slang-shaders/bezel/koko-aio/shaders/fxaa.slang

277 lines
10 KiB
Plaintext
Raw Normal View History

2022-12-06 11:48:10 +11:00
#version 450
#include "config.inc"
/**
* @license
* Copyright (c) 2011 NVIDIA Corporation. All rights reserved.
*
* TO THE MAXIMUM EXTENT PERMITTED BY APPLICABLE LAW, THIS SOFTWARE IS PROVIDED
* *AS IS* AND NVIDIA AND ITS SUPPLIERS DISCLAIM ALL WARRANTIES, EITHER EXPRESS
* OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, NONINFRINGEMENT,IMPLIED WARRANTIES OF
* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. IN NO EVENT SHALL NVIDIA
* OR ITS SUPPLIERS BE LIABLE FOR ANY DIRECT, SPECIAL, INCIDENTAL, INDIRECT, OR
* CONSEQUENTIAL DAMAGES WHATSOEVER (INCLUDING, WITHOUT LIMITATION, DAMAGES FOR LOSS
* OF BUSINESS PROFITS, BUSINESS INTERRUPTION, LOSS OF BUSINESS INFORMATION, OR ANY
* OTHER PECUNIARY LOSS) ARISING OUT OF THE USE OF OR INABILITY TO USE THIS SOFTWARE,
* EVEN IF NVIDIA HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES.
*/
/*
FXAA_PRESET - Choose compile-in knob preset 0-5.
------------------------------------------------------------------------------
FXAA_EDGE_THRESHOLD - The minimum amount of local contrast required
to apply algorithm.
1.0/3.0 - too little
1.0/4.0 - good start
1.0/8.0 - applies to more edges
1.0/16.0 - overkill
------------------------------------------------------------------------------
FXAA_EDGE_THRESHOLD_MIN - Trims the algorithm from processing darks.
Perf optimization.
1.0/32.0 - visible limit (smaller isn't visible)
1.0/16.0 - good compromise
1.0/12.0 - upper limit (seeing artifacts)
------------------------------------------------------------------------------
FXAA_SEARCH_STEPS - Maximum number of search steps for end of span.
------------------------------------------------------------------------------
FXAA_SEARCH_THRESHOLD - Controls when to stop searching.
1.0/4.0 - seems to be the best quality wise
------------------------------------------------------------------------------
FXAA_SUBPIX_TRIM - Controls sub-pixel aliasing removal.
1.0/2.0 - low removal
1.0/3.0 - medium removal
1.0/4.0 - default removal
1.0/8.0 - high removal
0.0 - complete removal
------------------------------------------------------------------------------
FXAA_SUBPIX_CAP - Insures fine detail is not completely removed.
This is important for the transition of sub-pixel detail,
like fences and wires.
3.0/4.0 - default (medium amount of filtering)
7.0/8.0 - high amount of filtering
1.0 - no capping of sub-pixel aliasing removal
*/
#ifndef FXAA_PRESET
#define FXAA_PRESET 4
#endif
#if (FXAA_PRESET == 6)
#define FXAA_EDGE_THRESHOLD (1.0/16.0)
#define FXAA_EDGE_THRESHOLD_MIN (1.0/12.0)
#define FXAA_SEARCH_STEPS 128
#define FXAA_SEARCH_THRESHOLD (1.0/16.0)
#define FXAA_SUBPIX_CAP (1.0)
#define FXAA_SUBPIX_TRIM (0.0)
#endif
#if (FXAA_PRESET == 3)
#define FXAA_EDGE_THRESHOLD (1.0/16.0)
#define FXAA_EDGE_THRESHOLD_MIN (1.0/16.0)
#define FXAA_SEARCH_STEPS 16
#define FXAA_SEARCH_THRESHOLD (1.0/4.0)
#define FXAA_SUBPIX_CAP (3.0/4.0)
#define FXAA_SUBPIX_TRIM (1.0/4.0)
#endif
#if (FXAA_PRESET == 4)
#define FXAA_EDGE_THRESHOLD (1.0/8.0)
#define FXAA_EDGE_THRESHOLD_MIN (1.0/24.0)
#define FXAA_SEARCH_STEPS 8
#define FXAA_SEARCH_THRESHOLD (1.0/4.0)
#define FXAA_SUBPIX_CAP (3.0/4.0)
#define FXAA_SUBPIX_TRIM (1.0/8.0)
#endif
#if (FXAA_PRESET == 5)
#define FXAA_EDGE_THRESHOLD (1.0/16.0)
#define FXAA_EDGE_THRESHOLD_MIN (1.0/12.0)
#define FXAA_SEARCH_STEPS 32
#define FXAA_SEARCH_THRESHOLD (1.0/4.0)
#define FXAA_SUBPIX_CAP (7.0/8.0)
#define FXAA_SUBPIX_TRIM (1.0/8.0)
#endif
#define FXAA_SUBPIX_TRIM_SCALE (1.0/(1.0 - FXAA_SUBPIX_TRIM))
// Return the luma, the estimation of luminance from rgb inputs.
