From 958afaa7a7bfb1217bbffee3fe562f21901277ac Mon Sep 17 00:00:00 2001 From: Rye Mutt Date: Tue, 27 Aug 2024 22:47:32 -0400 Subject: Integrate SMAA and rework post process chain for better visual quality Add SMAA buffer generation passes Add quality levels for both FXAA and SMAA Separate gamma correction and tonemapping for effects that require linear-but-tonemapped inputs Move application of noise to final render pass to screen to avoid damaging other post process effects --- .../app_settings/shaders/class1/deferred/CASF.glsl | 6 +- .../app_settings/shaders/class1/deferred/SMAA.glsl | 1463 ++++++++++++++++++++ .../shaders/class1/deferred/SMAABlendWeightsF.glsl | 57 + .../shaders/class1/deferred/SMAABlendWeightsV.glsl | 51 + .../shaders/class1/deferred/SMAAEdgeDetectF.glsl | 59 + .../shaders/class1/deferred/SMAAEdgeDetectV.glsl | 45 + .../class1/deferred/SMAANeighborhoodBlendF.glsl | 63 + .../class1/deferred/SMAANeighborhoodBlendV.glsl | 47 + .../shaders/class1/deferred/fxaaF.glsl | 2 +- .../class1/deferred/postDeferredGammaCorrect.glsl | 145 +- .../class1/deferred/postDeferredNoDoFF.glsl | 47 + .../class1/deferred/postDeferredTonemap.glsl | 150 ++ 12 files changed, 1989 insertions(+), 146 deletions(-) create mode 100644 indra/newview/app_settings/shaders/class1/deferred/SMAA.glsl create mode 100644 indra/newview/app_settings/shaders/class1/deferred/SMAABlendWeightsF.glsl create mode 100644 indra/newview/app_settings/shaders/class1/deferred/SMAABlendWeightsV.glsl create mode 100644 indra/newview/app_settings/shaders/class1/deferred/SMAAEdgeDetectF.glsl create mode 100644 indra/newview/app_settings/shaders/class1/deferred/SMAAEdgeDetectV.glsl create mode 100644 indra/newview/app_settings/shaders/class1/deferred/SMAANeighborhoodBlendF.glsl create mode 100644 indra/newview/app_settings/shaders/class1/deferred/SMAANeighborhoodBlendV.glsl create mode 100644 indra/newview/app_settings/shaders/class1/deferred/postDeferredTonemap.glsl (limited to 'indra/newview/app_settings/shaders/class1/deferred') diff --git a/indra/newview/app_settings/shaders/class1/deferred/CASF.glsl b/indra/newview/app_settings/shaders/class1/deferred/CASF.glsl index 96d08058cf..e80c59b39f 100644 --- a/indra/newview/app_settings/shaders/class1/deferred/CASF.glsl +++ b/indra/newview/app_settings/shaders/class1/deferred/CASF.glsl @@ -2067,9 +2067,6 @@ A_STATIC void CasSetup( AF3 CasLoad(ASU2 p) { return texelFetch(diffuseRect, p, 0).rgb; } void CasInput(inout AF1 r,inout AF1 g,inout AF1 b) { - r = AFromSrgbF1(r); - g = AFromSrgbF1(g); - b = AFromSrgbF1(b); } //------------------------------------------------------------------------------------------------------------------------------ @@ -2553,6 +2550,7 @@ void main() vec4 diff = vec4(0.f); uvec2 point = uvec2(vary_fragcoord * out_screen_res.xy); CasFilter(diff.r, diff.g, diff.b, point, cas_param_0, cas_param_1, true); - frag_color = vec4(linear_to_srgb(diff.rgb), 1.0); + diff.a = texture(diffuseRect, vary_fragcoord).a; + frag_color = diff; } #endif diff --git a/indra/newview/app_settings/shaders/class1/deferred/SMAA.glsl b/indra/newview/app_settings/shaders/class1/deferred/SMAA.glsl new file mode 100644 index 0000000000..fdb77cce6e --- /dev/null +++ b/indra/newview/app_settings/shaders/class1/deferred/SMAA.glsl @@ -0,0 +1,1463 @@ +/** + * @file SMAA.glsl + * + * $LicenseInfo:firstyear=2024&license=viewerlgpl$ + * Second Life Viewer Source Code + * Copyright (C) 2024, Linden Research, Inc. + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; + * version 2.1 of the License only. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + * + * Linden Research, Inc., 945 Battery Street, San Francisco, CA 94111 USA + * $/LicenseInfo$ + */ + +#extension GL_ARB_texture_rectangle : enable +#extension GL_ARB_shader_texture_lod : enable +#extension GL_EXT_gpu_shader4 : enable + +/*[EXTRA_CODE_HERE]*/ + +#ifdef VERTEX_SHADER + #define SMAA_INCLUDE_VS 1 + #define SMAA_INCLUDE_PS 0 +#else + #define SMAA_INCLUDE_VS 0 + #define SMAA_INCLUDE_PS 1 +#endif + +uniform vec4 SMAA_RT_METRICS; + +/** + * Copyright (C) 2013 Jorge Jimenez (jorge@iryoku.com) + * Copyright (C) 2013 Jose I. Echevarria (joseignacioechevarria@gmail.com) + * Copyright (C) 2013 Belen Masia (bmasia@unizar.es) + * Copyright (C) 2013 Fernando Navarro (fernandn@microsoft.com) + * Copyright (C) 2013 Diego Gutierrez (diegog@unizar.es) + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * this software and associated documentation files (the "Software"), to deal in + * the Software without restriction, including without limitation the rights to + * use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies + * of the Software, and to permit persons to whom the Software is furnished to + * do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. As clarification, there + * is no requirement that the copyright notice and permission be included in + * binary distributions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + + +/** + * _______ ___ ___ ___ ___ + * / || \/ | / \ / \ + * | (---- | \ / | / ^ \ / ^ \ + * \ \ | |\/| | / /_\ \ / /_\ \ + * ----) | | | | | / _____ \ / _____ \ + * |_______/ |__| |__| /__/ \__\ /__/ \__\ + * + * E N H A N C E D + * S U B P I X E L M O R P H O L O G I C A L A N T I A L I A S I N G + * + * http://www.iryoku.com/smaa/ + * + * Hi, welcome aboard! + * + * Here you'll find instructions to get the shader up and running as fast as + * possible. + * + * IMPORTANTE NOTICE: when updating, remember to update both this file and the + * precomputed textures! They may change from version to version. + * + * The shader has three passes, chained together as follows: + * + * |input|------------------� + * v | + * [ SMAA*EdgeDetection ] | + * v | + * |edgesTex| | + * v | + * [ SMAABlendingWeightCalculation ] | + * v | + * |blendTex| | + * v | + * [ SMAANeighborhoodBlending ] <------� + * v + * |output| + * + * Note that each [pass] has its own vertex and pixel shader. Remember to use + * oversized triangles instead of quads to avoid overshading along the + * diagonal. + * + * You've three edge detection methods to choose from: luma, color or depth. + * They represent different quality/performance and anti-aliasing/sharpness + * tradeoffs, so our recommendation is for you to choose the one that best + * suits your particular scenario: + * + * - Depth edge detection is usually the fastest but it may miss some edges. + * + * - Luma edge detection is usually more expensive than depth edge detection, + * but catches visible edges that depth edge detection can miss. + * + * - Color edge detection is usually the most expensive one but catches + * chroma-only edges. + * + * For quickstarters: just use luma edge detection. + * + * The general advice is to not rush the integration process and ensure each + * step is done correctly (don't try to integrate SMAA T2x with predicated edge + * detection from the start!). Ok then, let's go! + * + * 1. The first step is to create two RGBA temporal render targets for holding + * |edgesTex| and |blendTex|. + * + * In DX10 or DX11, you can use a RG render target for the edges texture. + * In the case of NVIDIA GPUs, using RG render targets seems to actually be + * slower. + * + * On the Xbox 360, you can use the same render target for resolving both + * |edgesTex| and |blendTex|, as they aren't needed simultaneously. + * + * 2. Both temporal render targets |edgesTex| and |blendTex| must be cleared + * each frame. Do not forget to clear the alpha channel! + * + * 3. The next step is loading the two supporting precalculated textures, + * 'areaTex' and 'searchTex'. You'll find them in the 'Textures' folder as + * C++ headers, and also as regular DDS files. They'll be needed for the + * 'SMAABlendingWeightCalculation' pass. + * + * If you use the C++ headers, be sure to load them in the format specified + * inside of them. + * + * You can also compress 'areaTex' and 'searchTex' using BC5 and BC4 + * respectively, if you have that option in your content processor pipeline. + * When compressing then, you get a non-perceptible quality decrease, and a + * marginal performance increase. + * + * 4. All samplers must be set to linear filtering and clamp. + * + * After you get the technique working, remember that 64-bit inputs have + * half-rate linear filtering on GCN. + * + * If SMAA is applied to 64-bit color buffers, switching to point filtering + * when accesing them will increase the performance. Search for + * 'SMAASamplePoint' to see which textures may benefit from point + * filtering, and where (which is basically the color input in the edge + * detection and resolve passes). + * + * 5. All texture reads and buffer writes must be non-sRGB, with the exception + * of the input read and the output write in + * 'SMAANeighborhoodBlending' (and only in this pass!). If sRGB reads in + * this last pass are not possible, the technique will work anyway, but + * will perform antialiasing in gamma space. + * + * IMPORTANT: for best results the input read for the color/luma edge + * detection should *NOT* be sRGB. + * + * 6. Before including SMAA.h you'll have to setup the render target metrics, + * the target and any optional configuration defines. Optionally you can + * use a preset. + * + * You have the following targets available: + * SMAA_HLSL_3 + * SMAA_HLSL_4 + * SMAA_HLSL_4_1 + * SMAA_GLSL_2 * + * SMAA_GLSL_3 * + * SMAA_GLSL_4 * + * + * * (See SMAA_INCLUDE_VS and SMAA_INCLUDE_PS below). + * + * And four presets: + * SMAA_PRESET_LOW (%60 of the quality) + * SMAA_PRESET_MEDIUM (%80 of the quality) + * SMAA_PRESET_HIGH (%95 of the quality) + * SMAA_PRESET_ULTRA (%99 of the quality) + * + * For example: + * #define SMAA_RT_METRICS float4(1.0 / 1280.0, 1.0 / 720.0, 1280.0, 720.0) + * #define SMAA_HLSL_4 + * #define