1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
|
/*
* Copyright 2011-2013 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.
*/
CCL_NAMESPACE_BEGIN
#ifdef __BRANCHED_PATH__
ccl_device_inline void kernel_branched_path_ao(KernelGlobals *kg,
ShaderData *sd,
ShaderData *emission_sd,
PathRadiance *L,
ccl_addr_space PathState *state,
float3 throughput)
{
int num_samples = kernel_data.integrator.ao_samples;
float num_samples_inv = 1.0f/num_samples;
float ao_factor = kernel_data.background.ao_factor;
float3 ao_N;
float3 ao_bsdf = shader_bsdf_ao(kg, sd, ao_factor, &ao_N);
float3 ao_alpha = shader_bsdf_alpha(kg, sd);
for(int j = 0; j < num_samples; j++) {
float bsdf_u, bsdf_v;
path_branched_rng_2D(kg, state->rng_hash, state, j, num_samples, PRNG_BSDF_U, &bsdf_u, &bsdf_v);
float3 ao_D;
float ao_pdf;
sample_cos_hemisphere(ao_N, bsdf_u, bsdf_v, &ao_D, &ao_pdf);
if(dot(sd->Ng, ao_D) > 0.0f && ao_pdf != 0.0f) {
Ray light_ray;
float3 ao_shadow;
light_ray.P = ray_offset(sd->P, sd->Ng);
light_ray.D = ao_D;
light_ray.t = kernel_data.background.ao_distance;
#ifdef __OBJECT_MOTION__
light_ray.time = sd->time;
#endif /* __OBJECT_MOTION__ */
light_ray.dP = sd->dP;
light_ray.dD = differential3_zero();
if(!shadow_blocked(kg, emission_sd, state, &light_ray, &ao_shadow)) {
path_radiance_accum_ao(L, state, throughput*num_samples_inv, ao_alpha, ao_bsdf, ao_shadow);
}
else {
path_radiance_accum_total_ao(L, state, throughput*num_samples_inv, ao_bsdf);
}
}
}
}
#ifndef __SPLIT_KERNEL__
/* bounce off surface and integrate indirect light */
ccl_device_noinline void kernel_branched_path_surface_indirect_light(KernelGlobals *kg,
ShaderData *sd, ShaderData *indirect_sd, ShaderData *emission_sd,
float3 throughput, float num_samples_adjust, PathState *state, PathRadiance *L)
{
float sum_sample_weight = 0.0f;
#ifdef __DENOISING_FEATURES__
if(state->denoising_feature_weight > 0.0f) {
for(int i = 0; i < sd->num_closure; i++) {
const ShaderClosure *sc = &sd->closure[i];
/* transparency is not handled here, but in outer loop */
if(!CLOSURE_IS_BSDF(sc->type) || CLOSURE_IS_BSDF_TRANSPARENT(sc->type)) {
continue;
}
sum_sample_weight += sc->sample_weight;
}
}
else {
sum_sample_weight = 1.0f;
}
#endif /* __DENOISING_FEATURES__ */
for(int i = 0; i < sd->num_closure; i++) {
const ShaderClosure *sc = &sd->closure[i];
/* transparency is not handled here, but in outer loop */
if(!CLOSURE_IS_BSDF(sc->type) || CLOSURE_IS_BSDF_TRANSPARENT(sc->type)) {
continue;
}
int num_samples;
if(CLOSURE_IS_BSDF_DIFFUSE(sc->type))
num_samples = kernel_data.integrator.diffuse_samples;
else if(CLOSURE_IS_BSDF_BSSRDF(sc->type))
num_samples = 1;
else if(CLOSURE_IS_BSDF_GLOSSY(sc->type))
num_samples = kernel_data.integrator.glossy_samples;
else
num_samples = kernel_data.integrator.transmission_samples;
num_samples = ceil_to_int(num_samples_adjust*num_samples);
float num_samples_inv = num_samples_adjust/num_samples;
for(int j = 0; j < num_samples; j++) {
PathState ps = *state;
float3 tp = throughput;
Ray bsdf_ray;
#ifdef __SHADOW_TRICKS__
float shadow_transparency = L->shadow_transparency;
#endif
ps.rng_hash = cmj_hash(state->rng_hash, i);
if(!kernel_branched_path_surface_bounce(kg,
sd,
sc,
j,
num_samples,
&tp,
&ps,
L,
&bsdf_ray,
sum_sample_weight))
{
continue;
}
ps.rng_hash = state->rng_hash;
kernel_path_indirect(kg,
indirect_sd,
emission_sd,
