o3de/Gems/AtomTressFX/Assets/Shaders/HairSimulationCommon.azsli

/*
 * 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) AND MIT
 *
 */
 
//---------------------------------------------------------------------------------------
// Shader code related to simulating hair strands in compute.
//-------------------------------------------------------------------------------------
//
// 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.
//

//--------------------------------------------------------------------------------------
// File: HairSimulation.azsl
//
// Physics simulation of hair using compute shaders
//--------------------------------------------------------------------------------------

#pragma once

#define USE_MESH_BASED_HAIR_TRANSFORM 0

// If you change the value below, you must change it in TressFXAsset.h as well.
#ifndef THREAD_GROUP_SIZE
    #define THREAD_GROUP_SIZE 64
#endif

groupshared float4 sharedPos[THREAD_GROUP_SIZE];
groupshared float4 sharedTangent[THREAD_GROUP_SIZE];
groupshared float  sharedLength[THREAD_GROUP_SIZE];

//--------------------------------------------------------------------------------------
//
//  			Helper Functions for the main simulation shaders
//
//--------------------------------------------------------------------------------------
bool IsMovable(float4 particle)
{
    if ( particle.w > 0 )
        return true;
    return false;
}

float2 ConstraintMultiplier(float4 particle0, float4 particle1)
{
    if (IsMovable(particle0))
    {
        if (IsMovable(particle1))
            return float2(0.5, 0.5);
        else
            return float2(1, 0);
    }
    else
    {
        if (IsMovable(particle1))
            return float2(0, 1);
        else
            return float2(0, 0);
    }
}

float4 MakeQuaternion(float angle_radian, float3 axis)
{
    // create quaternion using angle and rotation axis
    float4 quaternion;
    float halfAngle = 0.5f * angle_radian;
    float sinHalf = sin(halfAngle);

    quaternion.w = cos(halfAngle);
    quaternion.xyz = sinHalf * axis.xyz;

    return quaternion;
}

// Makes a quaternion from a 4x4 column major rigid transform matrix. Rigid transform means that rotational 3x3 sub matrix is orthonormal. 
// Note that this function does not check the orthonormality. 
float4 MakeQuaternion(column_major float4x4 m)
{
    float4 q;
    float trace = m[0][0] + m[1][1] + m[2][2];

    if (trace > 0.0f)
    {
        float r = sqrt(trace + 1.0f);
        q.w = 0.5 * r;
        r = 0.5 / r;
        q.x = (m[1][2] - m[2][1])*r;
        q.y = (m[2][0] - m[0][2])*r;
        q.z = (m[0][1] - m[1][0])*r;
    }
    else
    {
        int i = 0, j = 1, k = 2;

        if (m[1][1] > m[0][0])
        {
            i = 1; j = 2; k = 0;
        }
        if (m[2][2] > m[i][i])
        {
            i = 2; j = 0; k = 1;
        }

        float r = sqrt(m[i][i] - m[j][j] - m[k][k] + 1.0f);

        float qq[4];

        qq[i] = 0.5f * r;
        r = 0.5f / r;
        q.w = (m[j][k] - m[k][j])*r;
        qq[j] = (m[j][i] + m[i][j])*r;
        qq[k] = (m[k][i] + m[i][k])*r;

        q.x = qq[0]; q.y = qq[1]; q.z = qq[2];
    }

    return q;
}

float4 InverseQuaternion(float4 q)
{
    float lengthSqr = q.x*q.x + q.y*q.y + q.z*q.z + q.w*q.w;

    if ( lengthSqr < 0.001 )
        return float4(0, 0, 0, 1.0f);

    q.x = -q.x / lengthSqr;
    q.y = -q.y / lengthSqr;
    q.z = -q.z / lengthSqr;
    q.w = q.w / lengthSqr;

    return q;
}

float3 MultQuaternionAndVector(float4 q, float3 v)
{
    float3 uv, uuv;
    float3 qvec = float3(q.x, q.y, q.z);
    uv = cross(qvec, v);
    uuv = cross(qvec, uv);
    uv *= (2.0f * q.w);
    uuv *= 2.0f;

