2022-02-11 13:53:21 +01:00
|
|
|
/* SPDX-License-Identifier: Apache-2.0
|
|
|
|
|
* Copyright 2011-2022 Blender Foundation */
|
Cycles: Add multi-scattering, energy-conserving GGX as an option to the Glossy, Anisotropic and Glass BSDFs
This commit adds a new distribution to the Glossy, Anisotropic and Glass BSDFs that implements the
multiple-scattering microfacet model described in the paper "Multiple-Scattering Microfacet BSDFs with the Smith Model".
Essentially, the improvement is that unlike classical GGX, which only models single scattering and assumes
the contribution of multiple bounces to be zero, this new model performs a random walk on the microsurface until
the ray leaves it again, which ensures perfect energy conservation.
In practise, this means that the "darkening problem" - GGX materials becoming darker with increasing
roughness - is solved in a physically correct and efficient way.
The downside of this model is that it has no (known) analytic expression for evalation. However, it can be
evaluated stochastically, and although the correct PDF isn't known either, the properties of MIS and the
balance heuristic guarantee an unbiased result at the cost of slightly higher noise.
Reviewers: dingto, #cycles, brecht
Reviewed By: dingto, #cycles, brecht
Subscribers: bliblubli, ace_dragon, gregzaal, brecht, harvester, dingto, marcog, swerner, jtheninja, Blendify, nutel
Differential Revision: https://developer.blender.org/D2002
2016-06-23 22:56:43 +02:00
|
|
|
|
|
|
|
|
/* Evaluate the BSDF from wi to wo.
|
|
|
|
|
* Evaluation is split into the analytical single-scattering BSDF and the multi-scattering BSDF,
|
2019-05-01 21:14:11 +10:00
|
|
|
* which is evaluated stochastically through a random walk. At each bounce (except for the first
|
|
|
|
|
* one), the amount of reflection from here towards wo is evaluated before bouncing again.
|
Cycles: Add multi-scattering, energy-conserving GGX as an option to the Glossy, Anisotropic and Glass BSDFs
This commit adds a new distribution to the Glossy, Anisotropic and Glass BSDFs that implements the
multiple-scattering microfacet model described in the paper "Multiple-Scattering Microfacet BSDFs with the Smith Model".
Essentially, the improvement is that unlike classical GGX, which only models single scattering and assumes
the contribution of multiple bounces to be zero, this new model performs a random walk on the microsurface until
the ray leaves it again, which ensures perfect energy conservation.
In practise, this means that the "darkening problem" - GGX materials becoming darker with increasing
roughness - is solved in a physically correct and efficient way.
The downside of this model is that it has no (known) analytic expression for evalation. However, it can be
evaluated stochastically, and although the correct PDF isn't known either, the properties of MIS and the
balance heuristic guarantee an unbiased result at the cost of slightly higher noise.
Reviewers: dingto, #cycles, brecht
Reviewed By: dingto, #cycles, brecht
Subscribers: bliblubli, ace_dragon, gregzaal, brecht, harvester, dingto, marcog, swerner, jtheninja, Blendify, nutel
Differential Revision: https://developer.blender.org/D2002
2016-06-23 22:56:43 +02:00
|
|
|
*
|
2019-05-01 21:14:11 +10:00
|
|
|
* Because of the random walk, the evaluation is not deterministic, but its expected value is equal
|
|
|
|
|
* to the correct BSDF, which is enough for Monte-Carlo rendering. The PDF also can't be determined
|
|
|
|
|
* analytically, so the single-scattering PDF plus a diffuse term to account for the
|
|
|
|
|
* multi-scattered energy is used. In combination with MIS, that is enough to produce an unbiased
|
|
|
|
|
* result, although the balance heuristic isn't necessarily optimal anymore.
|
Cycles: Add multi-scattering, energy-conserving GGX as an option to the Glossy, Anisotropic and Glass BSDFs
This commit adds a new distribution to the Glossy, Anisotropic and Glass BSDFs that implements the
multiple-scattering microfacet model described in the paper "Multiple-Scattering Microfacet BSDFs with the Smith Model".
Essentially, the improvement is that unlike classical GGX, which only models single scattering and assumes
the contribution of multiple bounces to be zero, this new model performs a random walk on the microsurface until
the ray leaves it again, which ensures perfect energy conservation.
In practise, this means that the "darkening problem" - GGX materials becoming darker with increasing
roughness - is solved in a physically correct and efficient way.
The downside of this model is that it has no (known) analytic expression for evalation. However, it can be
evaluated stochastically, and although the correct PDF isn't known either, the properties of MIS and the
balance heuristic guarantee an unbiased result at the cost of slightly higher noise.
Reviewers: dingto, #cycles, brecht
Reviewed By: dingto, #cycles, brecht
Subscribers: bliblubli, ace_dragon, gregzaal, brecht, harvester, dingto, marcog, swerner, jtheninja, Blendify, nutel
Differential Revision: https://developer.blender.org/D2002
2016-06-23 22:56:43 +02:00
|
|
|
*/
|
2022-07-29 13:41:37 +02:00
|
|
|
ccl_device_forceinline Spectrum MF_FUNCTION_FULL_NAME(mf_eval)(float3 wi,
|
|
|
|
|
float3 wo,
|
|
|
|
|
const bool wo_outside,
|
|
|
|
|
const Spectrum color,
|
|
|
|
|
const float alpha_x,
|
|
|
|
|
const float alpha_y,
|
|
|
|
|
ccl_private uint *lcg_state,
|
|
|
|
|
const float eta,
|
|
|
|
|
bool use_fresnel,
|
|
|
|
|
const Spectrum cspec0)
|
Cycles: Add multi-scattering, energy-conserving GGX as an option to the Glossy, Anisotropic and Glass BSDFs
This commit adds a new distribution to the Glossy, Anisotropic and Glass BSDFs that implements the
multiple-scattering microfacet model described in the paper "Multiple-Scattering Microfacet BSDFs with the Smith Model".
Essentially, the improvement is that unlike classical GGX, which only models single scattering and assumes
the contribution of multiple bounces to be zero, this new model performs a random walk on the microsurface until
the ray leaves it again, which ensures perfect energy conservation.
In practise, this means that the "darkening problem" - GGX materials becoming darker with increasing
roughness - is solved in a physically correct and efficient way.
The downside of this model is that it has no (known) analytic expression for evalation. However, it can be
evaluated stochastically, and although the correct PDF isn't known either, the properties of MIS and the
balance heuristic guarantee an unbiased result at the cost of slightly higher noise.
Reviewers: dingto, #cycles, brecht
Reviewed By: dingto, #cycles, brecht
Subscribers: bliblubli, ace_dragon, gregzaal, brecht, harvester, dingto, marcog, swerner, jtheninja, Blendify, nutel
Differential Revision: https://developer.blender.org/D2002
2016-06-23 22:56:43 +02:00
|
|
|
{
|
2019-05-01 21:14:11 +10:00
|
|
|
/* Evaluating for a shallower incoming direction produces less noise, and the properties of the
|
|
|
|
|
* BSDF guarantee reciprocity. */
|
Cycles: Add multi-scattering, energy-conserving GGX as an option to the Glossy, Anisotropic and Glass BSDFs
This commit adds a new distribution to the Glossy, Anisotropic and Glass BSDFs that implements the
multiple-scattering microfacet model described in the paper "Multiple-Scattering Microfacet BSDFs with the Smith Model".
Essentially, the improvement is that unlike classical GGX, which only models single scattering and assumes
the contribution of multiple bounces to be zero, this new model performs a random walk on the microsurface until
the ray leaves it again, which ensures perfect energy conservation.
