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 "config.h"
20 #include "libavutil/attributes.h"
21 #include "libavutil/common.h"
22 #include "idctdsp.h"
23 #include "mathops.h"
24 #include "qpeldsp.h"
25 #include "wmv2dsp.h"
26
27 #define W0 2048
28 #define W1 2841 /* 2048*sqrt (2)*cos (1*pi/16) */
29 #define W2 2676 /* 2048*sqrt (2)*cos (2*pi/16) */
30 #define W3 2408 /* 2048*sqrt (2)*cos (3*pi/16) */
31 #define W4 2048 /* 2048*sqrt (2)*cos (4*pi/16) */
32 #define W5 1609 /* 2048*sqrt (2)*cos (5*pi/16) */
33 #define W6 1108 /* 2048*sqrt (2)*cos (6*pi/16) */
34 #define W7 565 /* 2048*sqrt (2)*cos (7*pi/16) */
35
wmv2_idct_row(short * b)36 static void wmv2_idct_row(short * b)
37 {
38 int s1, s2;
39 int a0, a1, a2, a3, a4, a5, a6, a7;
40
41 /* step 1 */
42 a1 = W1 * b[1] + W7 * b[7];
43 a7 = W7 * b[1] - W1 * b[7];
44 a5 = W5 * b[5] + W3 * b[3];
45 a3 = W3 * b[5] - W5 * b[3];
46 a2 = W2 * b[2] + W6 * b[6];
47 a6 = W6 * b[2] - W2 * b[6];
48 a0 = W0 * b[0] + W0 * b[4];
49 a4 = W0 * b[0] - W0 * b[4];
50
51 /* step 2 */
52 s1 = (int)(181U * (a1 - a5 + a7 - a3) + 128) >> 8; // 1, 3, 5, 7
53 s2 = (int)(181U * (a1 - a5 - a7 + a3) + 128) >> 8;
54
55 /* step 3 */
56 b[0] = (a0 + a2 + a1 + a5 + (1 << 7)) >> 8;
57 b[1] = (a4 + a6 + s1 + (1 << 7)) >> 8;
58 b[2] = (a4 - a6 + s2 + (1 << 7)) >> 8;
59 b[3] = (a0 - a2 + a7 + a3 + (1 << 7)) >> 8;
60 b[4] = (a0 - a2 - a7 - a3 + (1 << 7)) >> 8;
61 b[5] = (a4 - a6 - s2 + (1 << 7)) >> 8;
62 b[6] = (a4 + a6 - s1 + (1 << 7)) >> 8;
63 b[7] = (a0 + a2 - a1 - a5 + (1 << 7)) >> 8;
64 }
65
wmv2_idct_col(short * b)66 static void wmv2_idct_col(short * b)
67 {
68 int s1, s2;
69 int a0, a1, a2, a3, a4, a5, a6, a7;
70
71 /* step 1, with extended precision */
72 a1 = (W1 * b[8 * 1] + W7 * b[8 * 7] + 4) >> 3;
73 a7 = (W7 * b[8 * 1] - W1 * b[8 * 7] + 4) >> 3;
74 a5 = (W5 * b[8 * 5] + W3 * b[8 * 3] + 4) >> 3;
75 a3 = (W3 * b[8 * 5] - W5 * b[8 * 3] + 4) >> 3;
76 a2 = (W2 * b[8 * 2] + W6 * b[8 * 6] + 4) >> 3;
77 a6 = (W6 * b[8 * 2] - W2 * b[8 * 6] + 4) >> 3;
78 a0 = (W0 * b[8 * 0] + W0 * b[8 * 4] ) >> 3;
79 a4 = (W0 * b[8 * 0] - W0 * b[8 * 4] ) >> 3;
80
81 /* step 2 */
82 s1 = (int)(181U * (a1 - a5 + a7 - a3) + 128) >> 8;
83 s2 = (int)(181U * (a1 - a5 - a7 + a3) + 128) >> 8;
84
85 /* step 3 */
86 b[8 * 0] = (a0 + a2 + a1 + a5 + (1 << 13)) >> 14;
87 b[8 * 1] = (a4 + a6 + s1 + (1 << 13)) >> 14;
88 b[8 * 2] = (a4 - a6 + s2 + (1 << 13)) >> 14;
89 b[8 * 3] = (a0 - a2 + a7 + a3 + (1 << 13)) >> 14;
90
91 b[8 * 4] = (a0 - a2 - a7 - a3 + (1 << 13)) >> 14;
92 b[8 * 5] = (a4 - a6 - s2 + (1 << 13)) >> 14;
93 b[8 * 6] = (a4 + a6 - s1 + (1 << 13)) >> 14;
94 b[8 * 7] = (a0 + a2 - a1 - a5 + (1 << 13)) >> 14;
95 }
96
wmv2_idct_add_c(uint8_t * dest,ptrdiff_t line_size,int16_t * block)97 static void wmv2_idct_add_c(uint8_t *dest, ptrdiff_t line_size, int16_t *block)
98 {
99 int i;
100
101 for (i = 0; i < 64; i += 8)
102 wmv2_idct_row(block + i);
103 for (i = 0; i < 8; i++)
104 wmv2_idct_col(block + i);
105
106 for (i = 0; i < 8; i++) {
107 dest[0] = av_clip_uint8(dest[0] + block[0]);
108 dest[1] = av_clip_uint8(dest[1] + block[1]);
109 dest[2] = av_clip_uint8(dest[2] + block[2]);
110 dest[3] = av_clip_uint8(dest[3] + block[3]);
