kannonier8/blender
1
0
Fork 0
forked from bartvdbraak/blender
Code Pull Requests Activity
c47d669f24
blender/intern/cycles/kernel/bvh/bvh_nodes.h
Brecht Van Lommel 7a92b8820b Cycles: remove hair minimum width support. 
This never really worked as it was supposed to. The main goal of this is to
turn noise from sampling tiny hairs into multiple layers of transparency that
do not need to be sampled stochastically. However the implementation of this
worked by randomly discarding hair intersections in BVH traversal, which
defeats the purpose.

If it ever comes back, it's best implemented outside the kernel as a preprocess
that changes hair radius before BVH building. This would also make it work with
Embree, where it's not supported now. But it's not so clear anymore that with
many AA samples and GPU rendering this feature is as helpful as it once was for
CPU raytracers with few AA samples.

The benefit of removing this feature is improved hair ray tracing performance,
tested on NVIDIA Titan Xp:

bmw27: +0.37%
classroom: +0.26%
fishy_cat: -7.36%
koro: -12.98%
pabellon: -0.12%

Differential Revision: https://developer.blender.org/D4532
2019-04-24 14:39:47 +02:00

273 lines
12 KiB
C
Raw Blame History

/*
* Copyright 2011-2016, Blender Foundation.
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
// TODO(sergey): Look into avoid use of full Transform and use 3x3 matrix and
// 3-vector which might be faster.
ccl_device_forceinline Transform bvh_unaligned_node_fetch_space(KernelGlobals *kg,
int node_addr,
int child)
{
Transform space;
const int child_addr = node_addr + child * 3;
space.x = kernel_tex_fetch(__bvh_nodes, child_addr + 1);
space.y = kernel_tex_fetch(__bvh_nodes, child_addr + 2);
space.z = kernel_tex_fetch(__bvh_nodes, child_addr + 3);
return space;
}
#if !defined(__KERNEL_SSE2__)
ccl_device_forceinline int bvh_aligned_node_intersect(KernelGlobals *kg,
const float3 P,
const float3 idir,
const float t,
const int node_addr,
const uint visibility,
float dist[2])
{
/* fetch node data */
float4 cnodes = kernel_tex_fetch(__bvh_nodes, node_addr + 0);
float4 node0 = kernel_tex_fetch(__bvh_nodes, node_addr + 1);
float4 node1 = kernel_tex_fetch(__bvh_nodes, node_addr + 2);
float4 node2 = kernel_tex_fetch(__bvh_nodes, node_addr + 3);
/* intersect ray against child nodes */
float c0lox = (node0.x - P.x) * idir.x;
float c0hix = (node0.z - P.x) * idir.x;
float c0loy = (node1.x - P.y) * idir.y;
float c0hiy = (node1.z - P.y) * idir.y;
float c0loz = (node2.x - P.z) * idir.z;
float c0hiz = (node2.z - P.z) * idir.z;
float c0min = max4(0.0f, min(c0lox, c0hix), min(c0loy, c0hiy), min(c0loz, c0hiz));
float c0max = min4(t, max(c0lox, c0hix), max(c0loy, c0hiy), max(c0loz, c0hiz));
float c1lox = (node0.y - P.x) * idir.x;
float c1hix = (node0.w - P.x) * idir.x;
float c1loy = (node1.y - P.y) * idir.y;
float c1hiy = (node1.w - P.y) * idir.y;
float c1loz = (node2.y - P.z) * idir.z;
float c1hiz = (node2.w - P.z) * idir.z;
float c1min = max4(0.0f, min(c1lox, c1hix), min(c1loy, c1hiy), min(c1loz, c1hiz));
float c1max = min4(t, max(c1lox, c1hix), max(c1loy, c1hiy), max(c1loz, c1hiz));
dist[0] = c0min;
dist[1] = c1min;
# ifdef __VISIBILITY_FLAG__
/* this visibility test gives a 5% performance hit, how to solve? */