// This approximates luma using one FMA instruction,
// skipping normalization and tossing out blue.
// FxaaLuma() will range 0.0 to 2.963210702.
float FxaaLuma(vec3 rgb) {
return rgb.y * (0.587/0.299) + rgb.x;
}
vec3 FxaaLerp3(vec3 a, vec3 b, float amountOfA) {
return (vec3(-amountOfA) * b) + ((a * vec3(amountOfA)) + b);
}
vec4 FxaaTexOff(sampler2D tex, vec2 pos, ivec2 off, vec2 rcpFrame) {
float x = pos.x + float(off.x) * rcpFrame.x;
float y = pos.y + float(off.y) * rcpFrame.y;
return textureLod(tex, vec2(x, y),0);
}
// pos is the output of FxaaVertexShader interpolated across screen.
// xy -> actual texture position {0.0 to 1.0}
// rcpFrame should be a uniform equal to {1.0/frameWidth, 1.0/frameHeight}
vec3 FxaaPixelShader(vec2 pos, sampler2D tex, vec2 rcpFrame)
{
vec3 rgbN = FxaaTexOff(tex, pos.xy, ivec2( 0,-1), rcpFrame).xyz;
vec3 rgbW = FxaaTexOff(tex, pos.xy, ivec2(-1, 0), rcpFrame).xyz;
vec3 rgbM = FxaaTexOff(tex, pos.xy, ivec2( 0, 0), rcpFrame).xyz;
vec3 rgbE = FxaaTexOff(tex, pos.xy, ivec2( 1, 0), rcpFrame).xyz;
vec3 rgbS = FxaaTexOff(tex, pos.xy, ivec2( 0, 1), rcpFrame).xyz;
float lumaN = FxaaLuma(rgbN);
float lumaW = FxaaLuma(rgbW);
float lumaM = FxaaLuma(rgbM);
float lumaE = FxaaLuma(rgbE);
float lumaS = FxaaLuma(rgbS);
float rangeMin = min(lumaM, min(min(lumaN, lumaW), min(lumaS, lumaE)));
float rangeMax = max(lumaM, max(max(lumaN, lumaW), max(lumaS, lumaE)));
float range = rangeMax - rangeMin;
if(range < max(FXAA_EDGE_THRESHOLD_MIN, rangeMax * FXAA_EDGE_THRESHOLD))
{
return rgbM;
}
vec3 rgbL = rgbN + rgbW + rgbM + rgbE + rgbS;
float lumaL = (lumaN + lumaW + lumaE + lumaS) * 0.25;
float rangeL = abs(lumaL - lumaM);
float blendL = max(0.0, (rangeL / range) - FXAA_SUBPIX_TRIM) * FXAA_SUBPIX_TRIM_SCALE;
blendL = min(FXAA_SUBPIX_CAP, blendL);
vec3 rgbNW = FxaaTexOff(tex, pos.xy, ivec2(-1,-1), rcpFrame).xyz;
vec3 rgbNE = FxaaTexOff(tex, pos.xy, ivec2( 1,-1), rcpFrame).xyz;
vec3 rgbSW = FxaaTexOff(tex, pos.xy, ivec2(-1, 1), rcpFrame).xyz;
vec3 rgbSE = FxaaTexOff(tex, pos.xy, ivec2( 1, 1), rcpFrame).xyz;
rgbL += (rgbNW + rgbNE + rgbSW + rgbSE);
rgbL *= vec3(1.0/9.0);
float lumaNW = FxaaLuma(rgbNW);
float lumaNE = FxaaLuma(rgbNE);
float lumaSW = FxaaLuma(rgbSW);
float lumaSE = FxaaLuma(rgbSE);
float edgeVert =
abs((0.25 * lumaNW) + (-0.5 * lumaN) + (0.25 * lumaNE)) +
abs((0.50 * lumaW ) + (-1.0 * lumaM) + (0.50 * lumaE )) +
abs((0.25 * lumaSW) + (-0.5 * lumaS) + (0.25 * lumaSE));
float edgeHorz =
abs((0.25 * lumaNW) + (-0.5 * lumaW) + (0.25 * lumaSW)) +
abs((0.50 * lumaN ) + (-1.0 * lumaM) + (0.50 * lumaS )) +
abs((0.25 * lumaNE) + (-0.5 * lumaE) + (0.25 * lumaSE));
bool horzSpan = edgeHorz >= edgeVert;