In practise, this means that the "darkening problem" - GGX materials becoming darker with increasing
roughness - is solved in a physically correct and efficient way.
The downside of this model is that it has no (known) analytic expression for evalation. However, it can be
evaluated stochastically, and although the correct PDF isn't known either, the properties of MIS and the
balance heuristic guarantee an unbiased result at the cost of slightly higher noise.
Reviewers: dingto, #cycles, brecht
Reviewed By: dingto, #cycles, brecht
Subscribers: bliblubli, ace_dragon, gregzaal, brecht, harvester, dingto, marcog, swerner, jtheninja, Blendify, nutel
Differential Revision: https://developer.blender.org/D2002
2016-06-23 22:56:43 +02:00
|
|
|
bool swapped = false;
|
|
|
|
|
#ifdef MF_MULTI_GLASS
|
|
|
|
|
if (wi.z * wo.z < 0.0f) {
|
|
|
|
|
/* Glass transmission is a special case and requires the directions to change hemisphere. */
|
|
|
|
|
if (-wo.z < wi.z) {
|
|
|
|
|
swapped = true;
|
|
|
|
|
float3 tmp = -wo;
|
|
|
|
|
wo = -wi;
|
|
|
|
|
wi = tmp;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
else
|
|
|
|
|
#endif
|
|
|
|
|
if (wo.z < wi.z) {
|
|
|
|
|
swapped = true;
|
|
|
|
|
float3 tmp = wo;
|
|
|
|
|
wo = wi;
|
|
|
|
|
wi = tmp;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (wi.z < 1e-5f || (wo.z < 1e-5f && wo_outside) || (wo.z > -1e-5f && !wo_outside))
|
2022-07-29 13:41:37 +02:00
|
|
|
return zero_spectrum();
|
Cycles: Add multi-scattering, energy-conserving GGX as an option to the Glossy, Anisotropic and Glass BSDFs
This commit adds a new distribution to the Glossy, Anisotropic and Glass BSDFs that implements the
multiple-scattering microfacet model described in the paper "Multiple-Scattering Microfacet BSDFs with the Smith Model".
Essentially, the improvement is that unlike classical GGX, which only models single scattering and assumes
the contribution of multiple bounces to be zero, this new model performs a random walk on the microsurface until
the ray leaves it again, which ensures perfect energy conservation.
In practise, this means that the "darkening problem" - GGX materials becoming darker with increasing
roughness - is solved in a physically correct and efficient way.
The downside of this model is that it has no (known) analytic expression for evalation. However, it can be
evaluated stochastically, and although the correct PDF isn't known either, the properties of MIS and the
balance heuristic guarantee an unbiased result at the cost of slightly higher noise.
Reviewers: dingto, #cycles, brecht
Reviewed By: dingto, #cycles, brecht
Subscribers: bliblubli, ace_dragon, gregzaal, brecht, harvester, dingto, marcog, swerner, jtheninja, Blendify, nutel
Differential Revision: https://developer.blender.org/D2002
2016-06-23 22:56:43 +02:00
|
|
|
|
|
|
|
|
const float2 alpha = make_float2(alpha_x, alpha_y);
|
|
|
|
|
|
|
|
|
|
float lambda_r = mf_lambda(-wi, alpha);
|
|
|
|
|
float shadowing_lambda = mf_lambda(wo_outside ? wo : -wo, alpha);
|
|
|
|
|
|
|
|
|
|
/* Analytically compute single scattering for lower noise. */
|
2022-07-29 13:41:37 +02:00
|
|
|
Spectrum eval;
|
|
|
|
|
Spectrum throughput = one_spectrum();
|
2017-05-14 21:17:32 +02:00
|
|
|
const float3 wh = normalize(wi + wo);
|
Cycles: Add multi-scattering, energy-conserving GGX as an option to the Glossy, Anisotropic and Glass BSDFs
This commit adds a new distribution to the Glossy, Anisotropic and Glass BSDFs that implements the
multiple-scattering microfacet model described in the paper "Multiple-Scattering Microfacet BSDFs with the Smith Model".
Essentially, the improvement is that unlike classical GGX, which only models single scattering and assumes
the contribution of multiple bounces to be zero, this new model performs a random walk on the microsurface until
the ray leaves it again, which ensures perfect energy conservation.
In practise, this means that the "darkening problem" - GGX materials becoming darker with increasing
roughness - is solved in a physically correct and efficient way.
The downside of this model is that it has no (known) analytic expression for evalation. However, it can be
evaluated stochastically, and although the correct PDF isn't known either, the properties of MIS and the
balance heuristic guarantee an unbiased result at the cost of slightly higher noise.
Reviewers: dingto, #cycles, brecht
Reviewed By: dingto, #cycles, brecht
Subscribers: bliblubli, ace_dragon, gregzaal, brecht, harvester, dingto, marcog, swerner, jtheninja, Blendify, nutel
Differential Revision: https://developer.blender.org/D2002
2016-06-23 22:56:43 +02:00
|
|
|
#ifdef MF_MULTI_GLASS
|
|
|
|
|
eval = mf_eval_phase_glass(-wi, lambda_r, wo, wo_outside, alpha, eta);
|
|
|
|
|
if (wo_outside)
|
|
|
|
|
eval *= -lambda_r / (shadowing_lambda - lambda_r);
|
|
|
|
|
else
|
|
|
|
|
eval *= -lambda_r * beta(-lambda_r, shadowing_lambda + 1.0f);
|
|
|
|
|
#else /* MF_MULTI_GLOSSY */
|
|
|
|
|
const float G2 = 1.0f / (1.0f - (lambda_r + 1.0f) + shadowing_lambda);
|
|
|
|
|
float val = G2 * 0.25f / wi.z;
|
|
|
|
|
if (alpha.x == alpha.y)
|
|
|
|
|
val *= D_ggx(wh, alpha.x);
|
|
|
|
|
else
|
|
|
|
|
val *= D_ggx_aniso(wh, alpha);
|
2022-07-29 13:41:37 +02:00
|
|
|
eval = make_spectrum(val);
|
2017-05-14 21:17:32 +02:00
|
|
|
#endif
|
2017-04-18 11:43:09 +02:00
|
|
|
|
|
|
|
|
float F0 = fresnel_dielectric_cos(1.0f, eta);
|
|
|
|
|
if (use_fresnel) {
|
|
|
|
|
throughput = interpolate_fresnel_color(wi, wh, eta, F0, cspec0);
|
|
|
|
|
|
2017-05-14 21:17:32 +02:00
|
|
|
eval *= throughput;
|
2017-04-18 11:43:09 +02:00
|
|
|
}
|
Cycles: Add multi-scattering, energy-conserving GGX as an option to the Glossy, Anisotropic and Glass BSDFs
This commit adds a new distribution to the Glossy, Anisotropic and Glass BSDFs that implements the
multiple-scattering microfacet model described in the paper "Multiple-Scattering Microfacet BSDFs with the Smith Model".
Essentially, the improvement is that unlike classical GGX, which only models single scattering and assumes
the contribution of multiple bounces to be zero, this new model performs a random walk on the microsurface until
the ray leaves it again, which ensures perfect energy conservation.
In practise, this means that the "darkening problem" - GGX materials becoming darker with increasing
roughness - is solved in a physically correct and efficient way.
The downside of this model is that it has no (known) analytic expression for evalation. However, it can be
evaluated stochastically, and although the correct PDF isn't known either, the properties of MIS and the
balance heuristic guarantee an unbiased result at the cost of slightly higher noise.
