mirror of https://github.com/Xaymar/obs-StreamFX
560 lines
22 KiB
Plaintext
560 lines
22 KiB
Plaintext
// NVIDIA FXAA 3.11 by TIMOTHY LOTTES
|
|
// License: Public Domain
|
|
// Sourced from: https://gist.github.com/kosua20/0c506b81b3812ac900048059d2383126
|
|
//
|
|
// Adjusted for StreamFX by
|
|
// AUTOGENERATED COPYRIGHT HEADER START
|
|
// Copyright (C) 2021-2023 Michael Fabian 'Xaymar' Dirks <info@xaymar.com>
|
|
// AUTOGENERATED COPYRIGHT HEADER END
|
|
|
|
// ================================================================================ //
|
|
// Only adjust things above this line! Below here lie dragons, as well as weird symbols.
|
|
// ================================================================================ //
|
|
|
|
// Include the baseline filter data.
|
|
#define IS_FILTER
|
|
#include "../base.effect"
|
|
|
|
// Color conversion for RGBA->RGBL.
|
|
#include "../colorconversion_rgb_yuv.effect"
|
|
|
|
uniform int _000_QualityPreset<
|
|
bool visible = true;
|
|
bool automatic = false;
|
|
string name = "Quality Preset";
|
|
string field_type = "enum";
|
|
|
|
int enum_0 = 10;
|
|
string enum_0_name = "Medium Dither, Lowest Quality";
|
|
int enum_1 = 11;
|
|
string enum_1_name = "Medium Dither, Lower Quality";
|
|
int enum_2 = 12;
|
|
string enum_2_name = "Medium Dither, Low Quality";
|
|
int enum_3 = 13;
|
|
string enum_3_name = "Medium Dither, Normal Quality";
|
|
int enum_4 = 14;
|
|
string enum_4_name = "Medium Dither, Higher Quality";
|
|
int enum_5 = 15;
|
|
string enum_5_name = "Medium Dither, Highest Quality";
|
|
|
|
int enum_6 = 20;
|
|
string enum_6_name = "Low Dither, Lowest Quality";
|
|
int enum_7 = 21;
|
|
string enum_7_name = "Low Dither, Lower Quality";
|
|
int enum_8 = 22;
|
|
string enum_8_name = "Low Dither, Low Quality";
|
|
int enum_9 = 23;
|
|
string enum_9_name = "Low Dither, Below Medium Quality";
|
|
int enum_10 = 24;
|
|
string enum_10_name = "Low Dither, Medium Quality";
|
|
int enum_11 = 25;
|
|
string enum_11_name = "Low Dither, Above Medium Quality";
|
|
int enum_12 = 26;
|
|
string enum_12_name = "Low Dither, High Quality";
|
|
int enum_13 = 28;
|
|
string enum_13_name = "Low Dither, Higher Quality";
|
|
int enum_14 = 29;
|
|
string enum_14_name = "Low Dither, Highest Quality";
|
|
|
|
int enum_15 = 39;
|
|
string enum_15_name = "No Dither, Extreme Quality";
|
|
> = 23;
|
|
|
|
uniform float _100_SubpixelAliasingRemoval<
|
|
bool visible = true;
|
|
bool automatic = false;
|
|
string name = "Sub-Pixel Aliasing Removal";
|
|
string field_type = "slider";
|
|
|
|
float minimum = 0.0;
|
|
float maximum = 100.0;
|
|
float step = 0.01;
|
|
float scale = 0.01;
|
|
> = 75.0;
|
|
|
|
uniform float _200_EdgeThreshold<
|
|
bool visible = true;
|
|
bool automatic = false;
|
|
string name = "Edge Threshold";
|
|
string description = "Local contrast threshold to reach before applying the algorithm at all.\nMust be higher than 'Minimum Edge Threshold'.";
|
|
string field_type = "slider";
|
|
|
|
float minimum = 0.0;
|
|
float maximum = 100.00;
|
|
float step = 0.01;
|
|
float scale = 0.01;
|
|
> = 16.67;
|
|
|
|
uniform float _300_EdgeThresholdMin<
|
|
bool visible = true;
|
|
bool automatic = false;
|
|
string name = "Minimum Edge Threshold";
|
|
string description = "Minimum local contrast threshold to reach before applying the algorithm at all.\nMust be lower than 'Edge Threshold'.";
|
|
string field_type = "slider";
|
|
|
|
float minimum = 0.0;
|
|
float maximum = 100.0;
|
|
float step = 0.01;
|
|
float scale = 0.01;
|
|
> = 8.33;
|
|
|
|
// ======================================== //
|
|
// FXAA Functionality
|
|
// ======================================== //
|
|
// This is a bit of a mess, but it is necessary to get things to work. We're
|
|
// dealing with multiple unusual restrictions here, all of which are from
|
|
// choices made by libOBS and its renderers:
|
|
//
|
|
// - Due to OBS preprocessing the effect file, some things don't work like normal.