    return v + uv + uuv;
}

float4 MultQuaternionAndQuaternion(float4 qA, float4 qB)
{
    float4 q;

    q.w = qA.w * qB.w - qA.x * qB.x - qA.y * qB.y - qA.z * qB.z;
    q.x = qA.w * qB.x + qA.x * qB.w + qA.y * qB.z - qA.z * qB.y;
    q.y = qA.w * qB.y + qA.y * qB.w + qA.z * qB.x - qA.x * qB.z;
    q.z = qA.w * qB.z + qA.z * qB.w + qA.x * qB.y - qA.y * qB.x;

    return q;
}

float4 NormalizeQuaternion(float4 q)
{
    float4 qq = q;
    float n = q.x*q.x + q.y*q.y + q.z*q.z + q.w*q.w;

    if (n < 1e-10f)
    {
        qq.w = 1;
        return qq;
    }

    qq *= 1.0f / sqrt(n);
    return qq;
}

// Compute a quaternion which rotates u to v. u and v must be unit vector. 
float4 QuatFromTwoUnitVectors(float3 u, float3 v)
{
    float r = 1.f + dot(u, v);
    float3 n;

    // if u and v are parallel
    if (r < 1e-7)
    {
        r = 0.0f;
        n = abs(u.x) > abs(u.z) ? float3(-u.y, u.x, 0.f) : float3(0.f, -u.z, u.y);
    }
    else
    {
        n = cross(u, v);  
    }

    float4 q = float4(n.x, n.y, n.z, r);
    return NormalizeQuaternion(q);
}

// Make the inpute 4x4 matrix be identity
float4x4 MakeIdentity()
{
    float4x4 m;
    m._m00 = 1;   m._m01 = 0;   m._m02 = 0;   m._m03 = 0;
    m._m10 = 0;   m._m11 = 1;   m._m12 = 0;   m._m13 = 0;
    m._m20 = 0;   m._m21 = 0;   m._m22 = 1;   m._m23 = 0;
    m._m30 = 0;   m._m31 = 0;   m._m32 = 0;   m._m33 = 1;

    return m;
}

void ApplyDistanceConstraint(inout float4 pos0, inout float4 pos1, float targetDistance, float stiffness = 1.0)
{
    float3 delta = pos1.xyz - pos0.xyz;
    float distance = max(length(delta), 1e-7);
    float stretching = 1 - targetDistance / distance;
    delta = stretching * delta;
    float2 multiplier = ConstraintMultiplier(pos0, pos1);

    pos0.xyz += multiplier[0] * delta * stiffness;
    pos1.xyz -= multiplier[1] * delta * stiffness;
}

void CalcIndicesInVertexLevelTotal(uint local_id, uint group_id, inout uint globalStrandIndex, inout uint localStrandIndex, inout uint globalVertexIndex, inout uint localVertexIndex, inout uint numVerticesInTheStrand, inout uint indexForSharedMem, inout uint strandType)
{
    indexForSharedMem = local_id;
    numVerticesInTheStrand = (THREAD_GROUP_SIZE / g_NumOfStrandsPerThreadGroup);

    localStrandIndex = local_id % g_NumOfStrandsPerThreadGroup;
    globalStrandIndex = group_id * g_NumOfStrandsPerThreadGroup + localStrandIndex;
    localVertexIndex = (local_id - localStrandIndex) / g_NumOfStrandsPerThreadGroup;

    strandType = GetStrandType(globalStrandIndex);
    globalVertexIndex = globalStrandIndex * numVerticesInTheStrand + localVertexIndex;
}

void CalcIndicesInVertexLevelMaster(uint local_id, uint group_id, inout uint globalStrandIndex, inout uint localStrandIndex, inout uint globalVertexIndex, inout uint localVertexIndex, inout uint numVerticesInTheStrand, inout uint indexForSharedMem, inout uint strandType)
{
    indexForSharedMem = local_id;
    numVerticesInTheStrand = (THREAD_GROUP_SIZE / g_NumOfStrandsPerThreadGroup);