Reviewers: dingto, #cycles, brecht
Reviewed By: dingto, #cycles, brecht
Subscribers: bliblubli, ace_dragon, gregzaal, brecht, harvester, dingto, marcog, swerner, jtheninja, Blendify, nutel
Differential Revision: https://developer.blender.org/D2002
2016-06-23 22:56:43 +02:00
|
|
|
|
|
|
|
|
float3 wr = -wi;
|
|
|
|
|
float hr = 1.0f;
|
|
|
|
|
float C1_r = 1.0f;
|
|
|
|
|
float G1_r = 0.0f;
|
|
|
|
|
bool outside = true;
|
|
|
|
|
|
|
|
|
|
for (int order = 0; order < 10; order++) {
|
2017-07-23 22:43:55 +02:00
|
|
|
/* Sample microfacet height. */
|
Cycles: Kernel address space changes for MSL
This is the first of a sequence of changes to support compiling Cycles kernels as MSL (Metal Shading Language) in preparation for a Metal GPU device implementation.
MSL requires that all pointer types be declared with explicit address space attributes (device, thread, etc...). There is already precedent for this with Cycles' address space macros (ccl_global, ccl_private, etc...), therefore the first step of MSL-enablement is to apply these consistently. Line-for-line this represents the largest change required to enable MSL. Applying this change first will simplify future patches as well as offering the emergent benefit of enhanced descriptiveness.
The vast majority of deltas in this patch fall into one of two cases:
- Ensuring ccl_private is specified for thread-local pointer types
- Ensuring ccl_global is specified for device-wide pointer types
Additionally, the ccl_addr_space qualifier can be removed. Prior to Cycles X, ccl_addr_space was used as a context-dependent address space qualifier, but now it is either redundant (e.g. in struct typedefs), or can be replaced by ccl_global in the case of pointer types. Associated function variants (e.g. lcg_step_float_addrspace) are also redundant.
In cases where address space qualifiers are chained with "const", this patch places the address space qualifier first. The rationale for this is that the choice of address space is likely to have the greater impact on runtime performance and overall architecture.
The final part of this patch is the addition of a metal/compat.h header. This is partially complete and will be extended in future patches, paving the way for the full Metal implementation.
Ref T92212
Reviewed By: brecht
Maniphest Tasks: T92212
Differential Revision: https://developer.blender.org/D12864
2021-10-14 13:53:40 +01:00
|
|
|
float height_rand = lcg_step_float(lcg_state);
|
2017-07-23 22:43:55 +02:00
|
|
|
if (!mf_sample_height(wr, &hr, &C1_r, &G1_r, &lambda_r, height_rand))
|
Cycles: Add multi-scattering, energy-conserving GGX as an option to the Glossy, Anisotropic and Glass BSDFs
This commit adds a new distribution to the Glossy, Anisotropic and Glass BSDFs that implements the
multiple-scattering microfacet model described in the paper "Multiple-Scattering Microfacet BSDFs with the Smith Model".
Essentially, the improvement is that unlike classical GGX, which only models single scattering and assumes
the contribution of multiple bounces to be zero, this new model performs a random walk on the microsurface until
the ray leaves it again, which ensures perfect energy conservation.
In practise, this means that the "darkening problem" - GGX materials becoming darker with increasing
roughness - is solved in a physically correct and efficient way.
The downside of this model is that it has no (known) analytic expression for evalation. However, it can be
evaluated stochastically, and although the correct PDF isn't known either, the properties of MIS and the
balance heuristic guarantee an unbiased result at the cost of slightly higher noise.
Reviewers: dingto, #cycles, brecht
Reviewed By: dingto, #cycles, brecht
Subscribers: bliblubli, ace_dragon, gregzaal, brecht, harvester, dingto, marcog, swerner, jtheninja, Blendify, nutel
Differential Revision: https://developer.blender.org/D2002
2016-06-23 22:56:43 +02:00
|
|
|
break;
|
2017-07-23 22:43:55 +02:00
|
|
|
/* Sample microfacet normal. */
|
Cycles: Kernel address space changes for MSL
This is the first of a sequence of changes to support compiling Cycles kernels as MSL (Metal Shading Language) in preparation for a Metal GPU device implementation.
MSL requires that all pointer types be declared with explicit address space attributes (device, thread, etc...). There is already precedent for this with Cycles' address space macros (ccl_global, ccl_private, etc...), therefore the first step of MSL-enablement is to apply these consistently. Line-for-line this represents the largest change required to enable MSL. Applying this change first will simplify future patches as well as offering the emergent benefit of enhanced descriptiveness.
The vast majority of deltas in this patch fall into one of two cases:
- Ensuring ccl_private is specified for thread-local pointer types
- Ensuring ccl_global is specified for device-wide pointer types
Additionally, the ccl_addr_space qualifier can be removed. Prior to Cycles X, ccl_addr_space was used as a context-dependent address space qualifier, but now it is either redundant (e.g. in struct typedefs), or can be replaced by ccl_global in the case of pointer types. Associated function variants (e.g. lcg_step_float_addrspace) are also redundant.
In cases where address space qualifiers are chained with "const", this patch places the address space qualifier first. The rationale for this is that the choice of address space is likely to have the greater impact on runtime performance and overall architecture.
The final part of this patch is the addition of a metal/compat.h header. This is partially complete and will be extended in future patches, paving the way for the full Metal implementation.
Ref T92212
Reviewed By: brecht
Maniphest Tasks: T92212
Differential Revision: https://developer.blender.org/D12864
2021-10-14 13:53:40 +01:00
|
|
|
float vndf_rand_y = lcg_step_float(lcg_state);
|
|
|
|
|
float vndf_rand_x = lcg_step_float(lcg_state);
|
2017-07-23 22:43:55 +02:00
|
|
|
float3 wm = mf_sample_vndf(-wr, alpha, vndf_rand_x, vndf_rand_y);
|
Cycles: Add multi-scattering, energy-conserving GGX as an option to the Glossy, Anisotropic and Glass BSDFs
This commit adds a new distribution to the Glossy, Anisotropic and Glass BSDFs that implements the
multiple-scattering microfacet model described in the paper "Multiple-Scattering Microfacet BSDFs with the Smith Model".
Essentially, the improvement is that unlike classical GGX, which only models single scattering and assumes
the contribution of multiple bounces to be zero, this new model performs a random walk on the microsurface until
the ray leaves it again, which ensures perfect energy conservation.
In practise, this means that the "darkening problem" - GGX materials becoming darker with increasing
roughness - is solved in a physically correct and efficient way.
The downside of this model is that it has no (known) analytic expression for evalation. However, it can be
evaluated stochastically, and although the correct PDF isn't known either, the properties of MIS and the
balance heuristic guarantee an unbiased result at the cost of slightly higher noise.
Reviewers: dingto, #cycles, brecht
Reviewed By: dingto, #cycles, brecht
Subscribers: bliblubli, ace_dragon, gregzaal, brecht, harvester, dingto, marcog, swerner, jtheninja, Blendify, nutel
Differential Revision: https://developer.blender.org/D2002
2016-06-23 22:56:43 +02:00
|
|
|
|
2017-04-18 11:43:09 +02:00
|
|
|
#ifdef MF_MULTI_GLASS
|
|
|
|
|
if (order == 0 && use_fresnel) {
|
|
|
|
|
/* Evaluate amount of scattering towards wo on this microfacet. */
|
2022-07-29 13:41:37 +02:00
|
|
|
Spectrum phase;
|
2017-04-18 11:43:09 +02:00
|
|
|
if (outside)
|
|
|
|
|
phase = mf_eval_phase_glass(wr, lambda_r, wo, wo_outside, alpha, eta);
|
|
|
|
|
else
|
|
|
|
|
phase = mf_eval_phase_glass(wr, lambda_r, -wo, !wo_outside, alpha, 1.0f / eta);
|
|
|
|
|
|
|
|
|
|
eval = throughput * phase *
|
|
|
|
|
mf_G1(wo_outside ? wo : -wo,
|
|
|
|
|
mf_C1((outside == wo_outside) ? hr : -hr),
|
|
|
|
|
shadowing_lambda);
|
|
|
|
|
}
|
Cycles: Add multi-scattering, energy-conserving GGX as an option to the Glossy, Anisotropic and Glass BSDFs
This commit adds a new distribution to the Glossy, Anisotropic and Glass BSDFs that implements the
multiple-scattering microfacet model described in the paper "Multiple-Scattering Microfacet BSDFs with the Smith Model".