|
|
// - We are strictly limited to Shader Model 4.0, which makes this shader slower.
|
|
// - User experience is key, so some things were moved from preprocessor to uniform.
|
|
|
|
// API Interoperability
|
|
#define FxaaBool bool
|
|
#define FxaaFloat float
|
|
#define FxaaFloat2 float2
|
|
#define FxaaFloat3 float3
|
|
#define FxaaFloat4 float4
|
|
#define FxaaHalf FxaaFloat
|
|
#define FxaaHalf2 FxaaFloat2
|
|
#define FxaaHalf3 FxaaFloat3
|
|
#define FxaaHalf4 FxaaFloat4
|
|
#define FxaaInt2 int2
|
|
#define FxaaUInt uint
|
|
#define FxaaTex texture2d
|
|
#ifdef GS_DEVICE_OPENGL
|
|
#define FxaaSat(x) clamp(x, 0., 1.)
|
|
#define FxaaTexTop(t, p) textureLod(t, p, 0.0)
|
|
#define FxaaTexOff(t, p, o, r) textureLodOffset(t, p, 0.0, o)
|
|
#else
|
|
#define FxaaSat(x) saturate(x)
|
|
#define FxaaTexTop(t, p) t.SampleLevel(LinearClampSampler, p, 0.0)
|
|
#define FxaaTexOff(t, p, o, r) t.SampleLevel(LinearClampSampler, p, 0.0, o)
|
|
#endif
|
|
|
|
// Actual Functionality starts here.
|
|
FxaaUInt quality_stages() {
|
|
FxaaUInt stages = FxaaUInt(_000_QualityPreset) % 10u;
|
|
return 3u + stages;
|
|
}
|
|
|
|
FxaaFloat quality_value(in FxaaUInt stage) {
|
|
#ifdef GS_DEVICE_OPENGL
|
|
const FxaaFloat preset_00[12u] = FxaaFloat[12u]( 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f);
|
|
// 1.x
|
|
const FxaaFloat preset_10[12u] = FxaaFloat[12u]( 1.5f, 3.0f, 12.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f);
|
|
const FxaaFloat preset_11[12u] = FxaaFloat[12u]( 1.0f, 1.5f, 3.0f, 12.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f);
|
|
const FxaaFloat preset_12[12u] = FxaaFloat[12u]( 1.0f, 1.5f, 2.0f, 4.0f, 12.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f);
|
|
const FxaaFloat preset_13[12u] = FxaaFloat[12u]( 1.0f, 1.5f, 2.0f, 2.0f, 4.0f, 12.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f);
|
|
const FxaaFloat preset_14[12u] = FxaaFloat[12u]( 1.0f, 1.5f, 2.0f, 2.0f, 2.0f, 4.0f, 12.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f);
|
|
const FxaaFloat preset_15[12u] = FxaaFloat[12u]( 1.0f, 1.5f, 2.0f, 2.0f, 2.0f, 2.0f, 4.0f, 12.0f, 0.0f, 0.0f, 0.0f, 0.0f);
|
|
// 2.x
|
|
const FxaaFloat preset_20[12u] = FxaaFloat[12u]( 1.5f, 2.0f, 8.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f);
|
|
const FxaaFloat preset_21[12u] = FxaaFloat[12u]( 1.0f, 1.5f, 2.0f, 8.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f);
|
|
const FxaaFloat preset_22[12u] = FxaaFloat[12u]( 1.0f, 1.5f, 2.0f, 2.0f, 8.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f);
|
|
const FxaaFloat preset_23[12u] = FxaaFloat[12u]( 1.0f, 1.5f, 2.0f, 2.0f, 2.0f, 8.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f);
|
|
const FxaaFloat preset_24[12u] = FxaaFloat[12u]( 1.0f, 1.5f, 2.0f, 2.0f, 2.0f, 3.0f, 8.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f);
|
|
const FxaaFloat preset_25[12u] = FxaaFloat[12u]( 1.0f, 1.5f, 2.0f, 2.0f, 2.0f, 2.0f, 4.0f, 8.0f, 0.0f, 0.0f, 0.0f, 0.0f);
|
|
const FxaaFloat preset_26[12u] = FxaaFloat[12u]( 1.0f, 1.5f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 4.0f, 8.0f, 0.0f, 0.0f, 0.0f);
|
|
const FxaaFloat preset_27[12u] = FxaaFloat[12u]( 1.0f, 1.5f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 4.0f, 8.0f, 0.0f, 0.0f);
|
|
const FxaaFloat preset_28[12u] = FxaaFloat[12u]( 1.0f, 1.5f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 4.0f, 8.0f, 0.0f);
