Graphics Reference
In-Depth Information
Example 14-27
Rendering from the Eye Position Shaders
(continued)
{
v_color = a_color;
gl_Position = u_mvpMatrix * a_position;
v_shadowCoord = u_mvpLightMatrix * a_position;
// transform from [−1,1] to [0,1];
v_shadowCoord = v_shadowCoord * 0.5 + 0.5;
}
// fragment shader
#version 300 es
precision lowp float;
uniform lowp sampler2DShadow s_shadowMap;
in vec4 v_color;
in vec4 v_shadowCoord;
layout(location = 0) out vec4 outColor;
float lookup ( float x, float y )
{
float pixelSize = 0.002; // 1/500
vec4 offset = vec4 ( x * pixelSize * v_shadowCoord.w,
y * pixelSize * v_shadowCoord.w,
0.0, 0.0 );
return textureProj ( s_shadowMap, v_shadowCoord + offset );
}
void main()
{
// 3x3 kernel with 4 taps per sample, effectively 6x6 PCF
float sum = 0.0;
float x, y;
for ( x = −2.0; x <= 2.0; x += 2.0 )
for ( y = −2.0; y <= 2.0; y += 2.0 )
sum += lookup ( x, y );
// divide sum by 9.0
sum = sum * 0.11;
outColor = v_color * sum;
}
In the vertex shader, we transform the vertex position twice: (1) using
the MVP matrix created from the eye position and (2) using the MVP
matrix created from the light position. The former result is recorded into
gl_Position
, while the latter result is recorded into
v_shadowCoord
.
Note that the
v_shadowCoord
result is exactly the same vertex position
result when we render to the shadow map. Armed with this knowledge,