Essentially, the improvement is that unlike classical GGX, which only models single scattering and assumes
the contribution of multiple bounces to be zero, this new model performs a random walk on the microsurface until
the ray leaves it again, which ensures perfect energy conservation.
In practise, this means that the "darkening problem" - GGX materials becoming darker with increasing
roughness - is solved in a physically correct and efficient way.
The downside of this model is that it has no (known) analytic expression for evalation. However, it can be
evaluated stochastically, and although the correct PDF isn't known either, the properties of MIS and the
balance heuristic guarantee an unbiased result at the cost of slightly higher noise.
Reviewers: dingto, #cycles, brecht
Reviewed By: dingto, #cycles, brecht
Subscribers: bliblubli, ace_dragon, gregzaal, brecht, harvester, dingto, marcog, swerner, jtheninja, Blendify, nutel
Differential Revision: https://developer.blender.org/D2002
2016-06-23 22:56:43 +02:00
|
|
|
#endif
|
|
|
|
|
if (order > 0) {
|
|
|
|
|
/* Evaluate amount of scattering towards wo on this microfacet. */
|
2022-07-29 13:41:37 +02:00
|
|
|
Spectrum phase;
|
Cycles: Add multi-scattering, energy-conserving GGX as an option to the Glossy, Anisotropic and Glass BSDFs
This commit adds a new distribution to the Glossy, Anisotropic and Glass BSDFs that implements the
multiple-scattering microfacet model described in the paper "Multiple-Scattering Microfacet BSDFs with the Smith Model".
Essentially, the improvement is that unlike classical GGX, which only models single scattering and assumes
the contribution of multiple bounces to be zero, this new model performs a random walk on the microsurface until
the ray leaves it again, which ensures perfect energy conservation.
In practise, this means that the "darkening problem" - GGX materials becoming darker with increasing
roughness - is solved in a physically correct and efficient way.
The downside of this model is that it has no (known) analytic expression for evalation. However, it can be
evaluated stochastically, and although the correct PDF isn't known either, the properties of MIS and the
balance heuristic guarantee an unbiased result at the cost of slightly higher noise.
Reviewers: dingto, #cycles, brecht
Reviewed By: dingto, #cycles, brecht
Subscribers: bliblubli, ace_dragon, gregzaal, brecht, harvester, dingto, marcog, swerner, jtheninja, Blendify, nutel
Differential Revision: https://developer.blender.org/D2002
2016-06-23 22:56:43 +02:00
|
|
|
#ifdef MF_MULTI_GLASS
|
|
|
|
|
if (outside)
|
|
|
|
|
phase = mf_eval_phase_glass(wr, lambda_r, wo, wo_outside, alpha, eta);
|
|
|
|
|
else
|
|
|
|
|
phase = mf_eval_phase_glass(wr, lambda_r, -wo, !wo_outside, alpha, 1.0f / eta);
|
|
|
|
|
#else /* MF_MULTI_GLOSSY */
|
2017-05-14 21:17:32 +02:00
|
|
|
phase = mf_eval_phase_glossy(wr, lambda_r, wo, alpha) * throughput;
|
Cycles: Add multi-scattering, energy-conserving GGX as an option to the Glossy, Anisotropic and Glass BSDFs
This commit adds a new distribution to the Glossy, Anisotropic and Glass BSDFs that implements the
multiple-scattering microfacet model described in the paper "Multiple-Scattering Microfacet BSDFs with the Smith Model".
Essentially, the improvement is that unlike classical GGX, which only models single scattering and assumes
the contribution of multiple bounces to be zero, this new model performs a random walk on the microsurface until
the ray leaves it again, which ensures perfect energy conservation.
In practise, this means that the "darkening problem" - GGX materials becoming darker with increasing
roughness - is solved in a physically correct and efficient way.
The downside of this model is that it has no (known) analytic expression for evalation. However, it can be
evaluated stochastically, and although the correct PDF isn't known either, the properties of MIS and the
balance heuristic guarantee an unbiased result at the cost of slightly higher noise.
Reviewers: dingto, #cycles, brecht
Reviewed By: dingto, #cycles, brecht
Subscribers: bliblubli, ace_dragon, gregzaal, brecht, harvester, dingto, marcog, swerner, jtheninja, Blendify, nutel
Differential Revision: https://developer.blender.org/D2002
2016-06-23 22:56:43 +02:00
|
|
|
#endif
|
|
|
|
|
eval += throughput * phase *
|
|
|
|
|
mf_G1(wo_outside ? wo : -wo,
|
|
|
|
|
mf_C1((outside == wo_outside) ? hr : -hr),
|
|
|
|
|
shadowing_lambda);
|
|
|
|
|
}
|
|
|
|
|
if (order + 1 < 10) {
|
|
|
|
|
/* Bounce from the microfacet. */
|
|
|
|
|
#ifdef MF_MULTI_GLASS
|
|
|
|
|
bool next_outside;
|
2017-04-18 11:43:09 +02:00
|
|
|
float3 wi_prev = -wr;
|
Cycles: Kernel address space changes for MSL
This is the first of a sequence of changes to support compiling Cycles kernels as MSL (Metal Shading Language) in preparation for a Metal GPU device implementation.
MSL requires that all pointer types be declared with explicit address space attributes (device, thread, etc...). There is already precedent for this with Cycles' address space macros (ccl_global, ccl_private, etc...), therefore the first step of MSL-enablement is to apply these consistently. Line-for-line this represents the largest change required to enable MSL. Applying this change first will simplify future patches as well as offering the emergent benefit of enhanced descriptiveness.
The vast majority of deltas in this patch fall into one of two cases:
- Ensuring ccl_private is specified for thread-local pointer types
- Ensuring ccl_global is specified for device-wide pointer types
Additionally, the ccl_addr_space qualifier can be removed. Prior to Cycles X, ccl_addr_space was used as a context-dependent address space qualifier, but now it is either redundant (e.g. in struct typedefs), or can be replaced by ccl_global in the case of pointer types. Associated function variants (e.g. lcg_step_float_addrspace) are also redundant.
In cases where address space qualifiers are chained with "const", this patch places the address space qualifier first. The rationale for this is that the choice of address space is likely to have the greater impact on runtime performance and overall architecture.
The final part of this patch is the addition of a metal/compat.h header. This is partially complete and will be extended in future patches, paving the way for the full Metal implementation.
Ref T92212
Reviewed By: brecht
Maniphest Tasks: T92212
Differential Revision: https://developer.blender.org/D12864
2021-10-14 13:53:40 +01:00
|
|
|
float phase_rand = lcg_step_float(lcg_state);
|
2017-07-23 22:43:55 +02:00
|
|
|
wr = mf_sample_phase_glass(-wr, outside ? eta : 1.0f / eta, wm, phase_rand, &next_outside);
|
Cycles: Add multi-scattering, energy-conserving GGX as an option to the Glossy, Anisotropic and Glass BSDFs
This commit adds a new distribution to the Glossy, Anisotropic and Glass BSDFs that implements the
multiple-scattering microfacet model described in the paper "Multiple-Scattering Microfacet BSDFs with the Smith Model".
