE 9 // Gaussian kernel // 1 2 1 // 2 4 2 // 1 2 1 const float kernel[KERNEL_SIZE] = { 1.0/16.0, 2.0/16.0, 1.0/16.0, 2.0/16.0, 4.0/16.0, 2.0/16.0, 1.0/16.0, 2.0/16.0, 1.0/16.0 }; uniform sampler2D colorMap; uniform float width; uniform float height; const float step_w = 1.0/width; const float step_h = 1.0/height; const vec2 offset[KERNEL_SIZE] = { vec2(-step_w, -step_h), vec2(0.0, -step_h), vec2(step_w, -step_h), vec2(-step_w, 0.0), vec2(0.0, 0.0), vec2(step_w, 0.0), vec2(-step_w, step_h), vec2(0.0, step_h), vec2(step_w, step_h) }; void main(void) { int i = 0; vec4 sum = vec4(0.0); if(gl_TexCoord[0].s<0.495) { for( i=0; i<KERNEL_SIZE; i++ ) { vec4 tmp = texture2D(colorMap, gl_TexCoord[0].st + offset[i]); sum += tmp * kernel[i]; } } else if( gl_TexCoord[0].s>0.505 ) { sum = texture2D(colorMap, gl_TexCoord[0].xy); } else { sum = vec4(1.0, 0.0, 0.0, 1.0); } gl_FragColor = sum; }


Load the DEMO_gaussian_kernel.xml in Demoniak3D file and you will get the following result:


Fig.2 - Gaussian filter.The working of the shaders in this tutorial is the following: the base image is plated on a mesh plane. The left part of the image (texture coordinate u < 0.495) is filtered while the right part (u > 0.505) is not. All the texels that are located in the [0.495; 0.505] interval will be colored in red in order to form a vertical separating line.

The dimension of the base texture must be of a power of 2. In our case, they are 256x256. These measurements are passed as uniform variables to the pixel shaders in order to calculate the distances to access the neighboring pixels to the pixel being processed.

The vertex and pixel shaders being the same for all the filters, only the convolution core will be given.

If the code of the previous pixel shader does not work properly, it is possible to take of the branches and to unroll the loop as shown in the following code:

[Vertex_Shader] void main(void) { gl_TexCoord[0] = gl_MultiTexCoord0; gl_Position = ftransform(); } [Pixel_Shader] #define KERNEL_SIZE 9 // Gaussian kernel // 1 2 1 // 2 4 2 // 1 2 1 const float kernel[KERNEL_SIZE] = { 1.0/16.0, 2.0/16.0, 1.0/16.0, 2.0/16.0, 4.0/16.0, 2.0/16.0, 1.0/16.0, 2.0/16.0, 1.0/16.0 }; uniform sampler2D colorMap; uniform float width; uniform float height; const float step_w = 1.0/width; const float step_h = 1.0/height; const vec2 offset[KERNEL_SIZE] = { vec2(-step_w, -step_h), vec2(0.0, -step_h), vec2(step_w, -step_h), vec2(-step_w, 0.0), vec2(0.0, 0.0), vec2(step_w, 0.0), vec2(-step_w, step_h), vec2(0.0, step_h), vec2(step_w, step_h) }; void main(void) { int i = 0; vec4 sum = vec4(0.0); vec4 tmp = texture2D(colorMap, gl_TexCoord[0].st + offset[0]); sum += tmp * kernel[0]; tmp = texture2D(colorMap, gl_TexCoord[0].st + offset[1]); sum += tmp * kernel[1]; tmp = texture2D(colorMap, gl_TexCoord[0].st + offset[2]); sum += tmp * kernel[2]; tmp = texture2D(colorMap, gl_TexCoord[0].st + offset[3]); sum += tmp * kernel[3]; tmp = texture2D(colorMap, gl_TexCoord[0].st + offset[4]); sum += tmp * kernel[4]; tmp = texture2D(colorMap, gl_TexCoord[0].st + offset[5]); sum += tmp * kernel[5]; tmp = texture2D(colorMap, gl_TexCoord[0].st + offset[6]); sum += tmp * kernel[6]; tmp = texture2D(colorMap, gl_TexCoord[0].st + offset[7]); sum += tmp * kernel[7]; tmp = texture2D(colorMap, gl_TexCoord[0].st + offset[8]); sum += tmp * kernel[8]; gl_FragColor = sum; }


All the previous codes run properly on a nVidia GPU based graphic controller. Graphics controllers featuring the ATI chipsets seem not to support the constant array functionality. In case of an ATI chipset, the pixel shader code becomes:


[Pixel_Shader] #define KERNEL_SIZE 9 // Gaussian kernel // 1 2 1 // 2 4 2 // 1 2 1 float kernel[KERNEL_SIZE]; uniform sampler2D colorMap; uniform float width; uniform float height; float step_w = 1.0/width; float step_h = 1.0/height; vec2 offset[KERNEL_SIZE]; void main(void) { int i = 0; vec4 sum = vec4(0.0); offset[0] = vec2(-step_w, -step_h); offset[1] = vec2(0.0, -step_h); offset[2] = vec2(step_w, -step_h); offset[3] = vec2(-step_w, 0.0); offset[4] = vec2(0.0, 0.0); offset[5] = vec2(step_w, 0.0); offset[6] = vec2(-step_w, step_h); offset[7] = vec2(0.0, step_h); offset[8] = vec2(step_w, step_h); kernel[0] = 1.0/16.0; kernel[1] = 2.0/16.0; kernel[2] = 1.0/16.0; kernel[3] = 2.0/16.0; kernel[4] = 4.0/16.0; kernel[5] = 2.0/16.0; kernel[6] = 1.0/16.0; kernel[7] = 2.0/16.0; kernel[8] = 1.0/16.0; if(gl_TexCoord[0].s<0.495) { for( i=0; i<KERNEL_SIZE; i++ ) { vec4 tmp = texture2D(colorMap, gl_TexCoord[0].st + offset[i]); sum += tmp * kernel[i]; } } else if( gl_TexCoord[0].s>0.505 ) { sum = texture2D(colorMap, gl_TexCoord[0].xy); } else { sum = vec4(1.0, 0.0, 0.0, 1.0); } gl_FragColor = sum; }


The code of the pixel shader above is fully operational on a Radeon 9800pro graphics controller featuring the Catalyst 5.11 drivers. The accompanying projects are therefore provided in two versions: one exclusively intended for the nVidia chipsets while an other is meant for both ATI and nVidia.