Hugging Face's logo

camenduru
/
ncnn

Model card Files
xet
Community
ncnn
File size: 4,999 Bytes
be903e2
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
 
// Tencent is pleased to support the open source community by making ncnn available.
//
// Copyright (C) 2019 THL A29 Limited, a Tencent company. All rights reserved.
//
// Licensed under the BSD 3-Clause License (the "License"); you may not use this file except
// in compliance with the License. You may obtain a copy of the License at
//
// https://opensource.org/licenses/BSD-3-Clause
//
// 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.

#version 450

#if NCNN_fp16_storage
#extension GL_EXT_shader_16bit_storage: require
#endif
#if NCNN_fp16_arithmetic
#extension GL_EXT_shader_explicit_arithmetic_types_float16: require
#endif

layout (constant_id = 0) const int w = 0;
layout (constant_id = 1) const int h = 0;
layout (constant_id = 2) const int outw = 0;
layout (constant_id = 3) const int outh = 0;
layout (constant_id = 4) const int type_to = 0;
layout (constant_id = 5) const int rotate_from = 0;
layout (constant_id = 6) const int need_resize = 0;

layout (binding = 0) uniform sampler2D android_hardware_buffer_image;
#if NCNN_image_shader
layout (binding = 1, imfmtc1) writeonly uniform unfp image3D vkmat_blob;
layout (binding = 2, imfmtc4) writeonly uniform unfp image3D vkmat_pack4_blob;
#else
layout (binding = 1) writeonly buffer vkmat_blob { sfp vkmat_blob_data[]; };
layout (binding = 2) writeonly buffer vkmat_pack4_blob { sfpvec4 vkmat_pack4_blob_data[]; };
#endif

void main()
{
    int gx = int(gl_GlobalInvocationID.x);
    int gy = int(gl_GlobalInvocationID.y);
    int gz = int(gl_GlobalInvocationID.z);

    if (gx >= outw || gy >= outh || gz >= 1)
        return;

    vec2 pos;

    if (rotate_from == 1)
    {
        pos = vec2(gx, gy);
    }

    if (rotate_from == 2)
    {
        pos = vec2(outw - 1 - gx, gy);
    }

    if (rotate_from == 3)
    {
        pos = vec2(outw - 1 - gx, outh - 1 - gy);
    }

    if (rotate_from == 4)
    {
        pos = vec2(gx, outh - 1 - gy);
    }

    if (rotate_from == 5)
    {
        pos = vec2(gy, gx);
    }

    if (rotate_from == 6)
    {
        pos = vec2(gy, outw - 1 - gx);
    }

    if (rotate_from == 7)
    {
        pos = vec2(outh - 1 - gy, outw - 1 - gx);
    }

    if (rotate_from == 8)
    {
        pos = vec2(outh - 1 - gy, gx);
    }

    if (need_resize == 1)
    {
        if (rotate_from < 5) // 1 2 3 4
        {
            pos.x = pos.x * (float(w) / outw);
            pos.y = pos.y * (float(h) / outh);
        }
        else // 5 6 7 8
        {
            pos.x = pos.x * (float(w) / outh);
            pos.y = pos.y * (float(h) / outw);
        }
    }

    vec3 rgb = texture(android_hardware_buffer_image, pos).rgb * 255.f;

    const int outcstep = outw * outh / 4 * 4;

    if (type_to == 1) // PIXEL_RGB
    {
#if NCNN_image_shader
        image3d_st1(vkmat_blob, ivec3(gx, gy, 0), rgb.r);
        image3d_st1(vkmat_blob, ivec3(gx, gy, 1), rgb.g);
        image3d_st1(vkmat_blob, ivec3(gx, gy, 2), rgb.b);
#else
        ivec3 v_offset = (gy * outw + gx) + ivec3(0, 1, 2) * outcstep;

        buffer_st1(vkmat_blob_data, v_offset.r, afp(rgb.r));
        buffer_st1(vkmat_blob_data, v_offset.g, afp(rgb.g));
        buffer_st1(vkmat_blob_data, v_offset.b, afp(rgb.b));
#endif
    }

    if (type_to == 2) // PIXEL_BGR
    {
#if NCNN_image_shader
        image3d_st1(vkmat_blob, ivec3(gx, gy, 0), rgb.b);
        image3d_st1(vkmat_blob, ivec3(gx, gy, 1), rgb.g);
        image3d_st1(vkmat_blob, ivec3(gx, gy, 2), rgb.r);
#else
        ivec3 v_offset = (gy * outw + gx) + ivec3(0, 1, 2) * outcstep;

        buffer_st1(vkmat_blob_data, v_offset.r, afp(rgb.b));
        buffer_st1(vkmat_blob_data, v_offset.g, afp(rgb.g));
        buffer_st1(vkmat_blob_data, v_offset.b, afp(rgb.r));
#endif
    }

    if (type_to == 3) // PIXEL_GRAY
    {
        // coeffs for r g b = 0.299f, 0.587f, 0.114f
        float v = clamp(rgb.r * 0.299f + rgb.g * 0.587f + rgb.b * 0.114f, 0.f, 255.f);

#if NCNN_image_shader
        image3d_st1(vkmat_blob, ivec3(gx, gy, 0), v);
#else
        int v_offset = gy * outw + gx;

        buffer_st1(vkmat_blob_data, v_offset, afp(v));
#endif
    }

    if (type_to == 4) // PIXEL_RGBA
    {
        vec4 rgba;
        rgba.rgb = rgb;
        rgba.a = 255.f;

#if NCNN_image_shader
        image3d_st4(vkmat_pack4_blob, ivec3(gx, gy, 0), rgba);
#else
        int v_offset = gy * outw + gx;

        buffer_st4(vkmat_pack4_blob_data, v_offset, afpvec4(rgba));
#endif
    }

    if (type_to == 5) // PIXEL_BGRA
    {
        vec4 rgba;
        rgba.bgr = rgb;
        rgba.a = 255.f;

#if NCNN_image_shader
        image3d_st4(vkmat_pack4_blob, ivec3(gx, gy, 0), rgba);
#else
        int v_offset = gy * outw + gx;

        buffer_st4(vkmat_pack4_blob_data, v_offset, afpvec4(rgba));
#endif
    }
}