111 dest[4] = av_clip_uint8(dest[4] + block[4]);
112 dest[5] = av_clip_uint8(dest[5] + block[5]);
113 dest[6] = av_clip_uint8(dest[6] + block[6]);
114 dest[7] = av_clip_uint8(dest[7] + block[7]);
115 dest += line_size;
116 block += 8;
117 }
118 }
119
wmv2_idct_put_c(uint8_t * dest,ptrdiff_t line_size,int16_t * block)120 static void wmv2_idct_put_c(uint8_t *dest, ptrdiff_t line_size, int16_t *block)
121 {
122 int i;
123
124 for (i = 0; i < 64; i += 8)
125 wmv2_idct_row(block + i);
126 for (i = 0; i < 8; i++)
127 wmv2_idct_col(block + i);
128
129 for (i = 0; i < 8; i++) {
130 dest[0] = av_clip_uint8(block[0]);
131 dest[1] = av_clip_uint8(block[1]);
132 dest[2] = av_clip_uint8(block[2]);
133 dest[3] = av_clip_uint8(block[3]);
134 dest[4] = av_clip_uint8(block[4]);
135 dest[5] = av_clip_uint8(block[5]);
136 dest[6] = av_clip_uint8(block[6]);
137 dest[7] = av_clip_uint8(block[7]);
138 dest += line_size;
139 block += 8;
140 }
141 }
142
wmv2_mspel8_h_lowpass(uint8_t * dst,const uint8_t * src,int dstStride,int srcStride,int h)143 static void wmv2_mspel8_h_lowpass(uint8_t *dst, const uint8_t *src,
144 int dstStride, int srcStride, int h)
145 {
146 const uint8_t *cm = ff_crop_tab + MAX_NEG_CROP;
147 int i;
148
149 for (i = 0; i < h; i++) {
150 dst[0] = cm[(9 * (src[0] + src[1]) - (src[-1] + src[2]) + 8) >> 4];
151 dst[1] = cm[(9 * (src[1] + src[2]) - (src[0] + src[3]) + 8) >> 4];
152 dst[2] = cm[(9 * (src[2] + src[3]) - (src[1] + src[4]) + 8) >> 4];
153 dst[3] = cm[(9 * (src[3] + src[4]) - (src[2] + src[5]) + 8) >> 4];
154 dst[4] = cm[(9 * (src[4] + src[5]) - (src[3] + src[6]) + 8) >> 4];
155 dst[5] = cm[(9 * (src[5] + src[6]) - (src[4] + src[7]) + 8) >> 4];
156 dst[6] = cm[(9 * (src[6] + src[7]) - (src[5] + src[8]) + 8) >> 4];
157 dst[7] = cm[(9 * (src[7] + src[8]) - (src[6] + src[9]) + 8) >> 4];
158 dst += dstStride;
159 src += srcStride;
160 }
161 }
162
wmv2_mspel8_v_lowpass(uint8_t * dst,const uint8_t * src,int dstStride,int srcStride,int w)163 static void wmv2_mspel8_v_lowpass(uint8_t *dst, const uint8_t *src,
164 int dstStride, int srcStride, int w)
165 {
166 const uint8_t *cm = ff_crop_tab + MAX_NEG_CROP;
167 int i;
168
169 for (i = 0; i < w; i++) {
170 const int src_1 = src[-srcStride];
171 const int src0 = src[0];
172 const int src1 = src[srcStride];
173 const int src2 = src[2 * srcStride];
174 const int src3 = src[3 * srcStride];
175 const int src4 = src[4 * srcStride];
176 const int src5 = src[5 * srcStride];
177 const int src6 = src[6 * srcStride];
178 const int src7 = src[7 * srcStride];
179 const int src8 = src[8 * srcStride];
180 const int src9 = src[9 * srcStride];
181 dst[0 * dstStride] = cm[(9 * (src0 + src1) - (src_1 + src2) + 8) >> 4];
182 dst[1 * dstStride] = cm[(9 * (src1 + src2) - (src0 + src3) + 8) >> 4];
183 dst[2 * dstStride] = cm[(9 * (src2 + src3) - (src1 + src4) + 8) >> 4];
184 dst[3 * dstStride] = cm[(9 * (src3 + src4) - (src2 + src5) + 8) >> 4];
185 dst[4 * dstStride] = cm[(9 * (src4 + src5) - (src3 + src6) + 8) >> 4];
186 dst[5 * dstStride] = cm[(9 * (src5 + src6) - (src4 + src7) + 8) >> 4];
187 dst[6 * dstStride] = cm[(9 * (src6 + src7) - (src5 + src8) + 8) >> 4];