SMAA_PRESET_HIGH + * #include "SMAA.h" + * + * Note that SMAA_RT_METRICS doesn't need to be a macro, it can be a + * uniform variable. The code is designed to minimize the impact of not + * using a constant value, but it is still better to hardcode it. + * + * Depending on how you encoded 'areaTex' and 'searchTex', you may have to + * add (and customize) the following defines before including SMAA.h: + * #define SMAA_AREATEX_SELECT(sample) sample.rg + * #define SMAA_SEARCHTEX_SELECT(sample) sample.r + * + * If your engine is already using porting macros, you can define + * SMAA_CUSTOM_SL, and define the porting functions by yourself. + * + * 7. Then, you'll have to setup the passes as indicated in the scheme above. + * You can take a look into SMAA.fx, to see how we did it for our demo. + * Checkout the function wrappers, you may want to copy-paste them! + * + * 8. It's recommended to validate the produced |edgesTex| and |blendTex|. + * You can use a screenshot from your engine to compare the |edgesTex| + * and |blendTex| produced inside of the engine with the results obtained + * with the reference demo. + * + * 9. After you get the last pass to work, it's time to optimize. You'll have + * to initialize a stencil buffer in the first pass (discard is already in + * the code), then mask execution by using it the second pass. The last + * pass should be executed in all pixels. + * + * + * After this point you can choose to enable predicated thresholding, + * temporal supersampling and motion blur integration: + * + * a) If you want to use predicated thresholding, take a look into + * SMAA_PREDICATION; you'll need to pass an extra texture in the edge + * detection pass. + * + * b) If you want to enable temporal supersampling (SMAA T2x): + * + * 1. The first step is to render using subpixel jitters. I won't go into + * detail, but it's as simple as moving each vertex position in the + * vertex shader, you can check how we do it in our DX10 demo. + * + * 2. Then, you must setup the temporal resolve. You may want to take a look + * into SMAAResolve for resolving 2x modes. After you get it working, you'll + * probably see ghosting everywhere. But fear not, you can enable the + * CryENGINE temporal reprojection by setting the SMAA_REPROJECTION macro. + * Check out SMAA_DECODE_VELOCITY if your velocity buffer is encoded. + * + * 3. The next step is to apply SMAA to each subpixel jittered frame, just as + * done for 1x. + * + * 4. At this point you should already have something usable, but for best + * results the proper area textures must be set depending on current jitter. + * For this, the parameter 'subsampleIndices' of + * 'SMAABlendingWeightCalculationPS' must be set as follows, for our T2x + * mode: + * + * @SUBSAMPLE_INDICES + * + * | S# | Camera Jitter | subsampleIndices | + * +----+------------------+---------------------+ + * | 0 | ( 0.25, -0.25) | float4(1, 1, 1, 0) | + * | 1 | (-0.25, 0.25) | float4(2, 2, 2, 0) | + * + * These jitter positions assume a bottom-to-top y axis. S# stands for the + * sample number. + * + * More information about temporal supersampling here: + * http://iryoku.com/aacourse/downloads/13-Anti-Aliasing-Methods-in-CryENGINE-3.pdf + * + * c) If you want to enable spatial multisampling (SMAA S2x): + * + * 1. The scene must be rendered using MSAA 2x. The MSAA 2x buffer must be + * created with: + * - DX10: see below (*) + * - DX10.1: D3D10_STANDARD_MULTISAMPLE_PATTERN or + * - DX11: D3D11_STANDARD_MULTISAMPLE_PATTERN + * + * This allows to ensure that the subsample order matches the table in + * @SUBSAMPLE_INDICES. + * + * (*) In the case of DX10, we refer the reader to: + * - SMAA::detectMSAAOrder and + * - SMAA::msaaReorder + * + * These functions allow to match the standard multisample patterns by + * detecting the subsample order for a specific GPU, and reordering + * them appropriately. + * + * 2. A shader must be run to output each subsample into a separate buffer + * (DX10 is required). You can use SMAASeparate for this purpose, or just do + * it in an existing pass (for example, in the tone mapping pass, which has + * the advantage of feeding tone mapped subsamples to SMAA, which will yield + * better results). + * + * 3. The full SMAA 1x pipeline must be run for each separated buffer, storing + * the results in the final buffer. The second run should alpha blend with + * the existing final buffer using a blending factor of 0.5. + * 'subsampleIndices' must be adjusted as in the SMAA T2x case (see point + * b). + * + * d) If you want to enable temporal supersampling on top of SMAA S2x + * (which actually is SMAA 4x): + * + * 1. SMAA 4x consists on temporally jittering SMAA S2x, so the first step is + * to calculate SMAA S2x for current frame. In this case, 'subsampleIndices' + * must be set as follows: + * + * | F# | S# | Camera Jitter | Net Jitter | subsampleIndices | + * +----+----+--------------------+-------------------+----------------------+ + * | 0 | 0 | ( 0.125, 0.125) | ( 0.375, -0.125) | float4(5, 3, 1, 3) | + * | 0 | 1 | ( 0.125, 0.125) | (-0.125, 0.375) | float4(4, 6, 2, 3) | + * +----+----+--------------------+-------------------+----------------------+ + * | 1 | 2 | (-0.125, -0.125) | ( 0.125, -0.375) | float4(3, 5, 1, 4) | + * | 1 | 3 | (-0.125, -0.125) | (-0.375, 0.125) | float4(6, 4, 2, 4) | + * + * These jitter positions assume a bottom-to-top y axis. F# stands for the + * frame number. S# stands for the sample number. + * + * 2. After calculating SMAA S2x for current frame (with the new subsample + * indices), previous frame must be reprojected as in SMAA T2x mode (see + * point b). + * + * e) If motion blur is used, you may want to do the edge detection pass + * together with motion blur. This has two advantages: + * + * 1. Pixels under heavy motion can be omitted from the edge detection process. + * For these pixels we can just store "no edge", as motion blur will take + * care of them. + * 2. The center pixel tap is reused. + * + * Note that in this case depth testing should be used instead of stenciling, + * as we have to write all the pixels in the motion blur pass. + * + * That's it! + */ + +//----------------------------------------------------------------------------- +// SMAA Presets + +/** + * Note that if you use one of these presets, the following configuration + * macros will be ignored if set in the "Configurable Defines" section. + */ + +#if defined(SMAA_PRESET_LOW) +#define SMAA_THRESHOLD 0.15 +#define SMAA_MAX_SEARCH_STEPS 4 +#define SMAA_DISABLE_DIAG_DETECTION +#define SMAA_DISABLE_CORNER_DETECTION +#elif defined(SMAA_PRESET_MEDIUM) +#define SMAA_THRESHOLD 0.1 +#define SMAA_MAX_SEARCH_STEPS 8 +#define SMAA_DISABLE_DIAG_DETECTION +#define SMAA_DISABLE_CORNER_DETECTION +#elif defined(SMAA_PRESET_HIGH) +#define SMAA_THRESHOLD 0.1 +#define SMAA_MAX_SEARCH_STEPS 16 +#define SMAA_MAX_SEARCH_STEPS_DIAG 8 +#define SMAA_CORNER_ROUNDING 25 +#elif defined(SMAA_PRESET_ULTRA) +#define SMAA_THRESHOLD 0.05 +#define SMAA_MAX_SEARCH_STEPS 32 +#define SMAA_MAX_SEARCH_STEPS_DIAG 16 +#define SMAA_CORNER_ROUNDING 25 +#endif + +//----------------------------------------------------------------------------- +// Configurable Defines + +/** + * SMAA_THRESHOLD specifies the threshold or sensitivity to edges. + * Lowering this value you will be able to detect more edges at the expense of + * performance. + * + * Range: [0, 0.5] + * 0.1 is a reasonable value, and allows to catch most visible edges. + * 0.05 is a rather overkill value, that allows to catch 'em all. + * + * If temporal supersampling is used, 0.2 could be a reasonable value, as low + * contrast edges are properly filtered by just 2x. + */ +#ifndef SMAA_THRESHOLD +#define SMAA_THRESHOLD 0.1 +#endif + +/** + * SMAA_DEPTH_THRESHOLD specifies the threshold for depth edge detection. + * + * Range: depends on the depth range of the scene. + */ +#ifndef SMAA_DEPTH_THRESHOLD +#define SMAA_DEPTH_THRESHOLD (0.1 * SMAA_THRESHOLD) +#endif + +/** + * SMAA_MAX_SEARCH_STEPS specifies the maximum steps performed in the + * horizontal/vertical pattern searches, at each side of the pixel. + * + * In number of pixels, it's actually the double. So the maximum line length + * perfectly handled by, for example 16, is 64 (by perfectly, we meant that + * longer lines won't look as good, but still antialiased). + * + * Range: [0, 112] + */ +#ifndef SMAA_MAX_SEARCH_STEPS +#define SMAA_MAX_SEARCH_STEPS 16 +#endif + +/** + * SMAA_MAX_SEARCH_STEPS_DIAG specifies the maximum steps performed in the + * diagonal pattern searches, at each side of the pixel. In this case we jump + * one pixel at time, instead of two. + * + * Range: [0, 20] + * + * On high-end machines it is cheap (between a 0.8x and 0.9x slower for 16 + * steps), but it can have a significant impact on older machines. + * + * Define SMAA_DISABLE_DIAG_DETECTION to disable diagonal processing. + */ +#ifndef SMAA_MAX_SEARCH_STEPS_DIAG +#define SMAA_MAX_SEARCH_STEPS_DIAG 8 +#endif + +/** + * SMAA_CORNER_ROUNDING specifies how much sharp corners will be rounded. + * + * Range: [0, 100] + * + * Define SMAA_DISABLE_CORNER_DETECTION to disable corner processing. + */ +#ifndef SMAA_CORNER_ROUNDING +#define SMAA_CORNER_ROUNDING 25 +#endif + +/** + * If there is an neighbor edge that has SMAA_LOCAL_CONTRAST_FACTOR times + * bigger contrast than current edge, current edge will be discarded. + * + * This allows to eliminate spurious crossing edges, and is based on the fact + * that, if there is too much contrast in a direction, that will hide + * perceptually contrast in the other neighbors. + */ +#ifndef SMAA_LOCAL_CONTRAST_ADAPTATION_FACTOR +#define SMAA_LOCAL_CONTRAST_ADAPTATION_FACTOR 2.0 +#endif + +/** + * Predicated thresholding allows to better preserve texture details and to + * improve