Essentially, the improvement is that unlike classical GGX, which only models single scattering and assumes
the contribution of multiple bounces to be zero, this new model performs a random walk on the microsurface until
the ray leaves it again, which ensures perfect energy conservation.
In practise, this means that the "darkening problem" - GGX materials becoming darker with increasing
roughness - is solved in a physically correct and efficient way.
The downside of this model is that it has no (known) analytic expression for evalation. However, it can be
evaluated stochastically, and although the correct PDF isn't known either, the properties of MIS and the
balance heuristic guarantee an unbiased result at the cost of slightly higher noise.
Reviewers: dingto, #cycles, brecht
Reviewed By: dingto, #cycles, brecht
Subscribers: bliblubli, ace_dragon, gregzaal, brecht, harvester, dingto, marcog, swerner, jtheninja, Blendify, nutel
Differential Revision: https://developer.blender.org/D2002
2016-06-23 22:56:43 +02:00
|
|
|
if (!next_outside) {
|
|
|
|
|
outside = !outside;
|
|
|
|
|
wr = -wr;
|
|
|
|
|
hr = -hr;
|
|
|
|
|
}
|
2017-04-18 11:43:09 +02:00
|
|
|
|
|
|
|
|
if (use_fresnel && !next_outside) {
|
|
|
|
|
throughput *= color;
|
|
|
|
|
}
|
|
|
|
|
else if (use_fresnel && order > 0) {
|
|
|
|
|
throughput *= interpolate_fresnel_color(wi_prev, wm, eta, F0, cspec0);
|
|
|
|
|
}
|
Cycles: Add multi-scattering, energy-conserving GGX as an option to the Glossy, Anisotropic and Glass BSDFs
This commit adds a new distribution to the Glossy, Anisotropic and Glass BSDFs that implements the
multiple-scattering microfacet model described in the paper "Multiple-Scattering Microfacet BSDFs with the Smith Model".
Essentially, the improvement is that unlike classical GGX, which only models single scattering and assumes
the contribution of multiple bounces to be zero, this new model performs a random walk on the microsurface until
the ray leaves it again, which ensures perfect energy conservation.
In practise, this means that the "darkening problem" - GGX materials becoming darker with increasing
roughness - is solved in a physically correct and efficient way.
The downside of this model is that it has no (known) analytic expression for evalation. However, it can be
evaluated stochastically, and although the correct PDF isn't known either, the properties of MIS and the
balance heuristic guarantee an unbiased result at the cost of slightly higher noise.
Reviewers: dingto, #cycles, brecht
Reviewed By: dingto, #cycles, brecht
Subscribers: bliblubli, ace_dragon, gregzaal, brecht, harvester, dingto, marcog, swerner, jtheninja, Blendify, nutel
Differential Revision: https://developer.blender.org/D2002
2016-06-23 22:56:43 +02:00
|
|
|
#else /* MF_MULTI_GLOSSY */
|
2017-04-18 11:43:09 +02:00
|
|
|
if (use_fresnel && order > 0) {
|
|
|
|
|
throughput *= interpolate_fresnel_color(-wr, wm, eta, F0, cspec0);
|
|
|
|
|
}
|
2017-05-14 21:17:32 +02:00
|
|
|
wr = mf_sample_phase_glossy(-wr, &throughput, wm);
|
Cycles: Add multi-scattering, energy-conserving GGX as an option to the Glossy, Anisotropic and Glass BSDFs
This commit adds a new distribution to the Glossy, Anisotropic and Glass BSDFs that implements the
multiple-scattering microfacet model described in the paper "Multiple-Scattering Microfacet BSDFs with the Smith Model".
Essentially, the improvement is that unlike classical GGX, which only models single scattering and assumes
the contribution of multiple bounces to be zero, this new model performs a random walk on the microsurface until
the ray leaves it again, which ensures perfect energy conservation.
In practise, this means that the "darkening problem" - GGX materials becoming darker with increasing
roughness - is solved in a physically correct and efficient way.
The downside of this model is that it has no (known) analytic expression for evalation. However, it can be
evaluated stochastically, and although the correct PDF isn't known either, the properties of MIS and the
balance heuristic guarantee an unbiased result at the cost of slightly higher noise.
Reviewers: dingto, #cycles, brecht
Reviewed By: dingto, #cycles, brecht
Subscribers: bliblubli, ace_dragon, gregzaal, brecht, harvester, dingto, marcog, swerner, jtheninja, Blendify, nutel
Differential Revision: https://developer.blender.org/D2002
2016-06-23 22:56:43 +02:00
|
|
|
#endif
|
|
|
|
|
|
|
|
|
|
lambda_r = mf_lambda(wr, alpha);
|
|
|
|
|
|
2017-04-18 11:43:09 +02:00
|
|
|
if (!use_fresnel)
|
|
|
|
|
throughput *= color;
|
Cycles: Add multi-scattering, energy-conserving GGX as an option to the Glossy, Anisotropic and Glass BSDFs
This commit adds a new distribution to the Glossy, Anisotropic and Glass BSDFs that implements the
multiple-scattering microfacet model described in the paper "Multiple-Scattering Microfacet BSDFs with the Smith Model".
Essentially, the improvement is that unlike classical GGX, which only models single scattering and assumes
the contribution of multiple bounces to be zero, this new model performs a random walk on the microsurface until
the ray leaves it again, which ensures perfect energy conservation.
In practise, this means that the "darkening problem" - GGX materials becoming darker with increasing
roughness - is solved in a physically correct and efficient way.
The downside of this model is that it has no (known) analytic expression for evalation. However, it can be
evaluated stochastically, and although the correct PDF isn't known either, the properties of MIS and the
balance heuristic guarantee an unbiased result at the cost of slightly higher noise.
Reviewers: dingto, #cycles, brecht
Reviewed By: dingto, #cycles, brecht
Subscribers: bliblubli, ace_dragon, gregzaal, brecht, harvester, dingto, marcog, swerner, jtheninja, Blendify, nutel
Differential Revision: https://developer.blender.org/D2002
2016-06-23 22:56:43 +02:00
|
|
|
|
|
|
|
|
C1_r = mf_C1(hr);
|
|
|
|
|
G1_r = mf_G1(wr, C1_r, lambda_r);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (swapped)
|
|
|
|
|
eval *= fabsf(wi.z / wo.z);
|
|
|
|
|
return eval;
|
|
|
|
|
}
|
|
|
|
|
|
2019-05-01 21:14:11 +10:00
|
|
|
/* Perform a random walk on the microsurface starting from wi, returning the direction in which the
|
|
|
|
|
* walk escaped the surface in wo. The function returns the throughput between wi and wo. Without
|
|
|
|
|
* reflection losses due to coloring or fresnel absorption in conductors, the sampling is optimal.
|
Cycles: Add multi-scattering, energy-conserving GGX as an option to the Glossy, Anisotropic and Glass BSDFs
This commit adds a new distribution to the Glossy, Anisotropic and Glass BSDFs that implements the
multiple-scattering microfacet model described in the paper "Multiple-Scattering Microfacet BSDFs with the Smith Model".
Essentially, the improvement is that unlike classical GGX, which only models single scattering and assumes
the contribution of multiple bounces to be zero, this new model performs a random walk on the microsurface until
the ray leaves it again, which ensures perfect energy conservation.
In practise, this means that the "darkening problem" - GGX materials becoming darker with increasing
roughness - is solved in a physically correct and efficient way.
The downside of this model is that it has no (known) analytic expression for evalation. However, it can be
evaluated stochastically, and although the correct PDF isn't known either, the properties of MIS and the
balance heuristic guarantee an unbiased result at the cost of slightly higher noise.
