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
|
/*
* Copyright 2011-2017 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
ccl_device_inline void kernel_filter_construct_gramian(int x, int y,
int storage_stride,
int dx, int dy, int w, int h,
float ccl_readonly_ptr buffer,
int color_pass, int variance_pass,
float ccl_readonly_ptr transform,
int *rank, float weight,
float *XtWX, float3 *XtWY)
{
const int pass_stride = w*h;
float ccl_readonly_ptr p_buffer = buffer + y*w + x;
float ccl_readonly_ptr q_buffer = buffer + (y+dy)*w + (x+dx);
#ifdef __KERNEL_CPU__
const int stride = 1;
(void)storage_stride;
#else
const int stride = storage_stride;
#endif
float3 p_color = filter_get_pixel_color(p_buffer, color_pass, pass_stride);
float3 q_color = filter_get_pixel_color(q_buffer, color_pass, pass_stride);
float p_std_dev = sqrtf(filter_get_pixel_variance(p_buffer, variance_pass, pass_stride));
float q_std_dev = sqrtf(filter_get_pixel_variance(q_buffer, variance_pass, pass_stride));
if(average(fabs(p_color - q_color)) > 3.0f*(p_std_dev + q_std_dev + 1e-3f)) {
return;
}
float feature_means[DENOISE_FEATURES], features[DENOISE_FEATURES];
filter_get_features(make_int3(x, y, 0), p_buffer, feature_means, NULL, pass_stride);
float design_row[DENOISE_FEATURES+1];
filter_get_design_row_transform(make_int3(x+dx, y+dy, 0), q_buffer, feature_means, pass_stride, features, *rank, design_row, transform, stride);
math_trimatrix_add_gramian_strided(XtWX, (*rank)+1, design_row, weight, stride);
math_vec3_add_strided(XtWY, (*rank)+1, design_row, weight * q_color, stride);
}
ccl_device_inline void kernel_filter_finalize(int x, int y, int w, int h,
float *buffer,
int *rank, int storage_stride,
float *XtWX, float3 *XtWY,
int4 buffer_params, int sample)
{
#ifdef __KERNEL_CPU__
const int stride = 1;
(void)storage_stride;
#else
const int stride = storage_stride;
#endif
math_trimatrix_vec3_solve(XtWX, XtWY, (*rank)+1, stride);
float3 final_color = XtWY[0];
if(buffer_params.z) {
float *combined_buffer = buffer + (y*buffer_params.y + x + buffer_params.x)*buffer_params.z;
final_color *= sample;
if(buffer_params.w) {
final_color.x += combined_buffer[buffer_params.w+0];
final_color.y += combined_buffer[buffer_params.w+1];
final_color.z += combined_buffer[buffer_params.w+2];
}
combined_buffer[0] = final_color.x;
combined_buffer[1] = final_color.y;
combined_buffer[2] = final_color.z;
}
else {
int idx = y*w+x;
buffer[idx] = final_color.x;
buffer[buffer_params.w + idx] = final_color.y;
buffer[2*buffer_params.w + idx] = final_color.z;
}
}
#undef STORAGE_TYPE
CCL_NAMESPACE_END
|