188 dst[7 * dstStride] = cm[(9 * (src7 + src8) - (src6 + src9) + 8) >> 4];
189 src++;
190 dst++;
191 }
192 }
193
put_mspel8_mc10_c(uint8_t * dst,const uint8_t * src,ptrdiff_t stride)194 static void put_mspel8_mc10_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)
195 {
196 uint8_t half[64];
197
198 wmv2_mspel8_h_lowpass(half, src, 8, stride, 8);
199 ff_put_pixels8_l2_8(dst, src, half, stride, stride, 8, 8);
200 }
201
put_mspel8_mc20_c(uint8_t * dst,const uint8_t * src,ptrdiff_t stride)202 static void put_mspel8_mc20_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)
203 {
204 wmv2_mspel8_h_lowpass(dst, src, stride, stride, 8);
205 }
206
put_mspel8_mc30_c(uint8_t * dst,const uint8_t * src,ptrdiff_t stride)207 static void put_mspel8_mc30_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)
208 {
209 uint8_t half[64];
210
211 wmv2_mspel8_h_lowpass(half, src, 8, stride, 8);
212 ff_put_pixels8_l2_8(dst, src + 1, half, stride, stride, 8, 8);
213 }
214
put_mspel8_mc02_c(uint8_t * dst,const uint8_t * src,ptrdiff_t stride)215 static void put_mspel8_mc02_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)
216 {
217 wmv2_mspel8_v_lowpass(dst, src, stride, stride, 8);
218 }
219
put_mspel8_mc12_c(uint8_t * dst,const uint8_t * src,ptrdiff_t stride)220 static void put_mspel8_mc12_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)
221 {
222 uint8_t halfH[88];
223 uint8_t halfV[64];
224 uint8_t halfHV[64];
225
226 wmv2_mspel8_h_lowpass(halfH, src - stride, 8, stride, 11);
227 wmv2_mspel8_v_lowpass(halfV, src, 8, stride, 8);
228 wmv2_mspel8_v_lowpass(halfHV, halfH + 8, 8, 8, 8);
229 ff_put_pixels8_l2_8(dst, halfV, halfHV, stride, 8, 8, 8);
230 }
231
put_mspel8_mc32_c(uint8_t * dst,const uint8_t * src,ptrdiff_t stride)232 static void put_mspel8_mc32_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)
233 {
234 uint8_t halfH[88];
235 uint8_t halfV[64];
236 uint8_t halfHV[64];
237
238 wmv2_mspel8_h_lowpass(halfH, src - stride, 8, stride, 11);
239 wmv2_mspel8_v_lowpass(halfV, src + 1, 8, stride, 8);
240 wmv2_mspel8_v_lowpass(halfHV, halfH + 8, 8, 8, 8);
241 ff_put_pixels8_l2_8(dst, halfV, halfHV, stride, 8, 8, 8);
242 }
243
put_mspel8_mc22_c(uint8_t * dst,const uint8_t * src,ptrdiff_t stride)244 static void put_mspel8_mc22_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)
245 {
246 uint8_t halfH[88];
247
248 wmv2_mspel8_h_lowpass(halfH, src - stride, 8, stride, 11);
249 wmv2_mspel8_v_lowpass(dst, halfH + 8, stride, 8, 8);
250 }
251
ff_wmv2dsp_init(WMV2DSPContext * c)252 av_cold void ff_wmv2dsp_init(WMV2DSPContext *c)
253 {
254 c->idct_add = wmv2_idct_add_c;
255 c->idct_put = wmv2_idct_put_c;
256 c->idct_perm = FF_IDCT_PERM_NONE;
257
258 c->put_mspel_pixels_tab[0] = ff_put_pixels8x8_c;
259 c->put_mspel_pixels_tab[1] = put_mspel8_mc10_c;
260 c->put_mspel_pixels_tab[2] = put_mspel8_mc20_c;
261 c->put_mspel_pixels_tab[3] = put_mspel8_mc30_c;
262 c->put_mspel_pixels_tab[4] = put_mspel8_mc02_c;
263 c->put_mspel_pixels_tab[5] = put_mspel8_mc12_c;
264 c->put_mspel_pixels_tab[6] = put_mspel8_mc22_c;
265 c->put_mspel_pixels_tab[7] = put_mspel8_mc32_c;
266
267 #if ARCH_MIPS
268 ff_wmv2dsp_init_mips(c);
269 #endif
270 }
271