Reviewers: dingto, #cycles, brecht
Reviewed By: dingto, #cycles, brecht
Subscribers: bliblubli, ace_dragon, gregzaal, brecht, harvester, dingto, marcog, swerner, jtheninja, Blendify, nutel
Differential Revision: https://developer.blender.org/D2002
2016-06-23 22:56:43 +02:00
|
|
|
*/
|
2022-07-29 13:41:37 +02:00
|
|
|
ccl_device_forceinline Spectrum MF_FUNCTION_FULL_NAME(mf_sample)(float3 wi,
|
|
|
|
|
ccl_private float3 *wo,
|
|
|
|
|
const Spectrum color,
|
|
|
|
|
const float alpha_x,
|
|
|
|
|
const float alpha_y,
|
|
|
|
|
ccl_private uint *lcg_state,
|
|
|
|
|
const float eta,
|
|
|
|
|
bool use_fresnel,
|
|
|
|
|
const Spectrum cspec0)
|
Cycles: Add multi-scattering, energy-conserving GGX as an option to the Glossy, Anisotropic and Glass BSDFs
This commit adds a new distribution to the Glossy, Anisotropic and Glass BSDFs that implements the
multiple-scattering microfacet model described in the paper "Multiple-Scattering Microfacet BSDFs with the Smith Model".
Essentially, the improvement is that unlike classical GGX, which only models single scattering and assumes
the contribution of multiple bounces to be zero, this new model performs a random walk on the microsurface until
the ray leaves it again, which ensures perfect energy conservation.
In practise, this means that the "darkening problem" - GGX materials becoming darker with increasing
roughness - is solved in a physically correct and efficient way.
The downside of this model is that it has no (known) analytic expression for evalation. However, it can be
evaluated stochastically, and although the correct PDF isn't known either, the properties of MIS and the
balance heuristic guarantee an unbiased result at the cost of slightly higher noise.
Reviewers: dingto, #cycles, brecht
Reviewed By: dingto, #cycles, brecht
Subscribers: bliblubli, ace_dragon, gregzaal, brecht, harvester, dingto, marcog, swerner, jtheninja, Blendify, nutel
Differential Revision: https://developer.blender.org/D2002
2016-06-23 22:56:43 +02:00
|
|
|
{
|
|
|
|
|
const float2 alpha = make_float2(alpha_x, alpha_y);
|
|
|
|
|
|
2022-07-29 13:41:37 +02:00
|
|
|
Spectrum throughput = one_spectrum();
|
Cycles: Add multi-scattering, energy-conserving GGX as an option to the Glossy, Anisotropic and Glass BSDFs
This commit adds a new distribution to the Glossy, Anisotropic and Glass BSDFs that implements the
multiple-scattering microfacet model described in the paper "Multiple-Scattering Microfacet BSDFs with the Smith Model".
Essentially, the improvement is that unlike classical GGX, which only models single scattering and assumes
the contribution of multiple bounces to be zero, this new model performs a random walk on the microsurface until
the ray leaves it again, which ensures perfect energy conservation.
In practise, this means that the "darkening problem" - GGX materials becoming darker with increasing
roughness - is solved in a physically correct and efficient way.
The downside of this model is that it has no (known) analytic expression for evalation. However, it can be
evaluated stochastically, and although the correct PDF isn't known either, the properties of MIS and the
balance heuristic guarantee an unbiased result at the cost of slightly higher noise.
Reviewers: dingto, #cycles, brecht
Reviewed By: dingto, #cycles, brecht
Subscribers: bliblubli, ace_dragon, gregzaal, brecht, harvester, dingto, marcog, swerner, jtheninja, Blendify, nutel
Differential Revision: https://developer.blender.org/D2002
2016-06-23 22:56:43 +02:00
|
|
|
float3 wr = -wi;
|
|
|
|
|
float lambda_r = mf_lambda(wr, alpha);
|
|
|
|
|
float hr = 1.0f;
|
|
|
|
|
float C1_r = 1.0f;
|
|
|
|
|
float G1_r = 0.0f;
|
|
|
|
|
bool outside = true;
|
2017-05-14 21:17:32 +02:00
|
|
|
|
2017-04-18 11:43:09 +02:00
|
|
|
float F0 = fresnel_dielectric_cos(1.0f, eta);
|
Cycles: Add multi-scattering, energy-conserving GGX as an option to the Glossy, Anisotropic and Glass BSDFs
This commit adds a new distribution to the Glossy, Anisotropic and Glass BSDFs that implements the
multiple-scattering microfacet model described in the paper "Multiple-Scattering Microfacet BSDFs with the Smith Model".
Essentially, the improvement is that unlike classical GGX, which only models single scattering and assumes
the contribution of multiple bounces to be zero, this new model performs a random walk on the microsurface until
the ray leaves it again, which ensures perfect energy conservation.
In practise, this means that the "darkening problem" - GGX materials becoming darker with increasing
roughness - is solved in a physically correct and efficient way.
The downside of this model is that it has no (known) analytic expression for evalation. However, it can be
evaluated stochastically, and although the correct PDF isn't known either, the properties of MIS and the
balance heuristic guarantee an unbiased result at the cost of slightly higher noise.
Reviewers: dingto, #cycles, brecht
Reviewed By: dingto, #cycles, brecht
Subscribers: bliblubli, ace_dragon, gregzaal, brecht, harvester, dingto, marcog, swerner, jtheninja, Blendify, nutel
Differential Revision: https://developer.blender.org/D2002
2016-06-23 22:56:43 +02:00
|
|
|
|
|
|
|
|
int order;
|
|
|
|
|
for (order = 0; order < 10; order++) {
|
|
|
|
|
/* Sample microfacet height. */
|
Cycles: Kernel address space changes for MSL
This is the first of a sequence of changes to support compiling Cycles kernels as MSL (Metal Shading Language) in preparation for a Metal GPU device implementation.
MSL requires that all pointer types be declared with explicit address space attributes (device, thread, etc...). There is already precedent for this with Cycles' address space macros (ccl_global, ccl_private, etc...), therefore the first step of MSL-enablement is to apply these consistently. Line-for-line this represents the largest change required to enable MSL. Applying this change first will simplify future patches as well as offering the emergent benefit of enhanced descriptiveness.
The vast majority of deltas in this patch fall into one of two cases:
- Ensuring ccl_private is specified for thread-local pointer types
- Ensuring ccl_global is specified for device-wide pointer types
Additionally, the ccl_addr_space qualifier can be removed. Prior to Cycles X, ccl_addr_space was used as a context-dependent address space qualifier, but now it is either redundant (e.g. in struct typedefs), or can be replaced by ccl_global in the case of pointer types. Associated function variants (e.g. lcg_step_float_addrspace) are also redundant.
In cases where address space qualifiers are chained with "const", this patch places the address space qualifier first. The rationale for this is that the choice of address space is likely to have the greater impact on runtime performance and overall architecture.
The final part of this patch is the addition of a metal/compat.h header. This is partially complete and will be extended in future patches, paving the way for the full Metal implementation.
Ref T92212
Reviewed By: brecht
Maniphest Tasks: T92212
Differential Revision: https://developer.blender.org/D12864
2021-10-14 13:53:40 +01:00
|
|
|
float height_rand = lcg_step_float(lcg_state);
|
2017-07-23 22:43:55 +02:00
|
|
|
if (!mf_sample_height(wr, &hr, &C1_r, &G1_r, &lambda_r, height_rand)) {
|
Cycles: Add multi-scattering, energy-conserving GGX as an option to the Glossy, Anisotropic and Glass BSDFs
This commit adds a new distribution to the Glossy, Anisotropic and Glass BSDFs that implements the
multiple-scattering microfacet model described in the paper "Multiple-Scattering Microfacet BSDFs with the Smith Model".