    localStrandIndex = local_id % g_NumOfStrandsPerThreadGroup;
    globalStrandIndex = group_id * g_NumOfStrandsPerThreadGroup + localStrandIndex;
    globalStrandIndex *= (g_NumFollowHairsPerGuideHair+1);
    localVertexIndex = (local_id - localStrandIndex) / g_NumOfStrandsPerThreadGroup;

    strandType = GetStrandType(globalStrandIndex);
    globalVertexIndex = globalStrandIndex * numVerticesInTheStrand + localVertexIndex;
}

void CalcIndicesInStrandLevelTotal(uint local_id, uint group_id, inout uint globalStrandIndex, inout uint numVerticesInTheStrand, inout uint globalRootVertexIndex, inout uint strandType)
{
    globalStrandIndex = THREAD_GROUP_SIZE*group_id + local_id;
    numVerticesInTheStrand = (THREAD_GROUP_SIZE / g_NumOfStrandsPerThreadGroup);
    strandType = GetStrandType(globalStrandIndex);
    globalRootVertexIndex = globalStrandIndex * numVerticesInTheStrand;
}

void CalcIndicesInStrandLevelMaster(uint local_id, uint group_id, inout uint globalStrandIndex, inout uint numVerticesInTheStrand, inout uint globalRootVertexIndex, inout uint strandType)
{
    globalStrandIndex = THREAD_GROUP_SIZE*group_id + local_id;
    globalStrandIndex *= (g_NumFollowHairsPerGuideHair+1);
    numVerticesInTheStrand = (THREAD_GROUP_SIZE / g_NumOfStrandsPerThreadGroup);
    strandType = GetStrandType(globalStrandIndex);
    globalRootVertexIndex = globalStrandIndex * numVerticesInTheStrand;
}

//--------------------------------------------------------------------------------------
//
//  Integrate
//
// Verlet integration for calculating the new position based on exponential decay to move
// from the current position towards an approximated extrapolation point based
// on the velocity between the two last points and influenced by gravity force.
//--------------------------------------------------------------------------------------
float3 Integrate(float3 curPosition, float3 oldPosition, float3 initialPos, float dampingCoeff = 1.0f)
{
    float3 force = g_GravityMagnitude * float3(0, 0, -1.0f);
    // float decay = exp(-g_TimeStep/decayTime)
    float decay = exp(-dampingCoeff * g_TimeStep * 60.0f);
    return curPosition + decay * (curPosition - oldPosition) + force * g_TimeStep * g_TimeStep;
}


struct CollisionCapsule
{
    float4 p0; // xyz = position of capsule 0, w = radius 0
    float4 p1; // xyz = position of capsule 1, w = radius 1
};

//--------------------------------------------------------------------------------------
//
//  CapsuleCollision
//
//  Moves the position based on collision with capsule
//
//--------------------------------------------------------------------------------------
bool CapsuleCollision(float4 curPosition, float4 oldPosition, inout float3 newPosition, CollisionCapsule cc, float friction = 0.4f)
{
    const float radius0 = cc.p0.w;
    const float radius1 = cc.p1.w;
    newPosition = curPosition.xyz;

    if ( !IsMovable(curPosition) )
        return false;

    float3 segment = cc.p1.xyz - cc.p0.xyz;
    float3 delta0 = curPosition.xyz - cc.p0.xyz;
    float3 delta1 = cc.p1.xyz - curPosition.xyz;

    float dist0 = dot(delta0, segment);
    float dist1 = dot(delta1, segment);

