blob: 0af3d8bef8b8a3d74aab40038a8751b0a04122cc
1 | /* |
2 | * This file is part of FFmpeg. |
3 | * |
4 | * FFmpeg is free software; you can redistribute it and/or |
5 | * modify it under the terms of the GNU Lesser General Public |
6 | * License as published by the Free Software Foundation; either |
7 | * version 2.1 of the License, or (at your option) any later version. |
8 | * |
9 | * FFmpeg is distributed in the hope that it will be useful, |
10 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
11 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
12 | * Lesser General Public License for more details. |
13 | * |
14 | * You should have received a copy of the GNU Lesser General Public |
15 | * License along with FFmpeg; if not, write to the Free Software |
16 | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
17 | */ |
18 | |
19 | #include <assert.h> |
20 | #include <stdint.h> |
21 | #include <string.h> |
22 | |
23 | #include "config.h" |
24 | #include "libavutil/avassert.h" |
25 | #include "libavutil/attributes.h" |
26 | #include "libavutil/imgutils.h" |
27 | #include "avcodec.h" |
28 | #include "me_cmp.h" |
29 | #include "mpegvideoencdsp.h" |
30 | |
31 | static int try_8x8basis_c(int16_t rem[64], int16_t weight[64], |
32 | int16_t basis[64], int scale) |
33 | { |
34 | int i; |
35 | unsigned int sum = 0; |
36 | |
37 | for (i = 0; i < 8 * 8; i++) { |
38 | int b = rem[i] + ((basis[i] * scale + |
39 | (1 << (BASIS_SHIFT - RECON_SHIFT - 1))) >> |
40 | (BASIS_SHIFT - RECON_SHIFT)); |
41 | int w = weight[i]; |
42 | b >>= RECON_SHIFT; |
43 | av_assert2(-512 < b && b < 512); |
44 | |
45 | sum += (w * b) * (w * b) >> 4; |
46 | } |
47 | return sum >> 2; |
48 | } |
49 | |
50 | static void add_8x8basis_c(int16_t rem[64], int16_t basis[64], int scale) |
51 | { |
52 | int i; |
53 | |
54 | for (i = 0; i < 8 * 8; i++) |
55 | rem[i] += (basis[i] * scale + |
56 | (1 << (BASIS_SHIFT - RECON_SHIFT - 1))) >> |
57 | (BASIS_SHIFT - RECON_SHIFT); |
58 | } |
59 | |
60 | static int pix_sum_c(uint8_t *pix, int line_size) |
61 | { |
62 | int s = 0, i, j; |
63 | |
64 | for (i = 0; i < 16; i++) { |
65 | for (j = 0; j < 16; j += 8) { |
66 | s += pix[0]; |
67 | s += pix[1]; |
68 | s += pix[2]; |
69 | s += pix[3]; |
70 | s += pix[4]; |
71 | s += pix[5]; |
72 | s += pix[6]; |
73 | s += pix[7]; |
74 | pix += 8; |
75 | } |
76 | pix += line_size - 16; |
77 | } |
78 | return s; |
79 | } |
80 | |
81 | static int pix_norm1_c(uint8_t *pix, int line_size) |
82 | { |
83 | int s = 0, i, j; |
84 | uint32_t *sq = ff_square_tab + 256; |
85 | |
86 | for (i = 0; i < 16; i++) { |
87 | for (j = 0; j < 16; j += 8) { |
88 | #if HAVE_FAST_64BIT |
89 | register uint64_t x = *(uint64_t *) pix; |
90 | s += sq[x & 0xff]; |
91 | s += sq[(x >> 8) & 0xff]; |
92 | s += sq[(x >> 16) & 0xff]; |
93 | s += sq[(x >> 24) & 0xff]; |
94 | s += sq[(x >> 32) & 0xff]; |
95 | s += sq[(x >> 40) & 0xff]; |
96 | s += sq[(x >> 48) & 0xff]; |
97 | s += sq[(x >> 56) & 0xff]; |
98 | #else |
99 | register uint32_t x = *(uint32_t *) pix; |
100 | s += sq[x & 0xff]; |
101 | s += sq[(x >> 8) & 0xff]; |
102 | s += sq[(x >> 16) & 0xff]; |
103 | s += sq[(x >> 24) & 0xff]; |
104 | x = *(uint32_t *) (pix + 4); |
105 | s += sq[x & 0xff]; |
106 | s += sq[(x >> 8) & 0xff]; |
107 | s += sq[(x >> 16) & 0xff]; |
108 | s += sq[(x >> 24) & 0xff]; |
109 | #endif |
110 | pix += 8; |
111 | } |
112 | pix += line_size - 16; |
113 | } |
114 | return s; |
115 | } |
116 | |
117 | /* draw the edges of width 'w' of an image of size width, height */ |
118 | // FIXME: Check that this is OK for MPEG-4 interlaced. |
119 | static void draw_edges_8_c(uint8_t *buf, int wrap, int width, int height, |
120 | int w, int h, int sides) |
121 | { |
122 | uint8_t *ptr = buf, *last_line; |