Essentially, the improvement is that unlike classical GGX, which only models single scattering and assumes
the contribution of multiple bounces to be zero, this new model performs a random walk on the microsurface until
the ray leaves it again, which ensures perfect energy conservation.
In practise, this means that the "darkening problem" - GGX materials becoming darker with increasing
roughness - is solved in a physically correct and efficient way.
The downside of this model is that it has no (known) analytic expression for evalation. However, it can be
evaluated stochastically, and although the correct PDF isn't known either, the properties of MIS and the
balance heuristic guarantee an unbiased result at the cost of slightly higher noise.
Reviewers: dingto, #cycles, brecht
Reviewed By: dingto, #cycles, brecht
Subscribers: bliblubli, ace_dragon, gregzaal, brecht, harvester, dingto, marcog, swerner, jtheninja, Blendify, nutel
Differential Revision: https://developer.blender.org/D2002
2016-06-23 22:56:43 +02:00
|
|
|
/* The random walk has left the surface. */
|
|
|
|
|
*wo = outside ? wr : -wr;
|
|
|
|
|
return throughput;
|
|
|
|
|
}
|
|
|
|
|
/* Sample microfacet normal. */
|
Cycles: Kernel address space changes for MSL
This is the first of a sequence of changes to support compiling Cycles kernels as MSL (Metal Shading Language) in preparation for a Metal GPU device implementation.
MSL requires that all pointer types be declared with explicit address space attributes (device, thread, etc...). There is already precedent for this with Cycles' address space macros (ccl_global, ccl_private, etc...), therefore the first step of MSL-enablement is to apply these consistently. Line-for-line this represents the largest change required to enable MSL. Applying this change first will simplify future patches as well as offering the emergent benefit of enhanced descriptiveness.
The vast majority of deltas in this patch fall into one of two cases:
- Ensuring ccl_private is specified for thread-local pointer types
- Ensuring ccl_global is specified for device-wide pointer types
Additionally, the ccl_addr_space qualifier can be removed. Prior to Cycles X, ccl_addr_space was used as a context-dependent address space qualifier, but now it is either redundant (e.g. in struct typedefs), or can be replaced by ccl_global in the case of pointer types. Associated function variants (e.g. lcg_step_float_addrspace) are also redundant.
In cases where address space qualifiers are chained with "const", this patch places the address space qualifier first. The rationale for this is that the choice of address space is likely to have the greater impact on runtime performance and overall architecture.
The final part of this patch is the addition of a metal/compat.h header. This is partially complete and will be extended in future patches, paving the way for the full Metal implementation.
Ref T92212
Reviewed By: brecht
Maniphest Tasks: T92212
Differential Revision: https://developer.blender.org/D12864
2021-10-14 13:53:40 +01:00
|
|
|
float vndf_rand_y = lcg_step_float(lcg_state);
|
|
|
|
|
float vndf_rand_x = lcg_step_float(lcg_state);
|
2017-07-23 22:43:55 +02:00
|
|
|
float3 wm = mf_sample_vndf(-wr, alpha, vndf_rand_x, vndf_rand_y);
|
Cycles: Add multi-scattering, energy-conserving GGX as an option to the Glossy, Anisotropic and Glass BSDFs
This commit adds a new distribution to the Glossy, Anisotropic and Glass BSDFs that implements the
multiple-scattering microfacet model described in the paper "Multiple-Scattering Microfacet BSDFs with the Smith Model".
Essentially, the improvement is that unlike classical GGX, which only models single scattering and assumes
the contribution of multiple bounces to be zero, this new model performs a random walk on the microsurface until
the ray leaves it again, which ensures perfect energy conservation.
In practise, this means that the "darkening problem" - GGX materials becoming darker with increasing
roughness - is solved in a physically correct and efficient way.
The downside of this model is that it has no (known) analytic expression for evalation. However, it can be
evaluated stochastically, and although the correct PDF isn't known either, the properties of MIS and the
balance heuristic guarantee an unbiased result at the cost of slightly higher noise.
Reviewers: dingto, #cycles, brecht
Reviewed By: dingto, #cycles, brecht
Subscribers: bliblubli, ace_dragon, gregzaal, brecht, harvester, dingto, marcog, swerner, jtheninja, Blendify, nutel
Differential Revision: https://developer.blender.org/D2002
2016-06-23 22:56:43 +02:00
|
|
|
|
|
|
|
|
/* First-bounce color is already accounted for in mix weight. */
|
2017-04-18 11:43:09 +02:00
|
|
|
if (!use_fresnel && order > 0)
|
|
|
|
|
throughput *= color;
|
Cycles: Add multi-scattering, energy-conserving GGX as an option to the Glossy, Anisotropic and Glass BSDFs
This commit adds a new distribution to the Glossy, Anisotropic and Glass BSDFs that implements the
multiple-scattering microfacet model described in the paper "Multiple-Scattering Microfacet BSDFs with the Smith Model".
Essentially, the improvement is that unlike classical GGX, which only models single scattering and assumes
the contribution of multiple bounces to be zero, this new model performs a random walk on the microsurface until
the ray leaves it again, which ensures perfect energy conservation.
In practise, this means that the "darkening problem" - GGX materials becoming darker with increasing
roughness - is solved in a physically correct and efficient way.
The downside of this model is that it has no (known) analytic expression for evalation. However, it can be
evaluated stochastically, and although the correct PDF isn't known either, the properties of MIS and the
balance heuristic guarantee an unbiased result at the cost of slightly higher noise.
Reviewers: dingto, #cycles, brecht
Reviewed By: dingto, #cycles, brecht
Subscribers: bliblubli, ace_dragon, gregzaal, brecht, harvester, dingto, marcog, swerner, jtheninja, Blendify, nutel
Differential Revision: https://developer.blender.org/D2002
2016-06-23 22:56:43 +02:00
|
|
|
|
|
|
|
|
/* Bounce from the microfacet. */
|
|
|
|
|
#ifdef MF_MULTI_GLASS
|
|
|
|
|
bool next_outside;
|
2017-04-18 11:43:09 +02:00
|
|
|
float3 wi_prev = -wr;
|
Cycles: Kernel address space changes for MSL
This is the first of a sequence of changes to support compiling Cycles kernels as MSL (Metal Shading Language) in preparation for a Metal GPU device implementation.
MSL requires that all pointer types be declared with explicit address space attributes (device, thread, etc...). There is already precedent for this with Cycles' address space macros (ccl_global, ccl_private, etc...), therefore the first step of MSL-enablement is to apply these consistently. Line-for-line this represents the largest change required to enable MSL. Applying this change first will simplify future patches as well as offering the emergent benefit of enhanced descriptiveness.
The vast majority of deltas in this patch fall into one of two cases:
- Ensuring ccl_private is specified for thread-local pointer types
- Ensuring ccl_global is specified for device-wide pointer types
Additionally, the ccl_addr_space qualifier can be removed. Prior to Cycles X, ccl_addr_space was used as a context-dependent address space qualifier, but now it is either redundant (e.g. in struct typedefs), or can be replaced by ccl_global in the case of pointer types. Associated function variants (e.g. lcg_step_float_addrspace) are also redundant.
In cases where address space qualifiers are chained with "const", this patch places the address space qualifier first. The rationale for this is that the choice of address space is likely to have the greater impact on runtime performance and overall architecture.
The final part of this patch is the addition of a metal/compat.h header. This is partially complete and will be extended in future patches, paving the way for the full Metal implementation.