    // colliding with sphere 1
    if (dist0 < 0.f )
    {
        if ( dot(delta0, delta0) < radius0 * radius0)
        {
            float3 n = normalize(delta0);
            newPosition = radius0 * n + cc.p0.xyz;
            return true;
        }

        return false;
    }

    // colliding with sphere 2
    if (dist1 < 0.f )
    {
        if ( dot(delta1, delta1) < radius1 * radius1)
        {
            float3 n = normalize(-delta1);
            newPosition = radius1 * n + cc.p1.xyz;
            return true;
        }

        return false;
    }

    // colliding with middle cylinder
    float3 x = (dist0 * cc.p1.xyz + dist1 * cc.p0.xyz) / (dist0 + dist1);
    float3 delta = curPosition.xyz - x;

    float radius_at_x = (dist0 * radius1 + dist1 * radius0) / (dist0 + dist1);

    if ( dot(delta, delta) < radius_at_x * radius_at_x)
    {
        float3 n = normalize(delta);
        float3 vec = curPosition.xyz - oldPosition.xyz;
        float3 segN = normalize(segment);
        float3 vecTangent = dot(vec, segN) * segN;
        float3 vecNormal = vec - vecTangent;
        newPosition = oldPosition.xyz + friction * vecTangent + (vecNormal + radius_at_x * n - delta);
        return true;
    }

    return false;
}

float3 ApplyVertexBoneSkinning(float3 vertexPos, BoneSkinningData skinningData, inout float4 bone_quat)
{
    float3 newVertexPos;

#if TRESSFX_DQ
    {
        // weighted rotation part of dual quaternion
        float4 nq = g_BoneSkinningDQ[skinningData.boneIndex.x * 2] * skinningData.boneWeight.x + 
            g_BoneSkinningDQ[skinningData.boneIndex.y * 2] * skinningData.boneWeight.y +
            g_BoneSkinningDQ[skinningData.boneIndex.z * 2] * skinningData.boneWeight.z + 
            g_BoneSkinningDQ[skinningData.boneIndex.w * 2] * skinningData.boneWeight.w;

        // weighted tranlation part of dual quaternion
        float4 dq = g_BoneSkinningDQ[skinningData.boneIndex.x * 2 + 1] * skinningData.boneWeight.x + 
            g_BoneSkinningDQ[skinningData.boneIndex.y * 2 + 1] * skinningData.boneWeight.y +
            g_BoneSkinningDQ[skinningData.boneIndex.z * 2 + 1] * skinningData.boneWeight.z + 
            g_BoneSkinningDQ[skinningData.boneIndex.w * 2 + 1] * skinningData.boneWeight.w;

        float len = rsqrt(dot(nq, nq));
        nq *= len;
        dq *= len;

        bone_quat = nq;

        //convert translation part of dual quaternion to translation vector:
        float3 translation = (nq.w*dq.xyz - dq.w*nq.xyz + cross(nq.xyz, dq.xyz)) * 2;

        newVertexPos = MultQuaternionAndVector(nq, vertexPos) + translation.xyz;
    }
#else
    {
        // Interpolate world space bone matrices using weights. 
        row_major float4x4 bone_matrix = g_BoneSkinningMatrix[skinningData.boneIndex[0]] * skinningData.boneWeight[0];
        float weight_sum = skinningData.boneWeight[0];

        for (int i = 1; i < 4; i++)
        {
            if (skinningData.boneWeight[i] > 0)
            {
                bone_matrix += g_BoneSkinningMatrix[skinningData.boneIndex[i]] * skinningData.boneWeight[i];
                weight_sum += skinningData.boneWeight[i];
            }
        }

        bone_matrix /= weight_sum;
        bone_quat = MakeQuaternion(bone_matrix);

        newVertexPos = mul(float4(vertexPos, 1), bone_matrix).xyz;
    }
#endif

    return newVertexPos;
}


//--------------------------------------------------------------------------------------
//
//  UpdateFinalVertexPositions
//
//  Updates the  hair vertex positions based on the physics simulation
//
//--------------------------------------------------------------------------------------
void UpdateFinalVertexPositions(float4 oldPosition, float4 newPosition, int globalVertexIndex)
{
    SetSharedPrevPrevPosition(globalVertexIndex, GetSharedPrevPosition(globalVertexIndex));
    SetSharedPrevPosition(globalVertexIndex, oldPosition);
    SetSharedPosition(globalVertexIndex, newPosition);
}
//--------------------------------------------------------------------------------------