|
|
const FxaaFloat preset_29[12u] = FxaaFloat[12u]( 1.0f, 1.5f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 4.0f, 8.0f);
|
|
// 3.x
|
|
const FxaaFloat preset_39[12u] = FxaaFloat[12u]( 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.5f, 2.0f, 2.0f, 2.0f, 2.0f, 4.0f, 8.0f);
|
|
#else
|
|
const FxaaFloat preset_00[12u] = { 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f};
|
|
// 1.x
|
|
const FxaaFloat preset_10[12u] = { 1.5f, 3.0f, 12.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f};
|
|
const FxaaFloat preset_11[12u] = { 1.0f, 1.5f, 3.0f, 12.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f};
|
|
const FxaaFloat preset_12[12u] = { 1.0f, 1.5f, 2.0f, 4.0f, 12.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f};
|
|
const FxaaFloat preset_13[12u] = { 1.0f, 1.5f, 2.0f, 2.0f, 4.0f, 12.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f};
|
|
const FxaaFloat preset_14[12u] = { 1.0f, 1.5f, 2.0f, 2.0f, 2.0f, 4.0f, 12.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f};
|
|
const FxaaFloat preset_15[12u] = { 1.0f, 1.5f, 2.0f, 2.0f, 2.0f, 2.0f, 4.0f, 12.0f, 0.0f, 0.0f, 0.0f, 0.0f};
|
|
// 2.x
|
|
const FxaaFloat preset_20[12u] = { 1.5f, 2.0f, 8.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f};
|
|
const FxaaFloat preset_21[12u] = { 1.0f, 1.5f, 2.0f, 8.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f};
|
|
const FxaaFloat preset_22[12u] = { 1.0f, 1.5f, 2.0f, 2.0f, 8.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f};
|
|
const FxaaFloat preset_23[12u] = { 1.0f, 1.5f, 2.0f, 2.0f, 2.0f, 8.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f};
|
|
const FxaaFloat preset_24[12u] = { 1.0f, 1.5f, 2.0f, 2.0f, 2.0f, 3.0f, 8.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f};
|
|
const FxaaFloat preset_25[12u] = { 1.0f, 1.5f, 2.0f, 2.0f, 2.0f, 2.0f, 4.0f, 8.0f, 0.0f, 0.0f, 0.0f, 0.0f};
|
|
const FxaaFloat preset_26[12u] = { 1.0f, 1.5f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 4.0f, 8.0f, 0.0f, 0.0f, 0.0f};
|
|
const FxaaFloat preset_27[12u] = { 1.0f, 1.5f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 4.0f, 8.0f, 0.0f, 0.0f};
|
|
const FxaaFloat preset_28[12u] = { 1.0f, 1.5f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 4.0f, 8.0f, 0.0f};
|
|
const FxaaFloat preset_29[12u] = { 1.0f, 1.5f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 2.0f, 4.0f, 8.0f};
|
|
// 3.x
|
|
const FxaaFloat preset_39[12u] = { 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.5f, 2.0f, 2.0f, 2.0f, 2.0f, 4.0f, 8.0f};
|
|
#endif
|
|
|
|
switch (_000_QualityPreset) {
|
|
case 10: return preset_10[stage];
|
|
case 11: return preset_11[stage];
|
|
case 12: return preset_12[stage];
|
|
case 13: return preset_13[stage];
|
|
case 14: return preset_14[stage];
|
|
case 15: return preset_15[stage];
|
|
case 20: return preset_20[stage];
|
|
case 21: return preset_21[stage];
|
|
case 22: return preset_22[stage];
|
|
case 23: return preset_23[stage];
|
|
case 24: return preset_24[stage];
|
|
case 25: return preset_25[stage];
|
|
case 26: return preset_26[stage];
|
|
case 27: return preset_27[stage];
|
|
case 28: return preset_28[stage];
|
|
case 29: return preset_29[stage];
|
|
case 39: return preset_39[stage];
|
|
default: return preset_00[stage];
|
|
}
|
|
}
|
|
|
|
FxaaFloat FxaaLuma(in FxaaFloat4 rgba) {
|
|
// sRGB conversion breaks shit.