&bsdf_ray,
tp*num_samples_inv,
num_samples,
&ps,
L);
/* for render passes, sum and reset indirect light pass variables
* for the next samples */
path_radiance_sum_indirect(L);
path_radiance_reset_indirect(L);
#ifdef __SHADOW_TRICKS__
L->shadow_transparency = shadow_transparency;
#endif
}
}
}
#ifdef __SUBSURFACE__
ccl_device void kernel_branched_path_subsurface_scatter(KernelGlobals *kg,
ShaderData *sd,
ShaderData *indirect_sd,
ShaderData *emission_sd,
PathRadiance *L,
PathState *state,
Ray *ray,
float3 throughput)
{
for(int i = 0; i < sd->num_closure; i++) {
ShaderClosure *sc = &sd->closure[i];
if(!CLOSURE_IS_BSSRDF(sc->type))
continue;
/* set up random number generator */
uint lcg_state = lcg_state_init(state, 0x68bc21eb);
int num_samples = kernel_data.integrator.subsurface_samples;
float num_samples_inv = 1.0f/num_samples;
uint bssrdf_rng_hash = cmj_hash(state->rng_hash, i);
/* do subsurface scatter step with copy of shader data, this will
* replace the BSSRDF with a diffuse BSDF closure */
for(int j = 0; j < num_samples; j++) {
SubsurfaceIntersection ss_isect;
float bssrdf_u, bssrdf_v;
path_branched_rng_2D(kg, bssrdf_rng_hash, state, j, num_samples, PRNG_BSDF_U, &bssrdf_u, &bssrdf_v);
int num_hits = subsurface_scatter_multi_intersect(kg,
&ss_isect,
sd,
sc,
&lcg_state,
bssrdf_u, bssrdf_v,
true);
#ifdef __VOLUME__
Ray volume_ray = *ray;
bool need_update_volume_stack =
kernel_data.integrator.use_volumes &&
sd->object_flag & SD_OBJECT_INTERSECTS_VOLUME;
#endif /* __VOLUME__ */
/* compute lighting with the BSDF closure */
for(int hit = 0; hit < num_hits; hit++) {
ShaderData bssrdf_sd = *sd;
subsurface_scatter_multi_setup(kg,
&ss_isect,
hit,
&bssrdf_sd,
state,
state->flag,
sc,
true);
PathState hit_state = *state;
path_state_branch(&hit_state, j, num_samples);
#ifdef __VOLUME__
if(need_update_volume_stack) {
/* Setup ray from previous surface point to the new one. */
float3 P = ray_offset(bssrdf_sd.P, -bssrdf_sd.Ng);
volume_ray.D = normalize_len(P - volume_ray.P,
&volume_ray.t);
kernel_volume_stack_update_for_subsurface(
kg,
emission_sd,
&volume_ray,
hit_state.volume_stack);
}
#endif /* __VOLUME__ */
#ifdef __EMISSION__
/* direct light */
if(kernel_data.integrator.use_direct_light) {
int all = (kernel_data.integrator.sample_all_lights_direct) ||
(state->flag & PATH_RAY_SHADOW_CATCHER);
kernel_branched_path_surface_connect_light(
kg,
&bssrdf_sd,
emission_sd,
&hit_state,
throughput,
num_samples_inv,
L,
all);
}
#endif /* __EMISSION__ */
/* indirect light */
kernel_branched_path_surface_indirect_light(
kg,
&bssrdf_sd,
indirect_sd,
emission_sd,
throughput,
num_samples_inv,
&hit_state,
L);
}
}
}
}
#endif /* __SUBSURFACE__ */
ccl_device void kernel_branched_path_integrate(KernelGlobals *kg,
uint rng_hash,
int sample,
Ray ray,
ccl_global float *buffer,
PathRadiance *L,
bool *is_shadow_catcher)
{
/* initialize */
float3 throughput = make_float3(1.0f, 1.0f, 1.0f);
path_radiance_init(L, kernel_data.film.use_light_pass);
/* shader data memory used for both volumes and surfaces, saves stack space */
ShaderData sd;
/* shader data used by emission, shadows, volume stacks, indirect path */
ShaderData emission_sd, indirect_sd;
PathState state;
path_state_init(kg, &emission_sd, &state, rng_hash, sample, &ray);
/* Main Loop
* Here we only handle transparency intersections from the camera ray.