123 | int i; |
124 | |
125 | /* left and right */ |
126 | for (i = 0; i < height; i++) { |
127 | memset(ptr - w, ptr[0], w); |
128 | memset(ptr + width, ptr[width - 1], w); |
129 | ptr += wrap; |
130 | } |
131 | |
132 | /* top and bottom + corners */ |
133 | buf -= w; |
134 | last_line = buf + (height - 1) * wrap; |
135 | if (sides & EDGE_TOP) |
136 | for (i = 0; i < h; i++) |
137 | // top |
138 | memcpy(buf - (i + 1) * wrap, buf, width + w + w); |
139 | if (sides & EDGE_BOTTOM) |
140 | for (i = 0; i < h; i++) |
141 | // bottom |
142 | memcpy(last_line + (i + 1) * wrap, last_line, width + w + w); |
143 | } |
144 | |
145 | /* 2x2 -> 1x1 */ |
146 | static void shrink22(uint8_t *dst, int dst_wrap, |
147 | const uint8_t *src, int src_wrap, |
148 | int width, int height) |
149 | { |
150 | int w; |
151 | const uint8_t *s1, *s2; |
152 | uint8_t *d; |
153 | |
154 | for (; height > 0; height--) { |
155 | s1 = src; |
156 | s2 = s1 + src_wrap; |
157 | d = dst; |
158 | for (w = width; w >= 4; w -= 4) { |
159 | d[0] = (s1[0] + s1[1] + s2[0] + s2[1] + 2) >> 2; |
160 | d[1] = (s1[2] + s1[3] + s2[2] + s2[3] + 2) >> 2; |
161 | d[2] = (s1[4] + s1[5] + s2[4] + s2[5] + 2) >> 2; |
162 | d[3] = (s1[6] + s1[7] + s2[6] + s2[7] + 2) >> 2; |
163 | s1 += 8; |
164 | s2 += 8; |
165 | d += 4; |
166 | } |
167 | for (; w > 0; w--) { |
168 | d[0] = (s1[0] + s1[1] + s2[0] + s2[1] + 2) >> 2; |
169 | s1 += 2; |
170 | s2 += 2; |
171 | d++; |
172 | } |
173 | src += 2 * src_wrap; |
174 | dst += dst_wrap; |
175 | } |
176 | } |
177 | |
178 | /* 4x4 -> 1x1 */ |
179 | static void shrink44(uint8_t *dst, int dst_wrap, |
180 | const uint8_t *src, int src_wrap, |
181 | int width, int height) |
182 | { |
183 | int w; |
184 | const uint8_t *s1, *s2, *s3, *s4; |
185 | uint8_t *d; |
186 | |
187 | for (; height > 0; height--) { |
188 | s1 = src; |
189 | s2 = s1 + src_wrap; |
190 | s3 = s2 + src_wrap; |
191 | s4 = s3 + src_wrap; |
192 | d = dst; |
193 | for (w = width; w > 0; w--) { |
194 | d[0] = (s1[0] + s1[1] + s1[2] + s1[3] + |
195 | s2[0] + s2[1] + s2[2] + s2[3] + |
196 | s3[0] + s3[1] + s3[2] + s3[3] + |
197 | s4[0] + s4[1] + s4[2] + s4[3] + 8) >> 4; |
198 | s1 += 4; |
199 | s2 += 4; |
200 | s3 += 4; |
201 | s4 += 4; |
202 | d++; |
203 | } |
204 | src += 4 * src_wrap; |
205 | dst += dst_wrap; |
206 | } |
207 | } |
208 | |
209 | /* 8x8 -> 1x1 */ |
210 | static void shrink88(uint8_t *dst, int dst_wrap, |
211 | const uint8_t *src, int src_wrap, |
212 | int width, int height) |
213 | { |
214 | int w, i; |
215 | |
216 | for (; height > 0; height--) { |
217 | for(w = width;w > 0; w--) { |
218 | int tmp = 0; |
219 | for (i = 0; i < 8; i++) { |
220 | tmp += src[0] + src[1] + src[2] + src[3] + |
221 | src[4] + src[5] + src[6] + src[7]; |
222 | src += src_wrap; |
223 | } |
224 | *(dst++) = (tmp + 32) >> 6; |
225 | src += 8 - 8 * src_wrap; |
226 | } |
227 | src += 8 * src_wrap - 8 * width; |
228 | dst += dst_wrap - width; |
229 | } |
230 | } |
231 | |
232 | av_cold void ff_mpegvideoencdsp_init(MpegvideoEncDSPContext *c, |
233 | AVCodecContext *avctx) |
234 | { |
235 | c->try_8x8basis = try_8x8basis_c; |
236 | c->add_8x8basis = add_8x8basis_c; |
237 | |
238 | c->shrink[0] = av_image_copy_plane; |
239 | c->shrink[1] = shrink22; |
240 | c->shrink[2] = shrink44; |
241 | c->shrink[3] = shrink88; |
242 | |
243 | c->pix_sum = pix_sum_c; |
244 | c->pix_norm1 = pix_norm1_c; |
245 | |
246 | c->draw_edges = draw_edges_8_c; |
247 | |
248 | if (ARCH_ARM) |
249 | ff_mpegvideoencdsp_init_arm(c, avctx); |
250 | if (ARCH_PPC) |
251 | ff_mpegvideoencdsp_init_ppc(c, avctx); |
252 | if (ARCH_X86) |
253 | ff_mpegvideoencdsp_init_x86(c, avctx); |
254 | if (ARCH_MIPS) |
255 | ff_mpegvideoencdsp_init_mips(c, avctx); |
256 | } |
257 |