|
|
return RGBAtoYUVAf(rgba, RGB_YUV_709).r;
|
|
}
|
|
|
|
FxaaBool FxaaStep(in FxaaTex tex,
|
|
inout FxaaBool doneN, inout FxaaFloat2 posN, inout FxaaFloat lumaEndN,
|
|
inout FxaaBool doneP, inout FxaaFloat2 posP, inout FxaaFloat lumaEndP,
|
|
out FxaaBool doneNP, in FxaaFloat2 offNP, in FxaaFloat lumaNN, in FxaaFloat gradientScaled,
|
|
in FxaaFloat quality) {
|
|
if (!doneN) {
|
|
lumaEndN = FxaaLuma(FxaaTexTop(tex, posN)) - lumaNN * 0.5;
|
|
}
|
|
if (!doneP) {
|
|
lumaEndP = FxaaLuma(FxaaTexTop(tex, posP)) - lumaNN * 0.5;
|
|
}
|
|
doneN = abs(lumaEndN) >= gradientScaled;
|
|
doneP = abs(lumaEndP) >= gradientScaled;
|
|
if (!doneN) {
|
|
posN -= offNP * quality;
|
|
}
|
|
if (!doneP) {
|
|
posP += offNP * quality;
|
|
}
|
|
doneNP = (!doneN) || (!doneP);
|
|
return doneNP;
|
|
}
|
|
|
|
FxaaFloat4 FxaaPixelShader(
|
|
//
|
|
// Use noperspective interpolation here (turn off perspective interpolation).
|
|
// {xy} = center of pixel
|
|
FxaaFloat2 pos,
|
|
//
|
|
// Input color texture.
|
|
// {rgb_} = color in linear or perceptual color space
|
|
// if (FXAA_GREEN_AS_LUMA == 0)
|
|
// {___a} = luma in perceptual color space (not linear)
|
|
FxaaTex tex,
|
|
//
|
|
// Only used on FXAA Quality.
|
|
// This must be from a constant/uniform.
|
|
// {x_} = 1.0/screenWidthInPixels
|
|
// {_y} = 1.0/screenHeightInPixels
|
|
FxaaFloat2 fxaaQualityRcpFrame,
|
|
//
|
|
// Only used on FXAA Quality.
|
|
// This used to be the FXAA_QUALITY__SUBPIX define.
|
|
// It is here now to allow easier tuning.
|
|
// Choose the amount of sub-pixel aliasing removal.
|
|
// This can effect sharpness.
|
|
// 1.00 - upper limit (softer)
|
|
// 0.75 - default amount of filtering
|
|
// 0.50 - lower limit (sharper, less sub-pixel aliasing removal)
|
|
// 0.25 - almost off
|
|
// 0.00 - completely off
|
|
FxaaFloat fxaaQualitySubpix,
|
|
//
|
|
// Only used on FXAA Quality.
|
|
// This used to be the FXAA_QUALITY__EDGE_THRESHOLD define.
|
|
// It is here now to allow easier tuning.
|
|
// The minimum amount of local contrast required to apply algorithm.
|
|
// 0.333 - too little (faster)
|
|
// 0.250 - low quality
|
|
// 0.166 - default
|
|
// 0.125 - high quality
|
|
// 0.063 - overkill (slower)
|
|
FxaaFloat fxaaQualityEdgeThreshold,
|
|
//
|
|
// Only used on FXAA Quality.
|
|
// This used to be the FXAA_QUALITY__EDGE_THRESHOLD_MIN define.
|
|
// It is here now to allow easier tuning.
|
|
// Trims the algorithm from processing darks.