return (((c0max >= c0min) && (__float_as_uint(cnodes.x) & visibility)) ? 1 : 0) |
(((c1max >= c1min) && (__float_as_uint(cnodes.y) & visibility)) ? 2 : 0);
# else
return ((c0max >= c0min) ? 1 : 0) | ((c1max >= c1min) ? 2 : 0);
# endif
}
ccl_device_forceinline bool bvh_unaligned_node_intersect_child(KernelGlobals *kg,
const float3 P,
const float3 dir,
const float t,
int node_addr,
int child,
float dist[2])
{
Transform space = bvh_unaligned_node_fetch_space(kg, node_addr, child);
float3 aligned_dir = transform_direction(&space, dir);
float3 aligned_P = transform_point(&space, P);
float3 nrdir = -bvh_inverse_direction(aligned_dir);
float3 lower_xyz = aligned_P * nrdir;
float3 upper_xyz = lower_xyz - nrdir;
const float near_x = min(lower_xyz.x, upper_xyz.x);
const float near_y = min(lower_xyz.y, upper_xyz.y);
const float near_z = min(lower_xyz.z, upper_xyz.z);
const float far_x = max(lower_xyz.x, upper_xyz.x);
const float far_y = max(lower_xyz.y, upper_xyz.y);
const float far_z = max(lower_xyz.z, upper_xyz.z);
const float tnear = max4(0.0f, near_x, near_y, near_z);
const float tfar = min4(t, far_x, far_y, far_z);
*dist = tnear;
return tnear <= tfar;
}
ccl_device_forceinline int bvh_unaligned_node_intersect(KernelGlobals *kg,
const float3 P,
const float3 dir,
const float3 idir,
const float t,
const int node_addr,
const uint visibility,
float dist[2])
{
int mask = 0;
float4 cnodes = kernel_tex_fetch(__bvh_nodes, node_addr + 0);
if (bvh_unaligned_node_intersect_child(kg, P, dir, t, node_addr, 0, &dist[0])) {
# ifdef __VISIBILITY_FLAG__
if ((__float_as_uint(cnodes.x) & visibility))
# endif
{
mask |= 1;
}
}
if (bvh_unaligned_node_intersect_child(kg, P, dir, t, node_addr, 1, &dist[1])) {
# ifdef __VISIBILITY_FLAG__
if ((__float_as_uint(cnodes.y) & visibility))
# endif
{
mask |= 2;
}
}
return mask;
}
ccl_device_forceinline int bvh_node_intersect(KernelGlobals *kg,
const float3 P,
const float3 dir,
const float3 idir,
const float t,
const int node_addr,
const uint visibility,
float dist[2])
{
float4 node = kernel_tex_fetch(__bvh_nodes, node_addr);
if (__float_as_uint(node.x) & PATH_RAY_NODE_UNALIGNED) {
return bvh_unaligned_node_intersect(kg, P, dir, idir, t, node_addr, visibility, dist);
}
else {
return bvh_aligned_node_intersect(kg, P, idir, t, node_addr, visibility, dist);
}
}
#else /* !defined(__KERNEL_SSE2__) */
int ccl_device_forceinline bvh_aligned_node_intersect(KernelGlobals *kg,
const float3 &P,
const float3 &dir,
const ssef &tsplat,
const ssef Psplat[3],
const ssef idirsplat[3],
const shuffle_swap_t shufflexyz[3],
const int node_addr,
const uint visibility,
float dist[2])
{
/* Intersect two child bounding boxes, SSE3 version adapted from Embree */
const ssef pn = cast(ssei(0, 0, 0x80000000, 0x80000000));
/* fetch node data */
const ssef *bvh_nodes = (ssef *)kg->__bvh_nodes.data + node_addr;
/* intersect ray against child nodes */
const ssef tminmaxx = (shuffle_swap(bvh_nodes[1], shufflexyz[0]) - Psplat[0]) * idirsplat[0];
const ssef tminmaxy = (shuffle_swap(bvh_nodes[2], shufflexyz[1]) - Psplat[1]) * idirsplat[1];
const ssef tminmaxz = (shuffle_swap(bvh_nodes[3], shufflexyz[2]) - Psplat[2]) * idirsplat[2];
/* calculate { c0min, c1min, -c0max, -c1max} */