performance, by decreasing the number of detected edges using an + * additional buffer like the light accumulation buffer, object ids or even the + * depth buffer (the depth buffer usage may be limited to indoor or short range + * scenes). + * + * It locally decreases the luma or color threshold if an edge is found in an + * additional buffer (so the global threshold can be higher). + * + * This method was developed by Playstation EDGE MLAA team, and used in + * Killzone 3, by using the light accumulation buffer. More information here: + * http://iryoku.com/aacourse/downloads/06-MLAA-on-PS3.pptx + */ +#ifndef SMAA_PREDICATION +#define SMAA_PREDICATION 0 +#endif + +/** + * Threshold to be used in the additional predication buffer. + * + * Range: depends on the input, so you'll have to find the magic number that + * works for you. + */ +#ifndef SMAA_PREDICATION_THRESHOLD +#define SMAA_PREDICATION_THRESHOLD 0.01 +#endif + +/** + * How much to scale the global threshold used for luma or color edge + * detection when using predication. + * + * Range: [1, 5] + */ +#ifndef SMAA_PREDICATION_SCALE +#define SMAA_PREDICATION_SCALE 2.0 +#endif + +/** + * How much to locally decrease the threshold. + * + * Range: [0, 1] + */ +#ifndef SMAA_PREDICATION_STRENGTH +#define SMAA_PREDICATION_STRENGTH 0.4 +#endif + +/** + * Temporal reprojection allows to remove ghosting artifacts when using + * temporal supersampling. We use the CryEngine 3 method which also introduces + * velocity weighting. This feature is of extreme importance for totally + * removing ghosting. More information here: + * http://iryoku.com/aacourse/downloads/13-Anti-Aliasing-Methods-in-CryENGINE-3.pdf + * + * Note that you'll need to setup a velocity buffer for enabling reprojection. + * For static geometry, saving the previous depth buffer is a viable + * alternative. + */ +#ifndef SMAA_REPROJECTION +#define SMAA_REPROJECTION 0 +#endif + +/** + * SMAA_REPROJECTION_WEIGHT_SCALE controls the velocity weighting. It allows to + * remove ghosting trails behind the moving object, which are not removed by + * just using reprojection. Using low values will exhibit ghosting, while using + * high values will disable temporal supersampling under motion. + * + * Behind the scenes, velocity weighting removes temporal supersampling when + * the velocity of the subsamples differs (meaning they are different objects). + * + * Range: [0, 80] + */ +#ifndef SMAA_REPROJECTION_WEIGHT_SCALE +#define SMAA_REPROJECTION_WEIGHT_SCALE 30.0 +#endif + +/** + * On some compilers, discard and texture cannot be used in vertex shaders. Thus, they need + * to be compiled separately. + */ +#ifndef SMAA_INCLUDE_VS +#define SMAA_INCLUDE_VS 1 +#endif +#ifndef SMAA_INCLUDE_PS +#define SMAA_INCLUDE_PS 1 +#endif + +//----------------------------------------------------------------------------- +// Texture Access Defines + +#ifndef SMAA_AREATEX_SELECT +#if defined(SMAA_HLSL_3) +#define SMAA_AREATEX_SELECT(sample) sample.ra +#else +#define SMAA_AREATEX_SELECT(sample) sample.rg +#endif +#endif + +#ifndef SMAA_SEARCHTEX_SELECT +#define SMAA_SEARCHTEX_SELECT(sample) sample.r +#endif + +#ifndef SMAA_DECODE_VELOCITY +#define SMAA_DECODE_VELOCITY(sample) sample.rg +#endif + +//----------------------------------------------------------------------------- +// Non-Configurable Defines + +#define SMAA_AREATEX_MAX_DISTANCE 16 +#define SMAA_AREATEX_MAX_DISTANCE_DIAG 20 +#define SMAA_AREATEX_PIXEL_SIZE (1.0 / float2(160.0, 560.0)) +#define SMAA_AREATEX_SUBTEX_SIZE (1.0 / 7.0) +#define SMAA_SEARCHTEX_SIZE float2(66.0, 33.0) +#define SMAA_SEARCHTEX_PACKED_SIZE float2(64.0, 16.0) +#define SMAA_CORNER_ROUNDING_NORM (float(SMAA_CORNER_ROUNDING) / 100.0) + +//----------------------------------------------------------------------------- +// Porting Functions + +#if defined(SMAA_HLSL_3) +#ifndef SMAA_FLIP_Y +#define SMAA_FLIP_Y 0 +#endif // SMAA_FLIP_Y +#define SMAATexture2D(tex) sampler2D tex +#define SMAATexturePass2D(tex) tex +#define SMAASampleLevelZero(tex, coord) tex2Dlod(tex, float4(coord, 0.0, 0.0)) +#define SMAASampleLevelZeroPoint(tex, coord) tex2Dlod(tex, float4(coord, 0.0, 0.0)) +#define SMAASampleLevelZeroOffset(tex, coord, offset) tex2Dlod(tex, float4(coord + offset * SMAA_RT_METRICS.xy, 0.0, 0.0)) +#define SMAASample(tex, coord) tex2D(tex, coord) +#define SMAASamplePoint(tex, coord) tex2D(tex, coord) +#define SMAASampleOffset(tex, coord, offset) tex2D(tex, coord + offset * SMAA_RT_METRICS.xy) +#define SMAA_FLATTEN [flatten] +#define SMAA_BRANCH [branch] +#endif +#if defined(SMAA_HLSL_4) || defined(SMAA_HLSL_4_1) +#ifndef SMAA_FLIP_Y +#define SMAA_FLIP_Y 0 +#endif // SMAA_FLIP_Y +SamplerState LinearSampler { Filter = MIN_MAG_LINEAR_MIP_POINT; AddressU = Clamp; AddressV = Clamp; }; +SamplerState PointSampler { Filter = MIN_MAG_MIP_POINT; AddressU = Clamp; AddressV = Clamp; }; +#define SMAATexture2D(tex) Texture2D tex +#define SMAATexturePass2D(tex) tex +#define SMAASampleLevelZero(tex, coord) tex.SampleLevel(LinearSampler, coord, 0) +#define SMAASampleLevelZeroPoint(tex, coord) tex.SampleLevel(PointSampler, coord, 0) +#define SMAASampleLevelZeroOffset(tex, coord, offset) tex.SampleLevel(LinearSampler, coord, 0, offset) +#define SMAASample(tex, coord) tex.Sample(LinearSampler, coord) +#define SMAASamplePoint(tex, coord) tex.Sample(PointSampler, coord) +#define SMAASampleOffset(tex, coord, offset) tex.Sample(LinearSampler, coord, offset) +#define SMAA_FLATTEN [flatten] +#define SMAA_BRANCH [branch] +#define SMAATexture2DMS2(tex) Texture2DMS tex +#define SMAALoad(tex, pos, sample) tex.Load(pos, sample) +#if defined(SMAA_HLSL_4_1) +#define SMAAGather(tex, coord) tex.Gather(LinearSampler, coord, 0) +#endif +#endif + +#if defined(SMAA_GLSL_2) || defined(SMAA_GLSL_3) || defined(SMAA_GLSL_4) +#ifndef SMAA_FLIP_Y +#define SMAA_FLIP_Y 1 +#endif // SMAA_FLIP_Y + +#define SMAATexture2D(tex) sampler2D tex +#define SMAATexturePass2D(tex) tex +#if defined(SMAA_GLSL_2) +#define SMAASampleLevelZero(tex, coord) texture2DLod(tex, coord, 0.0) +#define SMAASampleLevelZeroPoint(tex, coord) texture2DLod(tex, coord, 0.0) +#define SMAASampleLevelZeroOffset(tex, coord, offset) texture2DLodOffset(tex, coord, 0.0, offset) +#define SMAASample(tex, coord) texture2D(tex, coord) +#define SMAASamplePoint(tex, coord) texture2D(tex, coord) +#define SMAASampleOffset(tex, coord, offset) texture2D(tex, coord, offset) +#else +#define SMAASampleLevelZero(tex, coord) textureLod(tex, coord, 0.0) +#define SMAASampleLevelZeroPoint(tex, coord) textureLod(tex, coord, 0.0) +#define SMAASampleLevelZeroOffset(tex, coord, offset) textureLodOffset(tex, coord, 0.0, offset) +#define SMAASample(tex, coord) texture(tex, coord) +#define SMAASamplePoint(tex, coord) texture(tex, coord) +#define SMAASampleOffset(tex, coord, offset) textureOffset(tex, coord, offset) +#endif +#define SMAA_FLATTEN +#define SMAA_BRANCH +#define lerp(a, b, t) mix(a, b, t) +#define saturate(a) clamp(a, 0.0, 1.0) +#if defined(SMAA_GLSL_4) +#define mad(a, b, c) fma(a, b, c) +#define SMAAGather(tex, coord) textureGather(tex, coord) +#else +#define mad(a, b, c) (a * b + c) +#endif +#if defined(SMAA_GLSL_3) || defined(SMAA_GLSL_4) +#define SMAATexture2DMS2(tex) sampler2DMS tex +#define SMAALoad(tex, pos, sample) texelFetch(tex, pos, sample) +#endif +#define float2 vec2 +#define float3 vec3 +#define float4 vec4 +#define int2 ivec2 +#define int3 ivec3 +#define int4 ivec4 +#define bool2 bvec2 +#define bool3 bvec3 +#define bool4 bvec4 +#endif + +#if !defined(SMAA_HLSL_3) && !defined(SMAA_HLSL_4) && !defined(SMAA_HLSL_4_1) && !defined(SMAA_GLSL_2) && !defined(SMAA_GLSL_3) && !defined(SMAA_GLSL_4) && !defined(SMAA_CUSTOM_SL) +#error you must define the shading language: SMAA_HLSL_*, SMAA_GLSL_* or SMAA_CUSTOM_SL +#endif + + +#if SMAA_FLIP_Y + +#define API_V_DIR(v) -(v) +#define API_V_COORD(v) (1.0 - v) +#define API_V_BELOW(v1, v2) v1 < v2 +#define API_V_ABOVE(v1, v2) v1 > v2 + +#else // VULKAN_FLIP + +#define API_V_DIR(v) v +#define API_V_COORD(v) v +#define API_V_BELOW(v1, v2) v1 > v2 +#define API_V_ABOVE(v1, v2) v1 < v2 + +#endif // VULKAN_FLIP + + +//----------------------------------------------------------------------------- +// Misc functions + +#if SMAA_INCLUDE_PS +/** + * Gathers current pixel, and the top-left neighbors. + */ +float3 SMAAGatherNeighbours(float2 texcoord, + float4 offset[3], + SMAATexture2D(tex)) { + #ifdef SMAAGather + + #if SMAA_FLIP_Y + return SMAAGather(tex, texcoord + SMAA_RT_METRICS.xy * float2(-0.5, 0.5)).zwy; + #else // SMAA_FLIP_Y + return SMAAGather(tex, texcoord + SMAA_RT_METRICS.xy * float2(-0.5, -0.5)).grb; + #endif // SMAA_FLIP_Y + + #else // SMAAGather + float P = SMAASamplePoint(tex, texcoord).r; + float Pleft = SMAASamplePoint(tex, offset[0].xy).r; + float Ptop = SMAASamplePoint(tex, offset[0].zw).r; + return float3(P, Pleft, Ptop); + #endif +} + +/** + * Adjusts the threshold by means of predication. + */ +float2 SMAACalculatePredicatedThreshold(float2 texcoord, + float4 offset[3], + SMAATexture2D(predicationTex)) { + float3 neighbours = SMAAGatherNeighbours(texcoord, offset, SMAATexturePass2D(predicationTex)); + float2 delta = abs(neighbours.xx - neighbours.yz); + float2 edges = step(SMAA_PREDICATION_THRESHOLD, delta); + return SMAA_PREDICATION_SCALE * SMAA_THRESHOLD * (1.0 - SMAA_PREDICATION_STRENGTH * edges); +} + +#endif // SMAA_INCLUDE_PS + +/** + * Conditional move: + */ +void SMAAMovc(bool2 cond, inout float2 variable, float2 value) { + SMAA_FLATTEN if (cond.x) variable.x = value.x; + SMAA_FLATTEN if (cond.y) variable.y = value.y; +} + +void SMAAMovc(bool4 cond, inout float4 variable, float4 value) { + SMAAMovc(cond.xy, variable.xy, value.xy); + SMAAMovc(cond.zw, variable.zw, value.zw); +} + + +#if SMAA_INCLUDE_VS +//----------------------------------------------------------------------------- +// Vertex Shaders + +/** + * Edge Detection Vertex Shader + */ +void SMAAEdgeDetectionVS(float2 texcoord, + out float4 offset[3]) { + offset[0] = mad(SMAA_RT_METRICS.xyxy, float4(-1.0, 0.0, 0.0, API_V_DIR(-1.0)), texcoord.xyxy); + offset[1] = mad(SMAA_RT_METRICS.xyxy, float4( 1.0, 0.0, 0.0, API_V_DIR(1.0)), texcoord.xyxy); + offset[2] = mad(SMAA_RT_METRICS.xyxy, float4(-2.0, 0.0, 0.0, API_V_DIR(-2.0)), texcoord.xyxy); +} + +/** + * Blend Weight Calculation Vertex Shader + */ +void SMAABlendingWeightCalculationVS(float2 texcoord, + out float2 pixcoord, + out float4 offset[3]) { + pixcoord = texcoord * SMAA_RT_METRICS.zw; + + // We will use these offsets for the searches later on (see @PSEUDO_GATHER4): + offset[0] = mad(SMAA_RT_METRICS.xyxy, float4(-0.25, API_V_DIR(-0.125), 1.25, API_V_DIR(-0.125)), texcoord.xyxy); + offset[1] = mad(SMAA_RT_METRICS.xyxy, float4(-0.125, API_V_DIR(-0.25), -0.125, API_V_DIR(1.25)), texcoord.xyxy); + + // And these for the searches, they indicate the ends of the loops: + offset[2] = mad(SMAA_RT_METRICS.xxyy, + float4(-2.0, 2.0, API_V_DIR(-2.0), API_V_DIR(2.0)) * float(SMAA_MAX_SEARCH_STEPS), + float4(offset[0].xz, offset[1].yw)); +} + +/** + * Neighborhood Blending Vertex Shader + */ +void SMAANeighborhoodBlendingVS(float2 texcoord, + out float4 offset) { + offset = mad(SMAA_RT_METRICS.xyxy, float4( 1.0, 0.0, 0.0, API_V_DIR(1.0)), texcoord.xyxy); +} +#endif // SMAA_INCLUDE_VS + +#if SMAA_INCLUDE_PS +//----------------------------------------------------------------------------- +// Edge Detection Pixel Shaders (First Pass) + +/** + * Luma Edge Detection + * + * IMPORTANT NOTICE: luma edge detection requires gamma-corrected colors, and + * thus 'colorTex' should be a non-sRGB texture. + */ +float2 SMAALumaEdgeDetectionPS(float2 texcoord, + float4 offset[3], + SMAATexture2D(colorTex) + #if SMAA_PREDICATION + , SMAATexture2D(predicationTex) + #endif + ) { + // Calculate the threshold: + #if SMAA_PREDICATION + float2 threshold = SMAACalculatePredicatedThreshold(texcoord, offset, SMAATexturePass2D(predicationTex)); + #else + float2 threshold = float2(SMAA_THRESHOLD, SMAA_THRESHOLD); + #endif + + // Calculate lumas: + float3 weights = float3(0.2126, 0.7152, 0.0722); + float L = dot(SMAASamplePoint(colorTex, texcoord).rgb, weights); + + float Lleft = dot(SMAASamplePoint(colorTex, offset[0].xy).rgb, weights); + float Ltop = dot(SMAASamplePoint(colorTex, offset[0].zw).rgb, weights); + + // We do the usual threshold: + float4 delta; + delta.xy = abs(L - float2(Lleft, Ltop)); + float2 edges = step(threshold, delta.xy); + + // Then discard if there is no edge: + if (dot(edges, float2(1.0, 1.0)) == 0.0) + discard; + + // Calculate right and bottom deltas: + float Lright = dot(SMAASamplePoint(colorTex, offset[1].xy).rgb, weights); + float Lbottom = dot(SMAASamplePoint(colorTex, offset[1].zw).rgb, weights); + delta.zw = abs(L - float2(Lright, Lbottom)); + + // Calculate the maximum delta in the direct neighborhood: + float2 maxDelta = max(delta.xy, delta.zw); + + // Calculate left-left and top-top deltas: + float Lleftleft = dot(SMAASamplePoint(colorTex, offset[2].xy).rgb, weights); + float Ltoptop = dot(SMAASamplePoint(colorTex, offset[2].zw).rgb, weights); + delta.zw = abs(float2(Lleft, Ltop) - float2(Lleftleft, Ltoptop)); + + // Calculate the final maximum delta: + maxDelta = max(maxDelta.xy, delta.zw); + float finalDelta = max(maxDelta.x, maxDelta.y); + + // Local contrast adaptation: + edges.xy *= step(finalDelta, SMAA_LOCAL_CONTRAST_ADAPTATION_FACTOR * delta.xy); + + return edges; +} + +/** + * Color Edge Detection + * + * IMPORTANT NOTICE: color edge detection requires gamma-corrected colors, and + * thus 'colorTex' should be a non-sRGB texture. + */ +float2 SMAAColorEdgeDetectionPS(float2 texcoord, + float4 offset[3], + SMAATexture2D(colorTex) + #if SMAA_PREDICATION + , SMAATexture2D(predicationTex) + #endif + ) { + // Calculate the threshold: + #if SMAA_PREDICATION + float2 threshold = SMAACalculatePredicatedThreshold(texcoord, offset, predicationTex); + #else + float2 threshold = float2(SMAA_THRESHOLD, SMAA_THRESHOLD); + #endif + + // Calculate color deltas: + float4 delta; + float3 C = SMAASamplePoint(colorTex, texcoord).rgb; + + float3 Cleft = SMAASamplePoint(colorTex, offset[0].xy).rgb; + float3 t = abs(C - Cleft); + delta.x = max(max(t.r, t.g), t.b); + + float3 Ctop = SMAASamplePoint(colorTex, offset[0].zw).rgb; + t = abs(C - Ctop); + delta.y = max(max(t.r, t.g), t.b); + + // We do the usual threshold: + float2 edges = step(threshold, delta.xy); + + // Then discard if there is no edge: + if (dot(edges, float2(1.0, 1.0)) == 0.0) + discard; + + // Calculate right and bottom deltas: + float3 Cright = SMAASamplePoint(colorTex, offset[1].xy).rgb; + t = abs(C - Cright); + delta.z = max(max(t.r, t.g), t.b); + + float3 Cbottom = SMAASamplePoint(colorTex, offset[1].zw).rgb; + t = abs(C - Cbottom); + delta.w = max(max(t.r, t.g), t.b); + + // Calculate the maximum delta in the direct neighborhood: + float2 maxDelta = max(delta.xy, delta.zw); + + // Calculate left-left and top-top deltas: + float3 Cleftleft = SMAASamplePoint(colorTex, offset[2].xy).rgb; + t = abs(C - Cleftleft); + delta.z = max(max(t.r, t.g), t.b); + + float3 Ctoptop = SMAASamplePoint(colorTex, offset[2].zw).rgb; + t = abs(C - Ctoptop); + delta.w = max(max(t.r, t.g), t.b); + + // Calculate the final maximum delta: + maxDelta = max(maxDelta.xy, delta.zw); + float finalDelta = max(maxDelta.x, maxDelta.y); + + // Local contrast adaptation: + edges.xy *= step(finalDelta, SMAA_LOCAL_CONTRAST_ADAPTATION_FACTOR * delta.xy); + + return edges; +} + +/** + * Depth Edge Detection + */ +float2 SMAADepthEdgeDetectionPS(float2 texcoord, + float4 offset[3], + SMAATexture2D(depthTex)) { + float3 neighbours = SMAAGatherNeighbours(texcoord, offset, SMAATexturePass2D(depthTex)); + float2 delta = abs(neighbours.xx - float2(neighbours.y, neighbours.z)); + float2 edges = step(SMAA_DEPTH_THRESHOLD, delta); + + if (dot(edges, float2(1.0, 1.0)) == 0.0) + discard; + + return edges; +} + +//----------------------------------------------------------------------------- +// Diagonal Search Functions + +#if !defined(SMAA_DISABLE_DIAG_DETECTION) + +/** + * Allows to decode two binary values from a bilinear-filtered access. + */ +float2 SMAADecodeDiagBilinearAccess(float2 e) { + // Bilinear access for fetching 'e' have a 0.25 offset, and we are + // interested in the R and G edges: + // + // +---G---+-------+ + // | x o R x | + // +-------+-------+ + // + // Then, if one of these edge is enabled: + // Red: (0.75 * X + 0.25 * 1) => 0.25 or 1.0 + // Green: (0.75 * 1 + 0.25 * X) => 0.75 or 1.0 + // + // This function will unpack the values (mad + mul + round): + // wolframalpha.com: round(x * abs(5 * x - 5 * 0.75)) plot 0 to 1 + e.r = e.r * abs(5.0 * e.r - 5.0 * 0.75); + return round(e); +} + +float4 SMAADecodeDiagBilinearAccess(float4 e) { + e.rb = e.rb * abs(5.0 * e.rb - 5.0 * 0.75); + return round(e); +} + +/** + * These functions allows to perform diagonal pattern searches. + */ +float2 SMAASearchDiag1(SMAATexture2D(edgesTex), float2 texcoord, float2 dir, out float2 e) { + dir.y = API_V_DIR(dir.y); + float4 coord = float4(texcoord, -1.0, 1.0); + float3 t = float3(SMAA_RT_METRICS.xy, 1.0); + while (coord.z < float(SMAA_MAX_SEARCH_STEPS_DIAG - 1) && + coord.w > 0.9) { + coord.xyz = mad(t, float3(dir, 1.0), coord.xyz); + e = SMAASampleLevelZero(edgesTex, coord.xy).rg; + coord.w = dot(e, float2(0.5, 0.5)); + } + return coord.zw; +} + +float2 SMAASearchDiag2(SMAATexture2D(edgesTex), float2 texcoord, float2 dir, out float2 e) { + dir.y = API_V_DIR(dir.y); + float4 coord = float4(texcoord, -1.0, 1.0); + coord.x += 0.25 * SMAA_RT_METRICS.x; // See @SearchDiag2Optimization + float3 t = float3(SMAA_RT_METRICS.xy, 1.0); + while (coord.z < float(SMAA_MAX_SEARCH_STEPS_DIAG - 1) && + coord.w > 0.9) { + coord.xyz = mad(t, float3(dir, 1.0), coord.xyz); + + // @SearchDiag2Optimization + // Fetch both edges at once using bilinear filtering: + e = SMAASampleLevelZero(edgesTex, coord.xy).rg; + e = SMAADecodeDiagBilinearAccess(e); + + // Non-optimized version: + // e.g = SMAASampleLevelZero(edgesTex, coord.xy).g; + // e.r = SMAASampleLevelZeroOffset(edgesTex, coord.xy, int2(1, 0)).r; + + coord.w = dot(e, float2(0.5, 0.5)); + } + return coord.zw; +} + +/** + * Similar to SMAAArea, this calculates the area corresponding to a certain + * diagonal distance and crossing edges 'e'. + */ +float2 SMAAAreaDiag(SMAATexture2D(areaTex), float2 dist, float2 e, float offset) { + float2 texcoord = mad(float2(SMAA_AREATEX_MAX_DISTANCE_DIAG, SMAA_AREATEX_MAX_DISTANCE_DIAG), e, dist); + + // We do a scale and bias for mapping to texel space: + texcoord = mad(SMAA_AREATEX_PIXEL_SIZE, texcoord, 0.5 * SMAA_AREATEX_PIXEL_SIZE); + + // Diagonal areas are on the second half of the texture: + texcoord.x += 0.5; + + // Move to proper place, according to the subpixel offset: + texcoord.y += SMAA_AREATEX_SUBTEX_SIZE * offset; + + texcoord.y = API_V_COORD(texcoord.y); + + // Do it! + return SMAA_AREATEX_SELECT(SMAASampleLevelZero(areaTex, texcoord)); +} + +/** + * This searches for diagonal patterns and returns the corresponding weights. + */ +float2 SMAACalculateDiagWeights(SMAATexture2D(edgesTex), SMAATexture2D(areaTex), float2 texcoord, float2 e, float4 subsampleIndices) { + float2 weights = float2(0.0, 0.0); + + // Search for the line ends: + float4 d; + float2 end; + if (e.r > 0.0) { + d.xz = SMAASearchDiag1(SMAATexturePass2D(edgesTex), texcoord, float2(-1.0, 1.0), end); + d.x += float(end.y > 0.9); + } else + d.xz = float2(0.0, 0.0); + d.yw = SMAASearchDiag1(SMAATexturePass2D(edgesTex), texcoord, float2(1.0, -1.0), end); + + SMAA_BRANCH + if (d.x + d.y > 2.0) { // d.x + d.y + 1 > 3 + // Fetch the crossing edges: + float4 coords = mad(float4(-d.x + 0.25, API_V_DIR(d.x), d.y, API_V_DIR(-d.y - 0.25)), SMAA_RT_METRICS.xyxy, texcoord.xyxy); + float4 c; + c.xy = SMAASampleLevelZeroOffset(edgesTex, coords.xy, int2(-1, 0)).rg; + c.zw = SMAASampleLevelZeroOffset(edgesTex, coords.zw, int2( 1, 0)).rg; + c.yxwz = SMAADecodeDiagBilinearAccess(c.xyzw); + + // Non-optimized version: + // float4 coords = mad(float4(-d.x, d.x, d.y, -d.y), SMAA_RT_METRICS.xyxy, texcoord.xyxy); + // float4 c; + // c.x = SMAASampleLevelZeroOffset(edgesTex, coords.xy, int2(-1, 0)).g; + // c.y = SMAASampleLevelZeroOffset(edgesTex, coords.xy, int2( 0, 0)).r; + // c.z = SMAASampleLevelZeroOffset(edgesTex, coords.zw, int2( 1, 0)).g; + // c.w = SMAASampleLevelZeroOffset(edgesTex, coords.zw, int2( 1, -1)).r; + + // Merge crossing edges at each side into a single value: + float2 cc = mad(float2(2.0, 2.0), c.xz, c.yw); + + // Remove the crossing edge if we didn't found the end of the line: + SMAAMovc(bool2(step(0.9, d.zw)), cc, float2(0.0, 0.0)); + + // Fetch the areas for this line: + weights += SMAAAreaDiag(SMAATexturePass2D(areaTex), d.xy, cc, subsampleIndices.z); + } + + // Search for the line ends: + d.xz = SMAASearchDiag2(SMAATexturePass2D(edgesTex), texcoord, float2(-1.0, -1.0), end); + if (SMAASampleLevelZeroOffset(edgesTex, texcoord, int2(1, 0)).r > 0.0) { + d.yw = SMAASearchDiag2(SMAATexturePass2D(edgesTex), texcoord, float2(1.0, 1.0), end); + d.y += float(end.y > 0.9); + } else + d.yw = float2(0.0, 0.0); + + SMAA_BRANCH + if (d.x + d.y > 2.0) { // d.x + d.y + 1 > 3 + // Fetch the crossing edges: + float4 coords = mad(float4(-d.x, API_V_DIR(-d.x), d.y, API_V_DIR(d.y)), SMAA_RT_METRICS.xyxy, texcoord.xyxy); + float4 c; + c.x = SMAASampleLevelZeroOffset(edgesTex, coords.xy, int2(-1, 0)).g; + c.y = SMAASampleLevelZeroOffset(edgesTex, coords.xy, int2( 0, API_V_DIR(-1))).r; + c.zw = SMAASampleLevelZeroOffset(edgesTex, coords.zw, int2( 1, 0)).gr; + float2 cc = mad(float2(2.0, 2.0), c.xz, c.yw); + + // Remove the crossing edge if we didn't found the end of the line: + SMAAMovc(bool2(step(0.9, d.zw)), cc, float2(0.0, 0.0)); + + // Fetch the areas for this line: + weights += SMAAAreaDiag(SMAATexturePass2D(areaTex), d.xy, cc, subsampleIndices.w).gr; + } + + return weights; +} +#endif + +//----------------------------------------------------------------------------- +// Horizontal/Vertical Search Functions + +/** + * This allows to determine how much length should we add in the last step + * of the searches. It takes the bilinearly interpolated edge (see + * @PSEUDO_GATHER4), and adds 0, 1 or 2, depending on which edges and + * crossing edges are active. + */ +float SMAASearchLength(SMAATexture2D(searchTex), float2 e, float offset) { + // The texture is flipped vertically, with left and right cases taking half + // of the space horizontally: + float2 scale = SMAA_SEARCHTEX_SIZE * float2(0.5, -1.0); + float2 bias = SMAA_SEARCHTEX_SIZE * float2(offset, 1.0); + + // Scale and bias to access texel centers: + scale += float2(-1.0, 1.0); + bias += float2( 0.5, -0.5); + + // Convert from pixel coordinates to texcoords: + // (We use SMAA_SEARCHTEX_PACKED_SIZE because the texture is cropped) + scale *= 1.0 / SMAA_SEARCHTEX_PACKED_SIZE; + bias *= 1.0 / SMAA_SEARCHTEX_PACKED_SIZE; + + float2 coord = mad(scale, e, bias); + coord.y = API_V_COORD(coord.y); + + // Lookup the search texture: + return SMAA_SEARCHTEX_SELECT(SMAASampleLevelZero(searchTex, coord)); +} + +/** + * Horizontal/vertical search functions for the 2nd pass. + */ +float SMAASearchXLeft(SMAATexture2D(edgesTex), SMAATexture2D(searchTex), float2 texcoord, float end) { + /** + * @PSEUDO_GATHER4 + * This texcoord has been offset by (-0.25, -0.125) in the vertex shader to + * sample between edge, thus fetching four edges in a row. + * Sampling with different offsets in each direction allows to disambiguate + * which edges are active from the four fetched ones. + */ + float2 e = float2(0.0, 1.0); + while (texcoord.x > end && + e.g > 0.8281 && // Is there some edge not activated? + e.r == 0.0) { // Or is there a crossing edge that breaks the line? + e = SMAASampleLevelZero(edgesTex, texcoord).rg; + texcoord = mad(-float2(2.0, 0.0), SMAA_RT_METRICS.xy, texcoord); + } + + float offset = mad(-(255.0 / 127.0), SMAASearchLength(SMAATexturePass2D(searchTex), e, 0.0), 3.25); + return mad(SMAA_RT_METRICS.x, offset, texcoord.x); + + // Non-optimized version: + // We correct the previous (-0.25, -0.125) offset we applied: + // texcoord.x += 0.25 * SMAA_RT_METRICS.x; + + // The searches are bias by 1, so adjust the coords accordingly: + // texcoord.x += SMAA_RT_METRICS.x; + + // Disambiguate the length added by the last step: + // texcoord.x += 2.0 * SMAA_RT_METRICS.x; // Undo last step + // texcoord.x -= SMAA_RT_METRICS.x * (255.0 / 127.0) * SMAASearchLength(SMAATexturePass2D(searchTex), e, 0.0); + // return mad(SMAA_RT_METRICS.x, offset, texcoord.x); +} + +float SMAASearchXRight(SMAATexture2D(edgesTex), SMAATexture2D(searchTex), float2 texcoord, float end) { + float2 e = float2(0.0, 1.0); + while (texcoord.x < end && + e.g > 0.8281 && // Is there some edge not activated? + e.r == 0.0) { // Or is there a crossing edge that breaks the line? + e = SMAASampleLevelZero(edgesTex, texcoord).rg; + texcoord = mad(float2(2.0, 0.0), SMAA_RT_METRICS.xy, texcoord); + } + float offset = mad(-(255.0 / 127.0), SMAASearchLength(SMAATexturePass2D(searchTex), e, 0.5), 3.25); + return mad(-SMAA_RT_METRICS.x, offset, texcoord.x); +} + +float SMAASearchYUp(SMAATexture2D(edgesTex), SMAATexture2D(searchTex), float2 texcoord, float end) { + float2 e = float2(1.0, 0.0); + while (API_V_BELOW(texcoord.y, end) && + e.r > 0.8281 && // Is there some edge not activated? + e.g == 0.0) { // Or is there a crossing edge that breaks the line? + e = SMAASampleLevelZero(edgesTex, texcoord).rg; + texcoord = mad(-float2(0.0, API_V_DIR(2.0)), SMAA_RT_METRICS.xy, texcoord); + } + float offset = mad(-(255.0 / 127.0), SMAASearchLength(SMAATexturePass2D(searchTex), e.gr, 0.0), 3.25); + return mad(SMAA_RT_METRICS.y, API_V_DIR(offset), texcoord.y); +} + +float SMAASearchYDown(SMAATexture2D(edgesTex), SMAATexture2D(searchTex), float2 texcoord, float end) { + float2 e = float2(1.0, 0.0); + while (API_V_ABOVE(texcoord.y, end) && + e.r > 0.8281 && // Is there some edge not activated? + e.g == 0.0) { // Or is there a crossing edge that breaks the line? + e = SMAASampleLevelZero(edgesTex, texcoord).rg; + texcoord = mad(float2(0.0, API_V_DIR(2.0)), SMAA_RT_METRICS.xy, texcoord); + } + float offset = mad(-(255.0 / 127.0), SMAASearchLength(SMAATexturePass2D(searchTex), e.gr, 0.5), 3.25); + return mad(-SMAA_RT_METRICS.y, API_V_DIR(offset), texcoord.y); +} + +/** + * Ok, we have the distance and both crossing edges. So, what are the areas + * at each side of current edge? + */ +float2 SMAAArea(SMAATexture2D(areaTex), float2 dist, float e1, float e2, float offset) { + // Rounding prevents precision errors of bilinear filtering: + float2 texcoord = mad(float2(SMAA_AREATEX_MAX_DISTANCE, SMAA_AREATEX_MAX_DISTANCE), round(4.0 * float2(e1, e2)), dist); + + // We do a scale and bias for mapping to texel space: + texcoord = mad(SMAA_AREATEX_PIXEL_SIZE, texcoord, 0.5 * SMAA_AREATEX_PIXEL_SIZE); + + // Move to proper place, according to the subpixel offset: + texcoord.y = mad(SMAA_AREATEX_SUBTEX_SIZE, offset, texcoord.y); + + texcoord.y = API_V_COORD(texcoord.y); + + // Do it! + return SMAA_AREATEX_SELECT(SMAASampleLevelZero(areaTex, texcoord)); +} + +//----------------------------------------------------------------------------- +// Corner Detection Functions + +void SMAADetectHorizontalCornerPattern(SMAATexture2D(edgesTex), inout float2 weights, float4 texcoord, float2 d) { + #if !defined(SMAA_DISABLE_CORNER_DETECTION) + float2 leftRight = step(d.xy, d.yx); + float2 rounding = (1.0 - SMAA_CORNER_ROUNDING_NORM) * leftRight; + + rounding /= leftRight.x + leftRight.y; // Reduce blending for pixels in the center of a line. + + float2 factor = float2(1.0, 1.0); + factor.x -= rounding.x * SMAASampleLevelZeroOffset(edgesTex, texcoord.xy, int2(0, API_V_DIR(1))).r; + factor.x -= rounding.y * SMAASampleLevelZeroOffset(edgesTex, texcoord.zw, int2(1, API_V_DIR(1))).r; + factor.y -= rounding.x * SMAASampleLevelZeroOffset(edgesTex, texcoord.xy, int2(0, API_V_DIR(-2))).r; + factor.y -= rounding.y * SMAASampleLevelZeroOffset(edgesTex, texcoord.zw, int2(1, API_V_DIR(-2))).r; + + weights *= saturate(factor); + #endif +} + +void SMAADetectVerticalCornerPattern(SMAATexture2D(edgesTex), inout float2 weights, float4 texcoord, float2 d) { + #if !defined(SMAA_DISABLE_CORNER_DETECTION) + float2 leftRight = step(d.xy, d.yx); + float2 rounding = (1.0 - SMAA_CORNER_ROUNDING_NORM) * leftRight; + + rounding /= leftRight.x + leftRight.y; + + float2 factor = float2(1.0, 1.0); + factor.x -= rounding.x * SMAASampleLevelZeroOffset(edgesTex, texcoord.xy, int2( 1, 0)).g; + factor.x -= rounding.y * SMAASampleLevelZeroOffset(edgesTex, texcoord.zw, int2( 1, API_V_DIR(1))).g; + factor.y -= rounding.x * SMAASampleLevelZeroOffset(edgesTex, texcoord.xy, int2(-2, 0)).g; + factor.y -= rounding.y * SMAASampleLevelZeroOffset(edgesTex, texcoord.zw, int2(-2, API_V_DIR(1))).g; + + weights *= saturate(factor); + #endif +} + +//----------------------------------------------------------------------------- +// Blending Weight Calculation Pixel Shader (Second Pass) + +float4 SMAABlendingWeightCalculationPS(float2 texcoord, + float2 pixcoord, + float4 offset[3], + SMAATexture2D(edgesTex), + SMAATexture2D(areaTex), + SMAATexture2D(searchTex), + float4 subsampleIndices) { // Just pass zero for SMAA 1x, see @SUBSAMPLE_INDICES. + float4 weights = float4(0.0, 0.0, 0.0, 0.0); + + float2 e = SMAASample(edgesTex, texcoord).rg; + + SMAA_BRANCH + if (e.g > 0.0) { // Edge at north + #if !defined(SMAA_DISABLE_DIAG_DETECTION) + // Diagonals have both north and west edges, so searching for them in + // one of the boundaries is enough. + weights.rg = SMAACalculateDiagWeights(SMAATexturePass2D(edgesTex), SMAATexturePass2D(areaTex), texcoord, e, subsampleIndices); + + // We give priority to diagonals, so if we find a diagonal we skip + // horizontal/vertical processing. + SMAA_BRANCH + if (weights.r == -weights.g) { // weights.r + weights.g == 0.0 + #endif + + float2 d; + + // Find the distance to the left: + float3 coords; + coords.x = SMAASearchXLeft(SMAATexturePass2D(edgesTex), SMAATexturePass2D(searchTex), offset[0].xy, offset[2].x); + coords.y = offset[1].y; // offset[1].y = texcoord.y - 0.25 * SMAA_RT_METRICS.y (@CROSSING_OFFSET) + d.x = coords.x; + + // Now fetch the left crossing edges, two at a time using bilinear + // filtering. Sampling at -0.25 (see @CROSSING_OFFSET) enables to + // discern what value each edge has: + float e1 = SMAASampleLevelZero(edgesTex, coords.xy).r; + + // Find the distance to the right: + coords.z = SMAASearchXRight(SMAATexturePass2D(edgesTex), SMAATexturePass2D(searchTex), offset[0].zw, offset[2].y); + d.y = coords.z; + + // We want the distances to be in pixel units (doing this here allow to + // better interleave arithmetic and memory accesses): + d = abs(round(mad(SMAA_RT_METRICS.zz, d, -pixcoord.xx))); + + // SMAAArea below needs a sqrt, as the areas texture is compressed + // quadratically: + float2 sqrt_d = sqrt(d); + + // Fetch the right crossing edges: + float e2 = SMAASampleLevelZeroOffset(edgesTex, coords.zy, int2(1, 0)).r; + + // Ok, we know how this pattern looks like, now it is time for getting + // the actual area: + weights.rg = SMAAArea(SMAATexturePass2D(areaTex), sqrt_d, e1, e2, subsampleIndices.y); + + // Fix corners: + coords.y = texcoord.y; + SMAADetectHorizontalCornerPattern(SMAATexturePass2D(edgesTex), weights.rg, coords.xyzy, d); + + #if !defined(SMAA_DISABLE_DIAG_DETECTION) + } else + e.r = 0.0; // Skip vertical processing. + #endif + } + + SMAA_BRANCH + if (e.r > 0.0) { // Edge at west + float2 d; + + // Find the distance to the top: + float3 coords; + coords.y = SMAASearchYUp(SMAATexturePass2D(edgesTex), SMAATexturePass2D(searchTex), offset[1].xy, offset[2].z); + coords.x = offset[0].x; // offset[1].x = texcoord.x - 0.25 * SMAA_RT_METRICS.x; + d.x = coords.y; + + // Fetch the top crossing edges: + float e1 = SMAASampleLevelZero(edgesTex, coords.xy).g; + + // Find the distance to the bottom: + coords.z = SMAASearchYDown(SMAATexturePass2D(edgesTex), SMAATexturePass2D(searchTex), offset[1].zw, offset[2].w); + d.y = coords.z; + + // We want the distances to be in pixel units: + d = abs(round(mad(SMAA_RT_METRICS.ww, d, -pixcoord.yy))); + + // SMAAArea below needs a sqrt, as the areas texture is compressed + // quadratically: + float2 sqrt_d = sqrt(d); + + // Fetch the bottom crossing edges: + float e2 = SMAASampleLevelZeroOffset(edgesTex, coords.xz, int2(0, API_V_DIR(1))).g; + + // Get the area for this direction: + weights.ba = SMAAArea(SMAATexturePass2D(areaTex), sqrt_d, e1, e2, subsampleIndices.x); + + // Fix corners: + coords.x = texcoord.x; + SMAADetectVerticalCornerPattern(SMAATexturePass2D(edgesTex), weights.ba, coords.xyxz, d); + } + + return weights; +} + +//----------------------------------------------------------------------------- +// Neighborhood Blending Pixel Shader (Third Pass) + +float4 SMAANeighborhoodBlendingPS(float2 texcoord, + float4 offset, + SMAATexture2D(colorTex), + SMAATexture2D(blendTex) + #if SMAA_REPROJECTION + , SMAATexture2D(velocityTex) + #endif + ) { + // Fetch the blending weights for current pixel: + float4 a; + a.x = SMAASample(blendTex, offset.xy).a; // Right + a.y = SMAASample(blendTex, offset.zw).g; // Top + a.wz = SMAASample(blendTex, texcoord).xz; // Bottom / Left + + // Is there any blending weight with a value greater than 0.0? + SMAA_BRANCH + if (dot(a, float4(1.0, 1.0, 1.0, 1.0)) < 1e-5) { + float4 color = SMAASampleLevelZero(colorTex, texcoord); + + #if SMAA_REPROJECTION + float2 velocity = SMAA_DECODE_VELOCITY(SMAASampleLevelZero(velocityTex, texcoord)); + + // Pack velocity into the alpha channel: + color.a = sqrt(5.0 * length(velocity)); + #endif + + return color; + } else { + bool h = max(a.x, a.z) > max(a.y, a.w); // max(horizontal) > max(vertical) + + // Calculate the blending offsets: + float4 blendingOffset = float4(0.0, API_V_DIR(a.y), 0.0, API_V_DIR(a.w)); + float2 blendingWeight = a.yw; + SMAAMovc(bool4(h, h, h, h), blendingOffset, float4(a.x, 0.0, a.z, 0.0)); + SMAAMovc(bool2(h, h), blendingWeight, a.xz); + blendingWeight /= dot(blendingWeight, float2(1.0, 1.0)); + + // Calculate the texture coordinates: + float4 blendingCoord = mad(blendingOffset, float4(SMAA_RT_METRICS.xy, -SMAA_RT_METRICS.xy), texcoord.xyxy); + + // We exploit bilinear filtering to mix current pixel with the chosen + // neighbor: + float4 color = blendingWeight.x * SMAASampleLevelZero(colorTex, blendingCoord.xy); + color += blendingWeight.y * SMAASampleLevelZero(colorTex, blendingCoord.zw); + + #if SMAA_REPROJECTION + // Antialias velocity for proper reprojection in a later stage: + float2 velocity = blendingWeight.x * SMAA_DECODE_VELOCITY(SMAASampleLevelZero(velocityTex, blendingCoord.xy)); + velocity += blendingWeight.y * SMAA_DECODE_VELOCITY(SMAASampleLevelZero(velocityTex, blendingCoord.zw)); + + // Pack velocity into the alpha channel: + color.a = sqrt(5.0 * length(velocity)); + #endif + + return color; + } +} + +//----------------------------------------------------------------------------- +// Temporal Resolve Pixel Shader (Optional Pass) + +float4 SMAAResolvePS(float2 texcoord, + SMAATexture2D(currentColorTex), + SMAATexture2D(previousColorTex) + #if SMAA_REPROJECTION + , SMAATexture2D(velocityTex) + #endif + ) { + #if SMAA_REPROJECTION + // Velocity is assumed to be calculated for motion blur, so we need to + // inverse it for reprojection: + float2 velocity = -SMAA_DECODE_VELOCITY(SMAASamplePoint(velocityTex, texcoord).rg); + + // Fetch current pixel: + float4 current = SMAASamplePoint(currentColorTex, texcoord); + + // Reproject current coordinates and fetch previous pixel: + float4 previous = SMAASamplePoint(previousColorTex, texcoord + velocity); + + // Attenuate the previous pixel if the velocity is different: + float delta = abs(current.a * current.a - previous.a * previous.a) / 5.0; + float weight = 0.5 * saturate(1.0 - sqrt(delta) * SMAA_REPROJECTION_WEIGHT_SCALE); + + // Blend the pixels according to the calculated weight: + return lerp(current, previous, weight); + #else + // Just blend the pixels: + float4 current = SMAASamplePoint(currentColorTex, texcoord); + float4 previous = SMAASamplePoint(previousColorTex, texcoord); + return lerp(current, previous, 0.5); + #endif +} + +//----------------------------------------------------------------------------- +// Separate Multisamples Pixel Shader (Optional Pass) + +#ifdef SMAALoad +void SMAASeparatePS(float4 position, + float2 texcoord, + out float4 target0, + out float4 target1, + SMAATexture2DMS2(colorTexMS)) { + int2 pos = int2(position.xy); + target0 = SMAALoad(colorTexMS, pos, 0); + target1 = SMAALoad(colorTexMS, pos, 1); +} +#endif + +//----------------------------------------------------------------------------- +#endif // SMAA_INCLUDE_PS diff --git a/indra/newview/app_settings/shaders/class1/deferred/SMAABlendWeightsF.glsl b/indra/newview/app_settings/shaders/class1/deferred/SMAABlendWeightsF.glsl new file mode 100644 index 0000000000..3332c5f58f --- /dev/null +++ b/indra/newview/app_settings/shaders/class1/deferred/SMAABlendWeightsF.glsl @@ -0,0 +1,57 @@ +/** + * @file SMAABlendWeightsF.glsl + * + * $LicenseInfo:firstyear=2024&license=viewerlgpl$ + * Second Life Viewer Source Code + * Copyright (C) 2024, Linden Research, Inc. + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; + * version 2.1 of the License only. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + * + * Linden Research, Inc., 945 Battery Street, San Francisco, CA 94111 USA + * $/LicenseInfo$ + */ + +/*[EXTRA_CODE_HERE]*/ + +out vec4 frag_color; + +in vec2 vary_texcoord0; +in vec2 vary_pixcoord; +in vec4 vary_offset[3]; + +uniform sampler2D edgesTex; +uniform sampler2D areaTex; +uniform sampler2D searchTex; + +vec4 SMAABlendingWeightCalculationPS(vec2 texcoord, + vec2 pixcoord, + vec4 offset[3], + sampler2D edgesTex, + sampler2D areaTex, + sampler2D searchTex, + vec4 subsampleIndices); + +void main() +{ + frag_color = SMAABlendingWeightCalculationPS(vary_texcoord0, + vary_pixcoord, + vary_offset, + edgesTex, + areaTex, + searchTex, + vec4(0.0) + ); +} + diff --git a/indra/newview/app_settings/shaders/class1/deferred/SMAABlendWeightsV.glsl b/indra/newview/app_settings/shaders/class1/deferred/SMAABlendWeightsV.glsl new file mode 100644 index 0000000000..52f85ef30c --- /dev/null +++ b/indra/newview/app_settings/shaders/class1/deferred/SMAABlendWeightsV.glsl @@ -0,0 +1,51 @@ +/** + * @file SMAABlendWeightsV.glsl + * + * $LicenseInfo:firstyear=2024&license=viewerlgpl$ + * Second Life Viewer Source Code + * Copyright (C) 2024, Linden Research, Inc. + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; + * version 2.1 of the License only. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + * + * Linden Research, Inc., 945 Battery Street, San Francisco, CA 94111 USA + * $/LicenseInfo$ + */ + +/*[EXTRA_CODE_HERE]*/ + +uniform mat4 modelview_projection_matrix; + +in vec3 position; + +out vec2 vary_texcoord0; +out vec2 vary_pixcoord; +out vec4 vary_offset[3]; + +#define float4 vec4 +#define float2 vec2 +void SMAABlendingWeightCalculationVS(float2 texcoord, + out float2 pixcoord, + out float4 offset[3]); + +void main() +{ + gl_Position = vec4(position.xyz, 1.0); + vary_texcoord0 = (gl_Position.xy*0.5+0.5); + + SMAABlendingWeightCalculationVS(vary_texcoord0, + vary_pixcoord, + vary_offset); +} + diff --git a/indra/newview/app_settings/shaders/class1/deferred/SMAAEdgeDetectF.glsl b/indra/newview/app_settings/shaders/class1/deferred/SMAAEdgeDetectF.glsl new file mode 100644 index 0000000000..0a8cd4a4ea --- /dev/null +++ b/indra/newview/app_settings/shaders/class1/deferred/SMAAEdgeDetectF.glsl @@ -0,0 +1,59 @@ +/** + * @file SMAAEdgeDetectF.glsl + * $LicenseInfo:firstyear=2024&license=viewerlgpl$ + * Second Life Viewer Source Code + * Copyright (C) 2024, Linden Research, Inc. + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; + * version 2.1 of the License only. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + * + * Linden Research, Inc., 945 Battery Street, San Francisco, CA 94111 USA + * $/LicenseInfo$ + */ + +/*[EXTRA_CODE_HERE]*/ + +out vec4 frag_color; + +in vec2 vary_texcoord0; +in vec4 vary_offset[3]; + +uniform sampler2D diffuseRect; +#if SMAA_PREDICATION +uniform sampler2D predicationTex; +#endif + +#define float4 vec4 +#define float2 vec2 +#define SMAATexture2D(tex) sampler2D tex + +float2 SMAAColorEdgeDetectionPS(float2 texcoord, + float4 offset[3], + SMAATexture2D(colorTex) + #if SMAA_PREDICATION + , SMAATexture2D(predicationTex) + #endif + ); + +void main() +{ + vec2 val = SMAAColorEdgeDetectionPS(vary_texcoord0, + vary_offset, + diffuseRect + #if SMAA_PREDICATION + , predicationTex + #endif + ); + frag_color = float4(val,0.0,0.0); +} diff --git a/indra/newview/app_settings/shaders/class1/deferred/SMAAEdgeDetectV.glsl b/indra/newview/app_settings/shaders/class1/deferred/SMAAEdgeDetectV.glsl new file mode 100644 index 0000000000..7c0184bfc4 --- /dev/null +++ b/indra/newview/app_settings/shaders/class1/deferred/SMAAEdgeDetectV.glsl @@ -0,0 +1,45 @@ +/** + * @file SMAAEdgeDetectV.glsl + * $LicenseInfo:firstyear=2024&license=viewerlgpl$ + * Second Life Viewer Source Code + * Copyright (C) 2024, Linden Research, Inc. + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; + * version 2.1 of the License only. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + * + * Linden Research, Inc., 945 Battery Street, San Francisco, CA 94111 USA + * $/LicenseInfo$ + */ + +/*[EXTRA_CODE_HERE]*/ + +uniform mat4 modelview_projection_matrix; + +in vec3 position; + +out vec2 vary_texcoord0; +out vec4 vary_offset[3]; + +#define float4 vec4 +#define float2 vec2 +void SMAAEdgeDetectionVS(float2 texcoord, + out float4 offset[3]); + +void main() +{ + gl_Position = vec4(position.xyz, 1.0); + vary_texcoord0 = (gl_Position.xy*0.5+0.5); + + SMAAEdgeDetectionVS(vary_texcoord0, vary_offset); +} diff --git a/indra/newview/app_settings/shaders/class1/deferred/SMAANeighborhoodBlendF.glsl b/indra/newview/app_settings/shaders/class1/deferred/SMAANeighborhoodBlendF.glsl new file mode 100644 index 0000000000..3276405447 --- /dev/null +++ b/indra/newview/app_settings/shaders/class1/deferred/SMAANeighborhoodBlendF.glsl @@ -0,0 +1,63 @@ +/** + * @file SMAANeighborhoodBlendF.glsl + * + * $LicenseInfo:firstyear=2024&license=viewerlgpl$ + * Second Life Viewer Source Code + * Copyright (C) 2024, Linden Research, Inc. + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; + * version 2.1 of the License only. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + * + * Linden Research, Inc., 945 Battery Street, San Francisco, CA 94111 USA + * $/LicenseInfo$ + */ + +/*[EXTRA_CODE_HERE]*/ + +out vec4 frag_color; + +in vec2 vary_texcoord0; +in vec4 vary_offset; + +uniform sampler2D diffuseRect; +uniform sampler2D blendTex; +#if SMAA_REPROJECTION +uniform sampler2D velocityTex; +#endif + +#define float4 vec4 +#define float2 vec2 +#define SMAATexture2D(tex) sampler2D tex + +float4 SMAANeighborhoodBlendingPS(float2 texcoord, + float4 offset, + SMAATexture2D(colorTex), + SMAATexture2D(blendTex) + #if SMAA_REPROJECTION + , SMAATexture2D(velocityTex) + #endif + ); + +void main() +{ + frag_color = SMAANeighborhoodBlendingPS(vary_texcoord0, + vary_offset, + diffuseRect, + blendTex + #if SMAA_REPROJECTION + , velocityTex + #endif + ); +} + diff --git a/indra/newview/app_settings/shaders/class1/deferred/SMAANeighborhoodBlendV.glsl b/indra/newview/app_settings/shaders/class1/deferred/SMAANeighborhoodBlendV.glsl new file mode 100644 index 0000000000..7ea1ac61e3 --- /dev/null +++ b/indra/newview/app_settings/shaders/class1/deferred/SMAANeighborhoodBlendV.glsl @@ -0,0 +1,47 @@ +/** + * @file SMAANeighborhoodBlendV.glsl + * + * $LicenseInfo:firstyear=2024&license=viewerlgpl$ + * Second Life Viewer Source Code + * Copyright (C) 2024, Linden Research, Inc. + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; + * version 2.1 of the License only. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + * + * Linden Research, Inc., 945 Battery Street, San Francisco, CA 94111 USA + * $/LicenseInfo$ + */ + +/*[EXTRA_CODE_HERE]*/ + +uniform mat4 modelview_projection_matrix; + +in vec3 position; + +out vec2 vary_texcoord0; +out vec4 vary_offset; + +#define float4 vec4 +#define float2 vec2 +void SMAANeighborhoodBlendingVS(float2 texcoord, + out float4 offset); + +void main() +{ + gl_Position = vec4(position.xyz, 1.0); + vary_texcoord0 = (gl_Position.xy*0.5+0.5); + + SMAANeighborhoodBlendingVS(vary_texcoord0, vary_offset); +} + diff --git a/indra/newview/app_settings/shaders/class1/deferred/fxaaF.glsl b/indra/newview/app_settings/shaders/class1/deferred/fxaaF.glsl index 16e23a3da7..655cb1ea97 100644 --- a/indra/newview/app_settings/shaders/class1/deferred/fxaaF.glsl +++ b/indra/newview/app_settings/shaders/class1/deferred/fxaaF.glsl @@ -32,7 +32,7 @@ out vec4 frag_color; #define FXAA_PC 1 //#define FXAA_GLSL_130 1 -#define FXAA_QUALITY__PRESET 12 +//#define FXAA_QUALITY__PRESET 12 /*============================================================================ diff --git a/indra/newview/app_settings/shaders/class1/deferred/postDeferredGammaCorrect.glsl b/indra/newview/app_settings/shaders/class1/deferred/postDeferredGammaCorrect.glsl index a0eb6cfbb8..0673159ab7 100644 --- a/indra/newview/app_settings/shaders/class1/deferred/postDeferredGammaCorrect.glsl +++ b/indra/newview/app_settings/shaders/class1/deferred/postDeferredGammaCorrect.glsl @@ -28,140 +28,14 @@ out vec4 frag_color; uniform sampler2D diffuseRect; -uniform sampler2D exposureMap; - -uniform vec2 screen_res; -in vec2 vary_fragcoord; - -vec3 linear_to_srgb(vec3 cl); - -//=============================================================== -// tone mapping taken from Khronos sample implementation -//=============================================================== - -// sRGB => XYZ => D65_2_D60 => AP1 => RRT_SAT -const mat3 ACESInputMat = mat3 -( - 0.59719, 0.07600, 0.02840, - 0.35458, 0.90834, 0.13383, - 0.04823, 0.01566, 0.83777 -); - - -// ODT_SAT => XYZ => D60_2_D65 => sRGB -const mat3 ACESOutputMat = mat3 -( - 1.60475, -0.10208, -0.00327, - -0.53108, 1.10813, -0.07276, - -0.07367, -0.00605, 1.07602 -); - -// ACES tone map (faster approximation) -// see: https://knarkowicz.wordpress.com/2016/01/06/aces-filmic-tone-mapping-curve/ -vec3 toneMapACES_Narkowicz(vec3 color) -{ - const float A = 2.51; - const float B = 0.03; - const float C = 2.43; - const float D = 0.59; - const float E = 0.14; - return clamp((color * (A * color + B)) / (color * (C * color + D) + E), 0.0, 1.0); -} - - -// ACES filmic tone map approximation -// see https://github.com/TheRealMJP/BakingLab/blob/master/BakingLab/ACES.hlsl -vec3 RRTAndODTFit(vec3 color) -{ - vec3 a = color * (color + 0.0245786) - 0.000090537; - vec3 b = color * (0.983729 * color + 0.4329510) + 0.238081; - return a / b; -} - - -// tone mapping -vec3 toneMapACES_Hill(vec3 color) -{ - color = ACESInputMat * color; - - // Apply RRT and ODT - color = RRTAndODTFit(color); - - color = ACESOutputMat * color; - - // Clamp to [0, 1] - color = clamp(color, 0.0, 1.0); - - return color; -} uniform float exposure; uniform float gamma; uniform float aces_mix; +uniform vec2 screen_res; +in vec2 vary_fragcoord; -vec3 toneMap(vec3 color) -{ -#ifndef NO_POST - float exp_scale = texture(exposureMap, vec2(0.5,0.5)).r; - - color *= exposure * exp_scale; - - // mix ACES and Linear here as a compromise to avoid over-darkening legacy content - color = mix(toneMapACES_Hill(color), color, aces_mix); -#endif - - return color; -} - -//=============================================================== - -//================================= -// borrowed noise from: -// -// By Morgan McGuire @morgan3d, http://graphicscodex.com -// -float hash(float n) { return fract(sin(n) * 1e4); } -float hash(vec2 p) { return fract(1e4 * sin(17.0 * p.x + p.y * 0.1) * (0.1 + abs(sin(p.y * 13.0 + p.x)))); } - -float noise(float x) { - float i = floor(x); - float f = fract(x); - float u = f * f * (3.0 - 2.0 * f); - return mix(hash(i), hash(i + 1.0), u); -} - -float noise(vec2 x) { - vec2 i = floor(x); - vec2 f = fract(x); - - // Four corners in 2D of a tile - float a = hash(i); - float b = hash(i + vec2(1.0, 0.0)); - float c = hash(i + vec2(0.0, 1.0)); - float d = hash(i + vec2(1.0, 1.0)); - - // Simple 2D lerp using smoothstep envelope between the values. - // return vec3(mix(mix(a, b, smoothstep(0.0, 1.0, f.x)), - // mix(c, d, smoothstep(0.0, 1.0, f.x)), - // smoothstep(0.0, 1.0, f.y))); - - // Same code, with the clamps in smoothstep and common subexpressions - // optimized away. - vec2 u = f * f * (3.0 - 2.0 * f); - return mix(a, b, u.x) + (c - a) * u.y * (1.0 - u.x) + (d - b) * u.x * u.y; -} - -//============================= - -void debugExposure(inout vec3 color) -{ - float exp_scale = texture(exposureMap, vec2(0.5,0.5)).r; - exp_scale *= 0.5; - if (abs(vary_fragcoord.y-exp_scale) < 0.01 && vary_fragcoord.x < 