* Indirect bounces are handled in kernel_branched_path_surface_indirect_light().
*/
for(;;) {
/* intersect scene */
Intersection isect;
uint visibility = path_state_ray_visibility(kg, &state);
#ifdef __HAIR__
float difl = 0.0f, extmax = 0.0f;
uint lcg_state = 0;
if(kernel_data.bvh.have_curves) {
if(kernel_data.cam.resolution == 1) {
float3 pixdiff = ray.dD.dx + ray.dD.dy;
/*pixdiff = pixdiff - dot(pixdiff, ray.D)*ray.D;*/
difl = kernel_data.curve.minimum_width * len(pixdiff) * 0.5f;
}
extmax = kernel_data.curve.maximum_width;
lcg_state = lcg_state_init(&state, 0x51633e2d);
}
bool hit = scene_intersect(kg, ray, visibility, &isect, &lcg_state, difl, extmax);
#else
bool hit = scene_intersect(kg, ray, visibility, &isect, NULL, 0.0f, 0.0f);
#endif /* __HAIR__ */
#ifdef __KERNEL_DEBUG__
L->debug_data.num_bvh_traversed_nodes += isect.num_traversed_nodes;
L->debug_data.num_bvh_traversed_instances += isect.num_traversed_instances;
L->debug_data.num_bvh_intersections += isect.num_intersections;
L->debug_data.num_ray_bounces++;
#endif /* __KERNEL_DEBUG__ */
#ifdef __VOLUME__
/* Sanitize volume stack. */
if(!hit) {
kernel_volume_clean_stack(kg, state.volume_stack);
}
/* volume attenuation, emission, scatter */
if(state.volume_stack[0].shader != SHADER_NONE) {
Ray volume_ray = ray;
volume_ray.t = (hit)? isect.t: FLT_MAX;
bool heterogeneous = volume_stack_is_heterogeneous(kg, state.volume_stack);
#ifdef __VOLUME_DECOUPLED__
/* decoupled ray marching only supported on CPU */
/* cache steps along volume for repeated sampling */
VolumeSegment volume_segment;
shader_setup_from_volume(kg, &sd, &volume_ray);
kernel_volume_decoupled_record(kg, &state,
&volume_ray, &sd, &volume_segment, heterogeneous);
/* direct light sampling */
if(volume_segment.closure_flag & SD_SCATTER) {
volume_segment.sampling_method = volume_stack_sampling_method(kg, state.volume_stack);
int all = kernel_data.integrator.sample_all_lights_direct;
kernel_branched_path_volume_connect_light(kg, &sd,
&emission_sd, throughput, &state, L, all,
&volume_ray, &volume_segment);
/* indirect light sampling */
int num_samples = kernel_data.integrator.volume_samples;
float num_samples_inv = 1.0f/num_samples;
for(int j = 0; j < num_samples; j++) {
PathState ps = state;
Ray pray = ray;
float3 tp = throughput;
/* branch RNG state */
path_state_branch(&ps, j, num_samples);
/* scatter sample. if we use distance sampling and take just one
* sample for direct and indirect light, we could share this
* computation, but makes code a bit complex */
float rphase = path_state_rng_1D_for_decision(kg, &ps, PRNG_PHASE);
float rscatter = path_state_rng_1D_for_decision(kg, &ps, PRNG_SCATTER_DISTANCE);
VolumeIntegrateResult result = kernel_volume_decoupled_scatter(kg,
&ps, &pray, &sd, &tp, rphase, rscatter, &volume_segment, NULL, false);
(void)result;
kernel_assert(result == VOLUME_PATH_SCATTERED);
if(kernel_path_volume_bounce(kg,
&sd,
&tp,
&ps,
L,
&pray))
{
kernel_path_indirect(kg,
&indirect_sd,
&emission_sd,
&pray,
tp*num_samples_inv,
num_samples,
&ps,
L);
/* for render passes, sum and reset indirect light pass variables
* for the next samples */
path_radiance_sum_indirect(L);
path_radiance_reset_indirect(L);
}
}
}
/* emission and transmittance */
if(volume_segment.closure_flag & SD_EMISSION)
path_radiance_accum_emission(L, throughput, volume_segment.accum_emission, state.bounce);
throughput *= volume_segment.accum_transmittance;
/* free cached steps */
kernel_volume_decoupled_free(kg, &volume_segment);
#else
/* GPU: no decoupled ray marching, scatter probalistically */
int num_samples = kernel_data.integrator.volume_samples;
float num_samples_inv = 1.0f/num_samples;
/* todo: we should cache the shader evaluations from stepping
* through the volume, for now we redo them multiple times */
for(int j = 0; j < num_samples; j++) {
PathState ps = state;
Ray pray = ray;
float3 tp = throughput * num_samples_inv;
/* branch RNG state */
path_state_branch(&ps, j, num_samples);
VolumeIntegrateResult result = kernel_volume_integrate(
kg, &ps, &sd, &volume_ray, L, &tp, heterogeneous);
#ifdef __VOLUME_SCATTER__
if(result == VOLUME_PATH_SCATTERED) {
/* todo: support equiangular, MIS and all light sampling.