Ref T92212
Reviewed By: brecht
Maniphest Tasks: T92212
Differential Revision: https://developer.blender.org/D12864
2021-10-14 13:53:40 +01:00
|
|
|
float phase_rand = lcg_step_float(lcg_state);
|
2017-07-23 22:43:55 +02:00
|
|
|
wr = mf_sample_phase_glass(-wr, outside ? eta : 1.0f / eta, wm, phase_rand, &next_outside);
|
Cycles: Add multi-scattering, energy-conserving GGX as an option to the Glossy, Anisotropic and Glass BSDFs
This commit adds a new distribution to the Glossy, Anisotropic and Glass BSDFs that implements the
multiple-scattering microfacet model described in the paper "Multiple-Scattering Microfacet BSDFs with the Smith Model".
Essentially, the improvement is that unlike classical GGX, which only models single scattering and assumes
the contribution of multiple bounces to be zero, this new model performs a random walk on the microsurface until
the ray leaves it again, which ensures perfect energy conservation.
In practise, this means that the "darkening problem" - GGX materials becoming darker with increasing
roughness - is solved in a physically correct and efficient way.
The downside of this model is that it has no (known) analytic expression for evalation. However, it can be
evaluated stochastically, and although the correct PDF isn't known either, the properties of MIS and the
balance heuristic guarantee an unbiased result at the cost of slightly higher noise.
Reviewers: dingto, #cycles, brecht
Reviewed By: dingto, #cycles, brecht
Subscribers: bliblubli, ace_dragon, gregzaal, brecht, harvester, dingto, marcog, swerner, jtheninja, Blendify, nutel
Differential Revision: https://developer.blender.org/D2002
2016-06-23 22:56:43 +02:00
|
|
|
if (!next_outside) {
|
|
|
|
|
hr = -hr;
|
|
|
|
|
wr = -wr;
|
|
|
|
|
outside = !outside;
|
|
|
|
|
}
|
2017-04-18 11:43:09 +02:00
|
|
|
|
|
|
|
|
if (use_fresnel) {
|
|
|
|
|
if (!next_outside) {
|
|
|
|
|
throughput *= color;
|
|
|
|
|
}
|
|
|
|
|
else {
|
2022-07-29 13:41:37 +02:00
|
|
|
Spectrum t_color = interpolate_fresnel_color(wi_prev, wm, eta, F0, cspec0);
|
2017-04-18 11:43:09 +02:00
|
|
|
|
|
|
|
|
if (order == 0)
|
|
|
|
|
throughput = t_color;
|
|
|
|
|
else
|
|
|
|
|
throughput *= t_color;
|
|
|
|
|
}
|
|
|
|
|
}
|
Cycles: Add multi-scattering, energy-conserving GGX as an option to the Glossy, Anisotropic and Glass BSDFs
This commit adds a new distribution to the Glossy, Anisotropic and Glass BSDFs that implements the
multiple-scattering microfacet model described in the paper "Multiple-Scattering Microfacet BSDFs with the Smith Model".
Essentially, the improvement is that unlike classical GGX, which only models single scattering and assumes
the contribution of multiple bounces to be zero, this new model performs a random walk on the microsurface until
the ray leaves it again, which ensures perfect energy conservation.
In practise, this means that the "darkening problem" - GGX materials becoming darker with increasing
roughness - is solved in a physically correct and efficient way.
The downside of this model is that it has no (known) analytic expression for evalation. However, it can be
evaluated stochastically, and although the correct PDF isn't known either, the properties of MIS and the
balance heuristic guarantee an unbiased result at the cost of slightly higher noise.
Reviewers: dingto, #cycles, brecht
Reviewed By: dingto, #cycles, brecht
Subscribers: bliblubli, ace_dragon, gregzaal, brecht, harvester, dingto, marcog, swerner, jtheninja, Blendify, nutel
Differential Revision: https://developer.blender.org/D2002
2016-06-23 22:56:43 +02:00
|
|
|
#else /* MF_MULTI_GLOSSY */
|
2017-04-18 11:43:09 +02:00
|
|
|
if (use_fresnel) {
|
2022-07-29 13:41:37 +02:00
|
|
|
Spectrum t_color = interpolate_fresnel_color(-wr, wm, eta, F0, cspec0);
|
2017-04-18 11:43:09 +02:00
|
|
|
|
|
|
|
|
if (order == 0)
|
|
|
|
|
throughput = t_color;
|
|
|
|
|
else
|
|
|
|
|
throughput *= t_color;
|
|
|
|
|
}
|
2017-05-14 21:17:32 +02:00
|
|
|
wr = mf_sample_phase_glossy(-wr, &throughput, wm);
|
Cycles: Add multi-scattering, energy-conserving GGX as an option to the Glossy, Anisotropic and Glass BSDFs
This commit adds a new distribution to the Glossy, Anisotropic and Glass BSDFs that implements the
multiple-scattering microfacet model described in the paper "Multiple-Scattering Microfacet BSDFs with the Smith Model".
Essentially, the improvement is that unlike classical GGX, which only models single scattering and assumes
the contribution of multiple bounces to be zero, this new model performs a random walk on the microsurface until
the ray leaves it again, which ensures perfect energy conservation.
In practise, this means that the "darkening problem" - GGX materials becoming darker with increasing
roughness - is solved in a physically correct and efficient way.
The downside of this model is that it has no (known) analytic expression for evalation. However, it can be
evaluated stochastically, and although the correct PDF isn't known either, the properties of MIS and the
balance heuristic guarantee an unbiased result at the cost of slightly higher noise.
Reviewers: dingto, #cycles, brecht
Reviewed By: dingto, #cycles, brecht
Subscribers: bliblubli, ace_dragon, gregzaal, brecht, harvester, dingto, marcog, swerner, jtheninja, Blendify, nutel
Differential Revision: https://developer.blender.org/D2002
2016-06-23 22:56:43 +02:00
|
|
|
#endif
|
|
|
|
|
|
|
|
|
|
/* Update random walk parameters. */
|
|
|
|
|
lambda_r = mf_lambda(wr, alpha);
|
|
|
|
|
G1_r = mf_G1(wr, C1_r, lambda_r);
|
|
|
|
|
}
|
|
|
|
|
*wo = make_float3(0.0f, 0.0f, 1.0f);
|
2022-07-29 13:41:37 +02:00
|
|
|
return zero_spectrum();
|
Cycles: Add multi-scattering, energy-conserving GGX as an option to the Glossy, Anisotropic and Glass BSDFs
This commit adds a new distribution to the Glossy, Anisotropic and Glass BSDFs that implements the
multiple-scattering microfacet model described in the paper "Multiple-Scattering Microfacet BSDFs with the Smith Model".
Essentially, the improvement is that unlike classical GGX, which only models single scattering and assumes
the contribution of multiple bounces to be zero, this new model performs a random walk on the microsurface until
the ray leaves it again, which ensures perfect energy conservation.
In practise, this means that the "darkening problem" - GGX materials becoming darker with increasing
roughness - is solved in a physically correct and efficient way.
The downside of this model is that it has no (known) analytic expression for evalation. However, it can be
evaluated stochastically, and although the correct PDF isn't known either, the properties of MIS and the
balance heuristic guarantee an unbiased result at the cost of slightly higher noise.
Reviewers: dingto, #cycles, brecht
Reviewed By: dingto, #cycles, brecht
Subscribers: bliblubli, ace_dragon, gregzaal, brecht, harvester, dingto, marcog, swerner, jtheninja, Blendify, nutel
Differential Revision: https://developer.blender.org/D2002
2016-06-23 22:56:43 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
#undef MF_MULTI_GLASS
|
|
|
|
|
#undef MF_MULTI_GLOSSY
|
|
|
|
|
#undef MF_PHASE_FUNCTION
|