2656bf0a90
RGB to YUV converison was previously baked into every scale shader, but this work has been moved to the YUV packing shaders. The scale shaders now write RGBA instead. In the case where base and output resolutions are identical, the render texture is forwarded directly to the YUV pack step, skipping an entire fullscreen pass. Intel GPA, SetStablePowerState, Intel HD Graphics 530, NV12 1920x1080, Before: RGBA -> UYVX: ~321 us UYVX -> Y: ~480 us UYVX -> UV: ~127 us 1920x1080, After: [forward render texture] RGBA -> Y: ~487 us RGBA -> UV: ~131 us 1920x1080 -> 1280x720, Before: RGBA -> UYVX: ~268 us UYVX -> Y: ~209 us UYVX -> UV: ~57 us 1920x1080 -> 1280x720, After: RGBA -> RGBA (rescale): ~268 us RGBA -> Y: ~210 us RGBA -> UV: ~58 us
86 lines
1.9 KiB
Plaintext
86 lines
1.9 KiB
Plaintext
/*
|
|
* bilinear low res scaling, samples 8 pixels of a larger image to scale to a
|
|
* low resolution image below half size
|
|
*/
|
|
|
|
uniform float4x4 ViewProj;
|
|
uniform texture2d image;
|
|
|
|
sampler_state textureSampler {
|
|
Filter = Linear;
|
|
AddressU = Clamp;
|
|
AddressV = Clamp;
|
|
};
|
|
|
|
struct VertData {
|
|
float4 pos : POSITION;
|
|
float2 uv : TEXCOORD0;
|
|
};
|
|
|
|
VertData VSDefault(VertData v_in)
|
|
{
|
|
VertData vert_out;
|
|
vert_out.pos = mul(float4(v_in.pos.xyz, 1.0), ViewProj);
|
|
vert_out.uv = v_in.uv;
|
|
return vert_out;
|
|
}
|
|
|
|
float4 pixel(float2 uv)
|
|
{
|
|
return image.Sample(textureSampler, uv);
|
|
}
|
|
|
|
float4 DrawLowresBilinear(VertData v_in)
|
|
{
|
|
float2 uv = v_in.uv;
|
|
float2 stepxy = float2(ddx(uv.x), ddy(uv.y));
|
|
float2 stepxy1 = stepxy * 0.0625;
|
|
float2 stepxy3 = stepxy * 0.1875;
|
|
float2 stepxy5 = stepxy * 0.3125;
|
|
float2 stepxy7 = stepxy * 0.4375;
|
|
|
|
// Simulate Direct3D 8-sample pattern
|
|
float4 out_color;
|
|
out_color = pixel(uv + float2( stepxy1.x, -stepxy3.y));
|
|
out_color += pixel(uv + float2(-stepxy1.x, stepxy3.y));
|
|
out_color += pixel(uv + float2( stepxy5.x, stepxy1.y));
|
|
out_color += pixel(uv + float2(-stepxy3.x, -stepxy5.y));
|
|
out_color += pixel(uv + float2(-stepxy5.x, stepxy5.y));
|
|
out_color += pixel(uv + float2(-stepxy7.x, -stepxy1.y));
|
|
out_color += pixel(uv + float2( stepxy3.x, stepxy7.y));
|
|
out_color += pixel(uv + float2( stepxy7.x, -stepxy7.y));
|
|
return out_color * 0.125;
|
|
}
|
|
|
|
float4 PSDrawLowresBilinearRGBA(VertData v_in) : TARGET
|
|
{
|
|
return DrawLowresBilinear(v_in);
|
|
}
|
|
|
|
float4 PSDrawLowresBilinearRGBADivide(VertData v_in) : TARGET
|
|
{
|
|
float4 rgba = DrawLowresBilinear(v_in);
|
|
float alpha = rgba.a;
|
|
float multiplier = (alpha > 0.0) ? (1.0 / alpha) : 0.0;
|
|
return float4(rgba.rgb * multiplier, alpha);
|
|
}
|
|
|
|
technique Draw
|
|
{
|
|
pass
|
|
{
|
|
vertex_shader = VSDefault(v_in);
|
|
pixel_shader = PSDrawLowresBilinearRGBA(v_in);
|
|
}
|
|
}
|
|
|
|
technique DrawAlphaDivide
|
|
{
|
|
pass
|
|
{
|
|
vertex_shader = VSDefault(v_in);
|
|
pixel_shader = PSDrawLowresBilinearRGBADivide(v_in);
|
|
}
|
|
}
|
|
|