* alternatively get decoupled ray marching working on the GPU */
kernel_path_volume_connect_light(kg, &sd, &emission_sd, tp, &state, L);
if(kernel_path_volume_bounce(kg,
&sd,
&tp,
&ps,
L,
&pray))
{
kernel_path_indirect(kg,
&indirect_sd,
&emission_sd,
&pray,
tp,
num_samples,
&ps,
L);
/* for render passes, sum and reset indirect light pass variables
* for the next samples */
path_radiance_sum_indirect(L);
path_radiance_reset_indirect(L);
}
}
#endif /* __VOLUME_SCATTER__ */
}
/* todo: avoid this calculation using decoupled ray marching */
kernel_volume_shadow(kg, &emission_sd, &state, &volume_ray, &throughput);
#endif /* __VOLUME_DECOUPLED__ */
}
#endif /* __VOLUME__ */
if(!hit) {
/* eval background shader if nothing hit */
if(kernel_data.background.transparent) {
L->transparent += average(throughput);
#ifdef __PASSES__
if(!(kernel_data.film.pass_flag & PASS_BACKGROUND))
#endif /* __PASSES__ */
break;
}
#ifdef __BACKGROUND__
/* sample background shader */
float3 L_background = indirect_background(kg, &emission_sd, &state, &ray);
path_radiance_accum_background(L, &state, throughput, L_background);
#endif /* __BACKGROUND__ */
break;
}
/* setup shading */
shader_setup_from_ray(kg, &sd, &isect, &ray);
shader_eval_surface(kg, &sd, &state, 0.0f, state.flag);
shader_merge_closures(&sd);
#ifdef __SHADOW_TRICKS__
if((sd.object_flag & SD_OBJECT_SHADOW_CATCHER)) {
state.flag |= (PATH_RAY_SHADOW_CATCHER |
PATH_RAY_STORE_SHADOW_INFO);
if(!kernel_data.background.transparent) {
L->shadow_background_color =
indirect_background(kg, &emission_sd, &state, &ray);
}
L->shadow_radiance_sum = path_radiance_clamp_and_sum(kg, L);
L->shadow_throughput = average(throughput);
}
else if(state.flag & PATH_RAY_SHADOW_CATCHER) {
/* Only update transparency after shadow catcher bounce. */
L->shadow_transparency *=
average(shader_bsdf_transparency(kg, &sd));
}
#endif /* __SHADOW_TRICKS__ */
/* holdout */
#ifdef __HOLDOUT__
if((sd.flag & SD_HOLDOUT) || (sd.object_flag & SD_OBJECT_HOLDOUT_MASK)) {
if(kernel_data.background.transparent) {
float3 holdout_weight;
if(sd.object_flag & SD_OBJECT_HOLDOUT_MASK) {
holdout_weight = make_float3(1.0f, 1.0f, 1.0f);
}
else {
holdout_weight = shader_holdout_eval(kg, &sd);
}
/* any throughput is ok, should all be identical here */
L->transparent += average(holdout_weight*throughput);
}
if(sd.object_flag & SD_OBJECT_HOLDOUT_MASK) {
break;
}
}
#endif /* __HOLDOUT__ */
/* holdout mask objects do not write data passes */
kernel_write_data_passes(kg, buffer, L, &sd, sample, &state, throughput);
#ifdef __EMISSION__
/* emission */
if(sd.flag & SD_EMISSION) {
float3 emission = indirect_primitive_emission(kg, &sd, isect.t, state.flag, state.ray_pdf);
path_radiance_accum_emission(L, throughput, emission, state.bounce);
}
#endif /* __EMISSION__ */
/* transparency termination */
if(state.flag & PATH_RAY_TRANSPARENT) {
/* path termination. this is a strange place to put the termination, it's