0.1) - { - color = vec3(1,0,0); - } -} +vec3 linear_to_srgb(vec3 cl); vec3 legacyGamma(vec3 color) { @@ -175,23 +49,12 @@ void main() { //this is the one of the rare spots where diffuseRect contains linear color values (not sRGB) vec4 diff = texture(diffuseRect, vary_fragcoord); + diff.rgb = linear_to_srgb(diff.rgb); #ifdef LEGACY_GAMMA - diff.rgb = linear_to_srgb(diff.rgb); diff.rgb = legacyGamma(diff.rgb); -#else -#ifndef NO_POST - diff.rgb = toneMap(diff.rgb); #endif - diff.rgb = linear_to_srgb(diff.rgb); -#endif - - vec2 tc = vary_fragcoord.xy*screen_res*4.0; - vec3 seed = (diff.rgb+vec3(1.0))*vec3(tc.xy, tc.x+tc.y); - vec3 nz = vec3(noise(seed.rg), noise(seed.gb), noise(seed.rb)); - diff.rgb += nz*0.003; - //debugExposure(diff.rgb); frag_color = max(diff, vec4(0)); } diff --git a/indra/newview/app_settings/shaders/class1/deferred/postDeferredNoDoFF.glsl b/indra/newview/app_settings/shaders/class1/deferred/postDeferredNoDoFF.glsl index 07384ebe9b..32b0a1ac8e 100644 --- a/indra/newview/app_settings/shaders/class1/deferred/postDeferredNoDoFF.glsl +++ b/indra/newview/app_settings/shaders/class1/deferred/postDeferredNoDoFF.glsl @@ -33,10 +33,57 @@ uniform sampler2D depthMap; uniform vec2 screen_res; in vec2 vary_fragcoord; +//================================= +// borrowed noise from: +// +// By Morgan McGuire @morgan3d, http://graphicscodex.com +// +float hash(float n) { return fract(sin(n) * 1e4); } +float hash(vec2 p) { return fract(1e4 * sin(17.0 * p.x + p.y * 0.1) * (0.1 + abs(sin(p.y * 13.0 + p.x)))); } + +float noise(float x) { + float i = floor(x); + float f = fract(x); + float u = f * f * (3.0 - 2.0 * f); + return mix(hash(i), hash(i + 1.0), u); +} + +float noise(vec2 x) { + vec2 i = floor(x); + vec2 f = fract(x); + + // Four corners in 2D of a tile + float a = hash(i); + float b = hash(i + vec2(1.0, 0.0)); + float c = hash(i + vec2(0.0, 1.0)); + float d = hash(i + vec2(1.0, 1.0)); + + // Simple 2D lerp using smoothstep envelope between the values. + // return vec3(mix(mix(a, b, smoothstep(0.0, 1.0, f.x)), + // mix(c, d, smoothstep(0.0, 1.0, f.x)), + // smoothstep(0.0, 1.0, f.y))); + + // Same code, with the clamps in smoothstep and common subexpressions + // optimized away. + vec2 u = f * f * (3.0 - 2.0 * f); + return mix(a, b, u.x) + (c - a) * u.y * (1.0 - u.x) + (d - b) * u.x * u.y; +} + +//============================= + + + void main() { vec4 diff = texture(diffuseRect, vary_fragcoord.xy); +#ifdef HAS_NOISE + vec2 tc = vary_fragcoord.xy*screen_res*4.0; + vec3 seed = (diff.rgb+vec3(1.0))*vec3(tc.xy, tc.x+tc.y); + vec3 nz = vec3(noise(seed.rg), noise(seed.gb), noise(seed.rb)); + diff.rgb += nz*0.003; +#endif + frag_color = diff; gl_FragDepth = texture(depthMap, vary_fragcoord.xy).r; diff --git a/indra/newview/app_settings/shaders/class1/deferred/postDeferredTonemap.glsl b/indra/newview/app_settings/shaders/class1/deferred/postDeferredTonemap.glsl new file mode 100644 index 0000000000..c16ab2f9c4 --- /dev/null +++ b/indra/newview/app_settings/shaders/class1/deferred/postDeferredTonemap.glsl @@ -0,0 +1,150 @@ +/** + * @file postDeferredTonemap.glsl + * + * $LicenseInfo:firstyear=2024&license=viewerlgpl$ + * Second Life Viewer Source Code + * Copyright (C) 2024, Linden Research, Inc. + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; + * version 2.1 of the License only. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + * + * Linden Research, Inc., 945 Battery Street, San Francisco, CA 94111 USA + * $/LicenseInfo$ + */ + +/*[EXTRA_CODE_HERE]*/ + +out vec4 frag_color; + +uniform sampler2D diffuseRect; +uniform sampler2D exposureMap; + +uniform vec2 screen_res; +in vec2 vary_fragcoord; + +vec3 linear_to_srgb(vec3 cl); + +//=============================================================== +// tone mapping taken from Khronos sample implementation +//=============================================================== + +// sRGB => XYZ => D65_2_D60 => AP1 => RRT_SAT +const mat3 ACESInputMat = mat3 +( + 0.59719, 0.07600, 0.02840, + 0.35458, 0.90834, 0.13383, + 0.04823, 0.01566, 0.83777 +); + + +// ODT_SAT => XYZ => D60_2_D65 => sRGB +const mat3 ACESOutputMat = mat3 +( + 1.60475, -0.10208, -0.00327, + -0.53108, 1.10813, -0.07276, + -0.07367, -0.00605, 1.07602 +); + +// ACES tone map (faster approximation) +// see: https://knarkowicz.wordpress.com/2016/01/06/aces-filmic-tone-mapping-curve/ +vec3 toneMapACES_Narkowicz(vec3 color) +{ + const float A = 2.51; + const float B = 0.03; + const float C = 2.43; + const float D = 0.59; + const float E = 0.14; + return clamp((color * (A * color + B)) / (color * (C * color + D) + E), 0.0, 1.0); +} + + +// ACES filmic tone map approximation +// see https://github.com/TheRealMJP/BakingLab/blob/master/BakingLab/ACES.hlsl +vec3 RRTAndODTFit(vec3 color) +{ + vec3 a = color * (color + 0.0245786) - 0.000090537; + vec3 b = color * (0.983729 * color + 0.4329510) + 0.238081; + return a / b; +} + + +// tone mapping +vec3 toneMapACES_Hill(vec3 color) +{ + color = ACESInputMat * color; + + // Apply RRT and ODT + color = RRTAndODTFit(color); + + color = ACESOutputMat * color; + + // Clamp to [0, 1] + color = clamp(color, 0.0, 1.0); + + return color; +} + +uniform float exposure; +uniform float gamma; +uniform float aces_mix; + +vec3 toneMap(vec3 color) +{ +#ifndef NO_POST + float exp_scale = texture(exposureMap, vec2(0.5,0.5)).r; + + color *= exposure * exp_scale; + + // mix ACES and Linear here as a compromise to avoid over-darkening legacy content + color = mix(toneMapACES_Hill(color), color, aces_mix); +#endif + + return color; +} + +//=============================================================== + +void debugExposure(inout vec3 color) +{ + float exp_scale = texture(exposureMap, vec2(0.5,0.5)).r; + exp_scale *= 0.5; + if (abs(vary_fragcoord.y-exp_scale) < 0.01 && vary_fragcoord.x < 0.1) + { + color = vec3(1,0,0); + } +} + +vec3 legacyGamma(vec3 color) +{ + vec3 c = 1. - clamp(color, vec3(0.), vec3(1.)); + c = 1. - pow(c, vec3(gamma)); // s/b inverted already CPU-side + + return c; +} + +void main() +{ + //this is the one of the rare spots where diffuseRect contains linear color values (not sRGB) + vec4 diff = texture(diffuseRect, vary_fragcoord); + +#ifndef NO_POST + diff.rgb = toneMap(diff.rgb); +#else + diff.rgb = clamp(diff.rgb, vec3(0.0), vec3(1.0)); +#endif + + //debugExposure(diff.rgb); + frag_color = max(diff, vec4(0)); +} + -- cgit v1.2.3 From 5b832291a8ea1f100e2caaffef3d162dd96bee50 Mon Sep 17 00:00:00 2001 From: Rye Mutt Date: Fri, 30 Aug 2024 12:49:05 -0700 Subject: Introduce Khronos Neutral tonemapper as new default along with debug options to control tonemap mix (#2464) (#2468) --- .../class1/deferred/postDeferredGammaCorrect.glsl | 2 - .../class1/deferred/postDeferredTonemap.glsl | 52 +++++++++++++++++----- 2 files changed, 40 insertions(+), 14 deletions(-) (limited to 'indra/newview/app_settings/shaders/class1/deferred') diff --git a/indra/newview/app_settings/shaders/class1/deferred/postDeferredGammaCorrect.glsl b/indra/newview/app_settings/shaders/class1/deferred/postDeferredGammaCorrect.glsl index 0673159ab7..befd2ae6da 100644 --- a/indra/newview/app_settings/shaders/class1/deferred/postDeferredGammaCorrect.glsl +++ b/indra/newview/app_settings/shaders/class1/deferred/postDeferredGammaCorrect.glsl @@ -29,9 +29,7 @@ out vec4 frag_color; uniform sampler2D diffuseRect; -uniform float exposure; uniform float gamma; -uniform float aces_mix; uniform vec2 screen_res; in vec2 vary_fragcoord; diff --git a/indra/newview/app_settings/shaders/class1/deferred/postDeferredTonemap.glsl b/indra/newview/app_settings/shaders/class1/deferred/postDeferredTonemap.glsl index c16ab2f9c4..fc6d4d7727 100644 --- a/indra/newview/app_settings/shaders/class1/deferred/postDeferredTonemap.glsl +++ b/indra/newview/app_settings/shaders/class1/deferred/postDeferredTonemap.glsl @@ -95,9 +95,33 @@ vec3 toneMapACES_Hill(vec3 color) return color; } +// Khronos Neutral tonemapping +// https://github.com/KhronosGroup/ToneMapping/tree/main +// Input color is non-negative and resides in the Linear Rec. 709 color space. +// Output color is also Linear Rec. 709, but in the [0, 1] range. +vec3 PBRNeutralToneMapping( vec3 color ) +{ + const float startCompression = 0.8 - 0.04; + const float desaturation = 0.15; + + float x = min(color.r, min(color.g, color.b)); + float offset = x < 0.08 ? x - 6.25 * x * x : 0.04; + color -= offset; + + float peak = max(color.r, max(color.g, color.b)); + if (peak < startCompression) return color; + + const float d = 1. - startCompression; + float newPeak = 1. - d * d / (peak + d - startCompression); + color *= newPeak / peak; + + float g = 1. - 1. / (desaturation * (peak - newPeak) + 1.); + return mix(color, newPeak * vec3(1, 1, 1), g); +} + uniform float exposure; -uniform float gamma; -uniform float aces_mix; +uniform float tonemap_mix; +uniform int tonemap_type; vec3 toneMap(vec3 color) { @@ -106,8 +130,20 @@ vec3 toneMap(vec3 color) color *= exposure * exp_scale; - // mix ACES and Linear here as a compromise to avoid over-darkening legacy content - color = mix(toneMapACES_Hill(color), color, aces_mix); + vec3 clamped_color = clamp(color.rgb, vec3(0.0), vec3(1.0)); + + switch(tonemap_type) + { + case 0: + color = PBRNeutralToneMapping(color); + break; + case 1: + color = toneMapACES_Hill(color); + break; + } + + // mix tonemapped and linear here to provide adjustment + color = mix(clamped_color, color, tonemap_mix); #endif return color; @@ -125,14 +161,6 @@ void debugExposure(inout vec3 color) } } -vec3 legacyGamma(vec3 color) -{ - vec3 c = 1. - clamp(color, vec3(0.), vec3(1.)); - c = 1. - pow(c, vec3(gamma)); // s/b inverted already CPU-side - - return c; -} - void main() { //this is the one of the rare spots where diffuseRect contains linear color values (not sRGB) -- cgit v1.2.3