float lengthSign = horzSpan ? -rcpFrame.y : -rcpFrame.x;
if(!horzSpan)
{
lumaN = lumaW;
lumaS = lumaE;
}
float gradientN = abs(lumaN - lumaM);
float gradientS = abs(lumaS - lumaM);
lumaN = (lumaN + lumaM) * 0.5;
lumaS = (lumaS + lumaM) * 0.5;
if (gradientN < gradientS)
{
lumaN = lumaS;
lumaN = lumaS;
gradientN = gradientS;
lengthSign *= -1.0;
}
vec2 posN;
posN.x = pos.x + (horzSpan ? 0.0 : lengthSign * 0.5);
posN.y = pos.y + (horzSpan ? lengthSign * 0.5 : 0.0);
gradientN *= FXAA_SEARCH_THRESHOLD;
vec2 posP = posN;
vec2 offNP = horzSpan ? vec2(rcpFrame.x, 0.0) : vec2(0.0, rcpFrame.y);
float lumaEndN = lumaN;
float lumaEndP = lumaN;
bool doneN = false;
bool doneP = false;
posN += offNP * vec2(-1.0, -1.0);
posP += offNP * vec2( 1.0, 1.0);
for(int i = 0; i < FXAA_SEARCH_STEPS; i++) {
if(!doneN)
{
lumaEndN = FxaaLuma(textureLod(tex, posN.xy,0.0).xyz);
}
if(!doneP)
{
lumaEndP = FxaaLuma(textureLod(tex, posP.xy,0.0).xyz);
}
doneN = doneN || (abs(lumaEndN - lumaN) >= gradientN);
doneP = doneP || (abs(lumaEndP - lumaN) >= gradientN);
if(doneN && doneP)
{
break;
}
if(!doneN)
{
posN -= offNP;
}
if(!doneP)
{
posP += offNP;
}
}
float dstN = horzSpan ? pos.x - posN.x : pos.y - posN.y;
float dstP = horzSpan ? posP.x - pos.x : posP.y - pos.y;
bool directionN = dstN < dstP;
lumaEndN = directionN ? lumaEndN : lumaEndP;
if(((lumaM - lumaN) < 0.0) == ((lumaEndN - lumaN) < 0.0))
{
lengthSign = 0.0;
}
float spanLength = (dstP + dstN);
dstN = directionN ? dstN : dstP;
float subPixelOffset = (0.5 + (dstN * (-1.0/spanLength))) * lengthSign;
vec3 rgbF = textureLod(tex, vec2(
pos.x + (horzSpan ? 0.0 : subPixelOffset),
pos.y + (horzSpan ? subPixelOffset : 0.0)),0.0).xyz;
return FxaaLerp3(rgbL, rgbF, blendL);
}
#pragma stage vertex
layout(location = 0) in vec4 Position;
layout(location = 1) in vec2 TexCoord;
layout(location = 0) out vec2 vTexCoord;
2022-12-06 11:48:10 +11:00
void main()
{
gl_Position = global.MVP * Position;
vTexCoord = TexCoord;
}
#pragma stage fragment
layout(location = 0) in vec2 vTexCoord;
layout(location = 0) out vec4 FragColor;
2023-04-28 18:03:23 +10:00
layout(set = 0, binding = 2) uniform sampler2D flick_and_noise_pass;
2022-12-06 11:48:10 +11:00
void main() {
2022-12-06 11:48:10 +11:00
if ( DO_FXAA == 0.0 ) return;
//Here lies the blur modifier from ntsc pass to glow.
//I'm not expecting any performance hit, since the lookup should have been cached alreadyl
2023-04-28 18:03:23 +10:00
float pixel_alpha = texture(flick_and_noise_pass, vTexCoord).a;
2023-04-28 18:03:23 +10:00
FragColor = vec4(FxaaPixelShader(vTexCoord, flick_and_noise_pass, vec2(global.flick_and_noise_passSize.z, global.flick_and_noise_passSize.w)),pixel_alpha);
2022-12-06 11:48:10 +11:00
}