mirror of
https://github.com/gnif/LookingGlass.git
synced 2024-11-14 10:08:24 +00:00
bb91b41c64
Using 4x4 means that some pixels will be outside of the lanczos window. The ideal lanczos function should in fact be zero in those areas, so we shouldn't waste time processing those pixels. I can't notice any difference in the results.
61 lines
1.1 KiB
GLSL
61 lines
1.1 KiB
GLSL
#version 300 es
|
|
precision mediump float;
|
|
|
|
#define PI 3.141592653589793
|
|
|
|
in vec2 fragCoord;
|
|
out vec4 fragColor;
|
|
|
|
uniform sampler2D texture;
|
|
float sinc(float x)
|
|
{
|
|
return x == 0.0 ? 1.0 : sin(x * PI) / (x * PI);
|
|
}
|
|
|
|
float lanczos(float x)
|
|
{
|
|
return sinc(x) * sinc(x * 0.5);
|
|
}
|
|
|
|
float lanczos(vec2 v)
|
|
{
|
|
return lanczos(v.x) * lanczos(v.y);
|
|
}
|
|
|
|
void main()
|
|
{
|
|
vec2 size = vec2(textureSize(texture, 0));
|
|
vec2 pos = fragCoord * size;
|
|
vec2 invSize = 1.0 / size;
|
|
vec2 uvc = floor(pos) + vec2(0.5, 0.5);
|
|
|
|
vec2 uvs[9] = vec2[](
|
|
uvc + vec2(-1.0, -1.0),
|
|
uvc + vec2(-1.0, 0.0),
|
|
uvc + vec2(-1.0, 1.0),
|
|
uvc + vec2( 0.0, -1.0),
|
|
uvc + vec2( 0.0, 0.0),
|
|
uvc + vec2( 0.0, 1.0),
|
|
uvc + vec2( 1.0, -1.0),
|
|
uvc + vec2( 1.0, 0.0),
|
|
uvc + vec2( 1.0, 1.0)
|
|
);
|
|
|
|
float factors[9];
|
|
float sum = 0.0;
|
|
for (int i = 0; i < 9; ++i)
|
|
{
|
|
factors[i] = lanczos(uvs[i] - fragCoord * size);
|
|
sum += factors[i];
|
|
}
|
|
|
|
for (int i = 0; i < 9; ++i)
|
|
factors[i] /= sum;
|
|
|
|
vec3 color = vec3(0.0);
|
|
for (int i = 0; i < 9; ++i)
|
|
color += texture2D(texture, uvs[i] * invSize).rgb * factors[i];
|
|
|
|
fragColor = vec4(color, 1.0);
|
|
}
|