ssef minmax = max(max(tminmaxx, tminmaxy), max(tminmaxz, tsplat));
const ssef tminmax = minmax ^ pn;
const sseb lrhit = tminmax <= shuffle<2, 3, 0, 1>(tminmax);
dist[0] = tminmax[0];
dist[1] = tminmax[1];
int mask = movemask(lrhit);
# ifdef __VISIBILITY_FLAG__
/* this visibility test gives a 5% performance hit, how to solve? */
float4 cnodes = kernel_tex_fetch(__bvh_nodes, node_addr + 0);
int cmask = (((mask & 1) && (__float_as_uint(cnodes.x) & visibility)) ? 1 : 0) |
(((mask & 2) && (__float_as_uint(cnodes.y) & visibility)) ? 2 : 0);
return cmask;
# else
return mask & 3;
# endif
}
ccl_device_forceinline int bvh_unaligned_node_intersect(KernelGlobals *kg,
const float3 P,
const float3 dir,
const ssef &isect_near,
const ssef &isect_far,
const int node_addr,
const uint visibility,
float dist[2])
{
Transform space0 = bvh_unaligned_node_fetch_space(kg, node_addr, 0);
Transform space1 = bvh_unaligned_node_fetch_space(kg, node_addr, 1);
float3 aligned_dir0 = transform_direction(&space0, dir),
aligned_dir1 = transform_direction(&space1, dir);
float3 aligned_P0 = transform_point(&space0, P), aligned_P1 = transform_point(&space1, P);
float3 nrdir0 = -bvh_inverse_direction(aligned_dir0),
nrdir1 = -bvh_inverse_direction(aligned_dir1);
ssef lower_x = ssef(aligned_P0.x * nrdir0.x, aligned_P1.x * nrdir1.x, 0.0f, 0.0f),
lower_y = ssef(aligned_P0.y * nrdir0.y, aligned_P1.y * nrdir1.y, 0.0f, 0.0f),
lower_z = ssef(aligned_P0.z * nrdir0.z, aligned_P1.z * nrdir1.z, 0.0f, 0.0f);
ssef upper_x = lower_x - ssef(nrdir0.x, nrdir1.x, 0.0f, 0.0f),
upper_y = lower_y - ssef(nrdir0.y, nrdir1.y, 0.0f, 0.0f),
upper_z = lower_z - ssef(nrdir0.z, nrdir1.z, 0.0f, 0.0f);
ssef tnear_x = min(lower_x, upper_x);
ssef tnear_y = min(lower_y, upper_y);
ssef tnear_z = min(lower_z, upper_z);
ssef tfar_x = max(lower_x, upper_x);
ssef tfar_y = max(lower_y, upper_y);
ssef tfar_z = max(lower_z, upper_z);
const ssef tnear = max4(isect_near, tnear_x, tnear_y, tnear_z);
const ssef tfar = min4(isect_far, tfar_x, tfar_y, tfar_z);
sseb vmask = tnear <= tfar;
dist[0] = tnear.f[0];
dist[1] = tnear.f[1];
int mask = (int)movemask(vmask);
# ifdef __VISIBILITY_FLAG__
/* this visibility test gives a 5% performance hit, how to solve? */
float4 cnodes = kernel_tex_fetch(__bvh_nodes, node_addr + 0);
int cmask = (((mask & 1) && (__float_as_uint(cnodes.x) & visibility)) ? 1 : 0) |
(((mask & 2) && (__float_as_uint(cnodes.y) & visibility)) ? 2 : 0);
return cmask;
# else
return mask & 3;
# endif
}
ccl_device_forceinline int bvh_node_intersect(KernelGlobals *kg,
const float3 &P,
const float3 &dir,
const ssef &isect_near,
const ssef &isect_far,
const ssef &tsplat,
const ssef Psplat[3],
const ssef idirsplat[3],
const shuffle_swap_t shufflexyz[3],
const int node_addr,
const uint visibility,
float dist[2])
{
float4 node = kernel_tex_fetch(__bvh_nodes, node_addr);
if (__float_as_uint(node.x) & PATH_RAY_NODE_UNALIGNED) {
return bvh_unaligned_node_intersect(
kg, P, dir, isect_near, isect_far, node_addr, visibility, dist);
}
else {
return bvh_aligned_node_intersect(
kg, P, dir, tsplat, Psplat, idirsplat, shufflexyz, node_addr, visibility, dist);
}
}
#endif /* !defined(__KERNEL_SSE2__) */
View Git Blame Copy Permalink