* mainly due to the mixed in MIS that we use. gives too many unneeded
* shader evaluations, only need emission if we are going to terminate */
float probability = path_state_continuation_probability(kg, &state, throughput);
if(probability == 0.0f) {
break;
}
else if(probability != 1.0f) {
float terminate = path_state_rng_1D_for_decision(kg, &state, PRNG_TERMINATE);
if(terminate >= probability)
break;
throughput /= probability;
}
}
kernel_update_denoising_features(kg, &sd, &state, L);
#ifdef __AO__
/* ambient occlusion */
if(kernel_data.integrator.use_ambient_occlusion || (sd.flag & SD_AO)) {
kernel_branched_path_ao(kg, &sd, &emission_sd, L, &state, throughput);
}
#endif /* __AO__ */
#ifdef __SUBSURFACE__
/* bssrdf scatter to a different location on the same object */
if(sd.flag & SD_BSSRDF) {
kernel_branched_path_subsurface_scatter(kg, &sd, &indirect_sd, &emission_sd,
L, &state, &ray, throughput);
}
#endif /* __SUBSURFACE__ */
if(!(sd.flag & SD_HAS_ONLY_VOLUME)) {
PathState hit_state = state;
#ifdef __EMISSION__
/* direct light */
if(kernel_data.integrator.use_direct_light) {
int all = (kernel_data.integrator.sample_all_lights_direct) ||
(state.flag & PATH_RAY_SHADOW_CATCHER);
kernel_branched_path_surface_connect_light(kg,
&sd, &emission_sd, &hit_state, throughput, 1.0f, L, all);
}
#endif /* __EMISSION__ */
/* indirect light */
kernel_branched_path_surface_indirect_light(kg,
&sd, &indirect_sd, &emission_sd, throughput, 1.0f, &hit_state, L);
/* continue in case of transparency */
throughput *= shader_bsdf_transparency(kg, &sd);
if(is_zero(throughput))
break;
}
/* Update Path State */
state.flag |= PATH_RAY_TRANSPARENT;
state.transparent_bounce++;
ray.P = ray_offset(sd.P, -sd.Ng);
ray.t -= sd.ray_length; /* clipping works through transparent */
#ifdef __RAY_DIFFERENTIALS__
ray.dP = sd.dP;
ray.dD.dx = -sd.dI.dx;
ray.dD.dy = -sd.dI.dy;
#endif /* __RAY_DIFFERENTIALS__ */
#ifdef __VOLUME__
/* enter/exit volume */
kernel_volume_stack_enter_exit(kg, &sd, state.volume_stack);
#endif /* __VOLUME__ */
}
#ifdef __SHADOW_TRICKS__
*is_shadow_catcher = (state.flag & PATH_RAY_SHADOW_CATCHER) != 0;
#endif /* __SHADOW_TRICKS__ */
}
ccl_device void kernel_branched_path_trace(KernelGlobals *kg,
ccl_global float *buffer, ccl_global uint *rng_state,
int sample, int x, int y, int offset, int stride)
{
/* buffer offset */
int index = offset + x + y*stride;
int pass_stride = kernel_data.film.pass_stride;
rng_state += index;
buffer += index*pass_stride;
/* initialize random numbers and ray */
uint rng_hash;
Ray ray;
kernel_path_trace_setup(kg, rng_state, sample, x, y, &rng_hash, &ray);
/* integrate */
PathRadiance L;
bool is_shadow_catcher;
if(ray.t != 0.0f) {
kernel_branched_path_integrate(kg, rng_hash, sample, ray, buffer, &L, &is_shadow_catcher);
kernel_write_result(kg, buffer, sample, &L, is_shadow_catcher);
}
else {
kernel_write_result(kg, buffer, sample, NULL, false);
}
}
#endif /* __SPLIT_KERNEL__ */
#endif /* __BRANCHED_PATH__ */
CCL_NAMESPACE_END
|