|
|
// 0.0833 - upper limit (default, the start of visible unfiltered edges)
|
|
// 0.0625 - high quality (faster)
|
|
// 0.0312 - visible limit (slower)
|
|
// Special notes when using FXAA_GREEN_AS_LUMA,
|
|
// Likely want to set this to zero.
|
|
// As colors that are mostly not-green
|
|
// will appear very dark in the green channel!
|
|
// Tune by looking at mostly non-green content,
|
|
// then start at zero and increase until aliasing is a problem.
|
|
FxaaFloat fxaaQualityEdgeThresholdMin
|
|
) {
|
|
// ======================================== // Similar to 3.11
|
|
FxaaFloat2 posM;
|
|
posM.x = pos.x;
|
|
posM.y = pos.y;
|
|
#ifdef FXAA_GATHER4_ALPHA
|
|
#ifdef FXAA_DISCARD
|
|
FxaaFloat4 rgbyM = FxaaTexTop(tex, posM);
|
|
#ifndef FXAA_GREEN_AS_LUMA
|
|
#define lumaM rgbyM.w
|
|
#else
|
|
#define lumaM rgbyM.y
|
|
#endif
|
|
#endif
|
|
#ifndef FXAA_GREEN_AS_LUMA
|
|
FxaaFloat4 luma4A = FxaaTexAlpha4(tex, posM);
|
|
FxaaFloat4 luma4B = FxaaTexOffAlpha4(tex, posM, FxaaInt2(-1, -1));
|
|
#else
|
|
FxaaFloat4 luma4A = FxaaTexGreen4(tex, posM);
|
|
FxaaFloat4 luma4B = FxaaTexOffGreen4(tex, posM, FxaaInt2(-1, -1));
|
|
#endif
|
|
#ifdef FXAA_DISCARD
|
|
#define lumaM luma4A.w
|
|
#endif
|
|
#define lumaE luma4A.z
|
|
#define lumaS luma4A.x
|
|
#define lumaSE luma4A.y
|
|
#define lumaNW luma4B.w
|
|
#define lumaN luma4B.z
|
|
#define lumaW luma4B.x
|
|
#else
|
|
FxaaFloat4 rgbyM = FxaaTexTop(tex, posM);
|
|
#ifndef FXAA_GREEN_AS_LUMA
|
|
#define lumaM rgbyM.w
|
|
#else
|
|
#define lumaM rgbyM.y
|
|
#endif
|
|
FxaaFloat lumaS = FxaaLuma(FxaaTexOff(tex, posM, FxaaInt2( 0, 1), fxaaQualityRcpFrame.xy));
|
|
FxaaFloat lumaE = FxaaLuma(FxaaTexOff(tex, posM, FxaaInt2( 1, 0), fxaaQualityRcpFrame.xy));
|
|
FxaaFloat lumaN = FxaaLuma(FxaaTexOff(tex, posM, FxaaInt2( 0,-1), fxaaQualityRcpFrame.xy));
|
|
FxaaFloat lumaW = FxaaLuma(FxaaTexOff(tex, posM, FxaaInt2(-1, 0), fxaaQualityRcpFrame.xy));
|
|
#endif
|
|
|
|
// ======================================== // Matches 3.11
|
|
FxaaFloat maxSM = max(lumaS, lumaM);
|
|
FxaaFloat minSM = min(lumaS, lumaM);
|
|
FxaaFloat maxESM = max(lumaE, maxSM);
|
|
FxaaFloat minESM = min(lumaE, minSM);
|
|
FxaaFloat maxWN = max(lumaN, lumaW);
|
|
FxaaFloat minWN = min(lumaN, lumaW);
|
|
FxaaFloat rangeMax = max(maxWN, maxESM);
|
|
FxaaFloat rangeMin = min(minWN, minESM);
|
|
FxaaFloat rangeMaxScaled = rangeMax * fxaaQualityEdgeThreshold;
|
|
FxaaFloat range = rangeMax - rangeMin;
|
|
FxaaFloat rangeMaxClamped = max(fxaaQualityEdgeThresholdMin, rangeMaxScaled);
|
|
FxaaBool earlyExit = range < rangeMaxClamped;
|
|
|
|
// ======================================== // Similar to 3.11
|
|
if(earlyExit) {
|
|
#ifdef FXAA_DISCARD
|
|
FxaaDiscard;
|
|
#else
|
|
return rgbyM;
|
|
#endif
|
|
}
|
|
|
|
// ======================================== // Similar to 3.11
|
|
#ifndef FXAA_GATHER4_ALPHA
|
