o3de/Gems/AtomTressFX/Assets/Shaders/HairShortCutGeometryDepthAlpha.azsl

/*
* Modifications Copyright (c) Contributors to the Open 3D Engine Project. 
* For complete copyright and license terms please see the LICENSE at the root of this distribution.
* 
* SPDX-License-Identifier: Apache-2.0 OR MIT
*
*/

//
// Copyright (c) 2019 Advanced Micro Devices, Inc. All rights reserved.
//
// Permission is hereby granted, free of charge, to any person obtaining a copy
// of 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.
//
// 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.
//

#include <Atom/Features/SrgSemantics.azsli>
#include <HairRenderingSrgs.azsli>

//!------------------------------ SRG Structure --------------------------------
//! Per pass SRG that holds the dynamic shared read-write buffer shared 
//! across all dispatches and draw calls. It is used for all the dynamic buffers
//! that can change between passes due to the application of skinning, simulation 
//! and physics affect. 
//! Once the compute pases are done, it is read by the rendering shaders.  
ShaderResourceGroup PassSrg : SRG_PerPass_WithFallback
{
    //! This shared buffer needs to match the SharedBuffer structure  
    //! shared between all draw calls / dispatches for the hair skinning
    StructuredBuffer<int>   m_skinnedHairSharedBuffer;

    //! Based on [[vk::binding(0, 3)]] RWTexture2DArray<uint> RWFragmentDepthsTexture : register(u0, space3);
    RWTexture2DArray<uint>  m_RWFragmentDepthsTexture;
}
//==============================================================================

//!------------------------------ SRG Structure --------------------------------
//! Per instance/draw SRG representing dynamic read-write set of buffers
//!  that are unique per instance and are shared and changed between passes due 
//!  to the application of skinning, simulation and physics affect.
//! It is then also read by the rendering shaders. 
//! This Srg is NOT shared by the passes since it requires having barriers between
//!  both passes and draw calls, instead, all buffers are allocated from a single 
//!  shared buffer (through BufferViews) and that buffer is then shared between 
//!  the passes via the PerPass Srg frequency. 
ShaderResourceGroup HairDynamicDataSrg : SRG_PerObject // space 1 - per instance / object
{
    Buffer<float4>      m_hairVertexPositions;
    Buffer<float4>      m_hairVertexTangents;

    //! Per hair object offset to the start location of each buffer within 
    //! 'm_skinnedHairSharedBuffer'. The offset is in bytes!
    uint m_positionBufferOffset;
    uint m_tangentBufferOffset;
};
//------------------------------------------------------------------------------
// Allow for the code to run with minimal changes - skinning / simulation compute passes
// Usage of per-instance buffer
#define g_GuideHairVertexPositions  HairDynamicDataSrg::m_hairVertexPositions
#define g_GuideHairVertexTangents   HairDynamicDataSrg::m_hairVertexTangents
//------------------------------------------------------------------------------

#include <HairStrands.azsli>        // VS resides here

//!=============================================================================
//!         Geometry Depth Alpha - First Pass of ShortCut Render
//! It is a Geometry pass that stores the K=3 front fragment depths, and accumulates 
//! product of 1-alpha multiplications (fade out) of the input render target.
//! 
//! Short explanation: in the original AMD implementation 1-alpha is multiplied
//! repeatedly with the incoming render target (back buffer) hence blending out 
//! the existing back buffer color based on the density and transparency of the hair.
//! This implies that later on the hair color should be added based on the inverse
//! of this operation.
//!=============================================================================
[earlydepthstencil]
float HairShortCutDepthsAlphaPS(PS_INPUT_HAIR input) : SV_Target
{
    //////////////////////////////////////////////////////////////////////
    // [To Do] Hair: anti aliasing via coverage requires work and is disabled for now
//    float3 vNDC = ScreenPosToNDC(PassSrg::m_linearDepth, input.Position.xy, input.Position.z);
//    uint2  dimensions;
//    PassSrg::m_linearDepth.GetDimensions(dimensions.x, dimensions.y);
//    float coverage = ComputeCoverage(input.p0p1.xy, input.p0p1.zw, vNDC.xy, float2(dimensions.x, dimensions.y));
    float coverage = 1.0;  
    /////////////////////////////////////////////////////////////////////

    float alpha = coverage * MatBaseColor.a;

    if (alpha < SHORTCUT_MIN_ALPHA)
        return 1.0;

    int2 vScreenAddress = int2(input.Position.xy);
    uint uDepth = asuint(input.Position.z);
    uint uDepth0Prev, uDepth1Prev, uDepth2Prev;

    // Min of depth 0 and input depth - in Atom the Z order is reverse
    // Original value is uDepth0Prev
    InterlockedMax(PassSrg::m_RWFragmentDepthsTexture[uint3(vScreenAddress, 0)], uDepth, uDepth0Prev);

    // Min of depth 1 and greater of the last compare - in Atom the Z order is reverse
    // If fragment opaque, always use input depth (don't need greater depths)
    uDepth = (alpha > 0.98) ? uDepth : max(uDepth, uDepth0Prev);

    InterlockedMax(PassSrg::m_RWFragmentDepthsTexture[uint3(vScreenAddress, 1)], uDepth, uDepth1Prev);

    // Min of depth 2 and greater of the last compare - in Atom the Z order is reverse
    // If fragment opaque, always use input depth (don't need greater depths)
    uDepth = (alpha > 0.98) ? uDepth : max(uDepth, uDepth1Prev);

    InterlockedMax(PassSrg::m_RWFragmentDepthsTexture[uint3(vScreenAddress, 2)], uDepth, uDepth2Prev);

    // Accumulate the alpha multiplication from all hair components by multiplying the inverse and 
    // therefore going down towards 0.  At the end product, the inverse will be taken as the hair 
    // alpha and the remainder will be used to blend the back buffer.
    return 1.0 - alpha;
}