|
FxaaFloat lumaNW = FxaaLuma(FxaaTexOff(tex, posM, FxaaInt2(-1,-1), fxaaQualityRcpFrame.xy));
|
|
FxaaFloat lumaSE = FxaaLuma(FxaaTexOff(tex, posM, FxaaInt2( 1, 1), fxaaQualityRcpFrame.xy));
|
|
FxaaFloat lumaNE = FxaaLuma(FxaaTexOff(tex, posM, FxaaInt2( 1,-1), fxaaQualityRcpFrame.xy));
|
|
FxaaFloat lumaSW = FxaaLuma(FxaaTexOff(tex, posM, FxaaInt2(-1, 1), fxaaQualityRcpFrame.xy));
|
|
#else
|
|
FxaaFloat lumaNE = FxaaLuma(FxaaTexOff(tex, posM, FxaaInt2(1, -1), fxaaQualityRcpFrame.xy));
|
|
FxaaFloat lumaSW = FxaaLuma(FxaaTexOff(tex, posM, FxaaInt2(-1, 1), fxaaQualityRcpFrame.xy));
|
|
#endif
|
|
|
|
// ======================================== // Matches 3.11
|
|
FxaaFloat lumaNS = lumaN + lumaS;
|
|
FxaaFloat lumaWE = lumaW + lumaE;
|
|
FxaaFloat subpixRcpRange = 1.0/range;
|
|
FxaaFloat subpixNSWE = lumaNS + lumaWE;
|
|
FxaaFloat edgeHorz1 = (-2.0 * lumaM) + lumaNS;
|
|
FxaaFloat edgeVert1 = (-2.0 * lumaM) + lumaWE;
|
|
|
|
// ======================================== // Matches 3.11
|
|
FxaaFloat lumaNESE = lumaNE + lumaSE;
|
|
FxaaFloat lumaNWNE = lumaNW + lumaNE;
|
|
FxaaFloat edgeHorz2 = (-2.0 * lumaE) + lumaNESE;
|
|
FxaaFloat edgeVert2 = (-2.0 * lumaN) + lumaNWNE;
|
|
|
|
// ======================================== // Matches 3.11
|
|
FxaaFloat lumaNWSW = lumaNW + lumaSW;
|
|
FxaaFloat lumaSWSE = lumaSW + lumaSE;
|
|
FxaaFloat edgeHorz4 = (abs(edgeHorz1) * 2.0) + abs(edgeHorz2);
|
|
FxaaFloat edgeVert4 = (abs(edgeVert1) * 2.0) + abs(edgeVert2);
|
|
FxaaFloat edgeHorz3 = (-2.0 * lumaW) + lumaNWSW;
|
|
FxaaFloat edgeVert3 = (-2.0 * lumaS) + lumaSWSE;
|
|
FxaaFloat edgeHorz = abs(edgeHorz3) + edgeHorz4;
|
|
FxaaFloat edgeVert = abs(edgeVert3) + edgeVert4;
|
|
|
|
// ======================================== // Matches 3.11
|
|
FxaaFloat subpixNWSWNESE = lumaNWSW + lumaNESE;
|
|
FxaaFloat lengthSign = fxaaQualityRcpFrame.x;
|
|
FxaaBool horzSpan = edgeHorz >= edgeVert;
|
|
FxaaFloat subpixA = subpixNSWE * 2.0 + subpixNWSWNESE;
|
|
|
|
// ======================================== // Matches 3.11
|
|
if(!horzSpan) lumaN = lumaW;
|
|
if(!horzSpan) lumaS = lumaE;
|
|
if(horzSpan) lengthSign = fxaaQualityRcpFrame.y;
|
|
FxaaFloat subpixB = (subpixA * (1.0/12.0)) - lumaM;
|
|
|
|
// ======================================== // Matches 3.11
|
|
FxaaFloat gradientN = lumaN - lumaM;
|
|
FxaaFloat gradientS = lumaS - lumaM;
|
|
FxaaFloat lumaNN = lumaN + lumaM;
|
|
FxaaFloat lumaSS = lumaS + lumaM;
|
|
FxaaBool pairN = abs(gradientN) >= abs(gradientS);
|
|
FxaaFloat gradient = max(abs(gradientN), abs(gradientS));
|
|
if(pairN) lengthSign = -lengthSign;
|
|
FxaaFloat subpixC = FxaaSat(abs(subpixB) * subpixRcpRange);
|
|
|
|
// ======================================== // Matches 3.11
|
|
FxaaFloat2 posB;
|
|
posB.x = posM.x;
|
|
posB.y = posM.y;
|
|
FxaaFloat2 offNP;
|
|
offNP.x = (!horzSpan) ? 0.0 : fxaaQualityRcpFrame.x;
|
|
offNP.y = ( horzSpan) ? 0.0 : fxaaQualityRcpFrame.y;
|
|
if(!horzSpan) posB.x += lengthSign * 0.5;
|
|
if( horzSpan) posB.y += lengthSign * 0.5;
|
|
|
|
// ======================================== // Custom (but still copied from 3.11)
|
|
FxaaFloat subpixD = ((-2.0)*subpixC) + 3.0;
|
|
FxaaFloat subpixE = subpixC * subpixC;
|
|
|
|
// ======================================== // Matches 3.11
|
|
if(!pairN) lumaNN = lumaSS;
|
|
FxaaFloat gradientScaled = gradient * 1.0/4.0;
|
|
FxaaFloat lumaMM = lumaM - lumaNN * 0.5;
|
|
FxaaFloat subpixF = subpixD * subpixE;
|
|
FxaaBool lumaMLTZero = lumaMM < 0.0;
|
|
|
|
// ======================================== // Custom
|
|
uint stages = quality_stages();
|
|
FxaaBool doneN = false;
|
|
FxaaFloat2 posN = posB;
|
|
FxaaFloat lumaEndN = 0;
|
|
FxaaBool doneP = false;
|
|
FxaaFloat2 posP = posB;
|
|
FxaaFloat lumaEndP = 0;
|
|
FxaaBool doneNP = false;
|
|
|
|
FxaaStep(tex, doneN, posN, lumaEndN, doneP, posP, lumaEndP, doneNP, offNP, lumaNN, gradientScaled, quality_value(0u));
|
|
if ((stages >= 1u) && doneNP) {
|
|
FxaaStep(tex, doneN, posN, lumaEndN, doneP, posP, lumaEndP, doneNP, offNP, lumaNN, gradientScaled, quality_value(1u));
|
|
if ((stages >= 2u) && doneNP) {
|
|
FxaaStep(tex, doneN, posN, lumaEndN, doneP, posP, lumaEndP, doneNP, offNP, lumaNN, gradientScaled, quality_value(2u));
|
|
if ((stages >= 3u) && doneNP) {
|
|
FxaaStep(tex, doneN, posN, lumaEndN, doneP, posP, lumaEndP, doneNP, offNP, lumaNN, gradientScaled, quality_value(3u));
|
|
if ((stages >= 4u) && doneNP) {
|
|
FxaaStep(tex, doneN, posN, lumaEndN, doneP, posP, lumaEndP, doneNP, offNP, lumaNN, gradientScaled, quality_value(4u));
|
|
if ((stages >= 5u) && doneNP) {
|
|
FxaaStep(tex, doneN, posN, lumaEndN, doneP, posP, lumaEndP, doneNP, offNP, lumaNN, gradientScaled, quality_value(5u));
|
|
if ((stages >= 6u) && doneNP) {
|
|
FxaaStep(tex, doneN, posN, lumaEndN, doneP, posP, lumaEndP, doneNP, offNP, lumaNN, gradientScaled, quality_value(6u));
|
|
if ((stages >= 7u) && doneNP) {
|
|
FxaaStep(tex, doneN, posN, lumaEndN, doneP, posP, lumaEndP, doneNP, offNP, lumaNN, gradientScaled, quality_value(7u));
|
|
if ((stages >= 8u) && doneNP) {
|
|
FxaaStep(tex, doneN, posN, lumaEndN, doneP, posP, lumaEndP, doneNP, offNP, lumaNN, gradientScaled, quality_value(8u));
|
|
if ((stages >= 9u) && doneNP) {
|
|
FxaaStep(tex, doneN, posN, lumaEndN, doneP, posP, lumaEndP, doneNP, offNP, lumaNN, gradientScaled, quality_value(9u));
|
|
if ((stages >= 10u) && doneNP) {
|
|
FxaaStep(tex, doneN, posN, lumaEndN, doneP, posP, lumaEndP, doneNP, offNP, lumaNN, gradientScaled, quality_value(10u));
|
|
if ((stages >= 11u) && doneNP) {
|
|
FxaaStep(tex, doneN, posN, lumaEndN, doneP, posP, lumaEndP, doneNP, offNP, lumaNN, gradientScaled, quality_value(11u));
|
|
// Present in FXAA 3.11, but not actually defined anywhere.
|
|
// if ((stages >= 12u) && doneNP) {
|
|
// FxaaStep(tex, doneN, posN, lumaEndN, doneP, posP, lumaEndP, doneNP, offNP, lumaNN, gradientScaled, quality_value(12));
|
|
// }
|
|
}
|
|
}
|
|
}
|
|
}
|
|
}
|
|
}
|
|
}
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
// ======================================== // Matches 3.11
|
|
FxaaFloat dstN = posM.x - posN.x;
|
|
FxaaFloat dstP = posP.x - posM.x;
|
|
if(!horzSpan) dstN = posM.y - posN.y;
|
|
if(!horzSpan) dstP = posP.y - posM.y;
|
|
|
|
// ======================================== // Matches 3.11
|
|
FxaaBool goodSpanN = (lumaEndN < 0.0) != lumaMLTZero;
|
|
FxaaFloat spanLength = (dstP + dstN);
|
|
FxaaBool goodSpanP = (lumaEndP < 0.0) != lumaMLTZero;
|
|
FxaaFloat spanLengthRcp = 1.0/spanLength;
|
|
|
|
// ======================================== // Matches 3.11
|
|
FxaaBool directionN = dstN < dstP;
|
|
FxaaFloat dst = min(dstN, dstP);
|
|
FxaaBool goodSpan = directionN ? goodSpanN : goodSpanP;
|
|
FxaaFloat subpixG = subpixF * subpixF;
|
|
FxaaFloat pixelOffset = (dst * (-spanLengthRcp)) + 0.5;
|
|
FxaaFloat subpixH = subpixG * fxaaQualitySubpix;
|
|
|
|
// ======================================== // Similar to 3.11
|
|
FxaaFloat pixelOffsetGood = goodSpan ? pixelOffset : 0.0;
|
|
FxaaFloat pixelOffsetSubpix = max(pixelOffsetGood, subpixH);
|
|
if(!horzSpan) posM.x += pixelOffsetSubpix * lengthSign;
|
|
if( horzSpan) posM.y += pixelOffsetSubpix * lengthSign;
|
|
#ifdef FXAA_DISCARD
|
|
return FxaaTexTop(tex, posM);
|
|
#else
|
|
return FxaaFloat4(FxaaTexTop(tex, posM).xyz, lumaM);
|
|
#endif
|
|
}
|
|
|
|
// ======================================== //
|
|
// Technique: Process
|
|
// ======================================== //
|
|
// Applies FXAA
|
|
|
|
FxaaFloat4 ProcessPixelShader(VertexInformation vtx) : TARGET {
|
|
return FxaaPixelShader(
|
|
vtx.texcoord0.xy,
|
|
InputA,
|
|
ViewSize.zw,
|
|
_100_SubpixelAliasingRemoval,
|
|
_200_EdgeThreshold,
|
|
_300_EdgeThresholdMin);
|
|
}
|
|
|
|
technique Process {
|
|
pass
|
|
{
|
|
vertex_shader = DefaultVertexShader(vtx);
|
|
pixel_shader = ProcessPixelShader(vtx);
|
|
};
|
|
};
|
|
|
|
// ======================================== //
|
|
// Technique: Draw
|
|
// ======================================== //
|
|
// Does both. Not supported in StreamFX.
|
|
|
|
/*
|
|
technique Draw {
|
|
pass
|
|
{
|
|
vertex_shader = DefaultVertexShader(vtx);
|
|
pixel_shader = PreprocessPixelShader(vtx);
|
|
};
|
|
pass
|
|
{
|
|
vertex_shader = DefaultVertexShader(vtx);
|
|
pixel_shader = ProcessPixelShader(vtx);
|
|
};
|
|
};
|
|
*/
|