1 /* ScummVM - Graphic Adventure Engine
2  *
3  * ScummVM is the legal property of its developers, whose names
4  * are too numerous to list here. Please refer to the COPYRIGHT
5  * file distributed with this source distribution.
6  *
7  * This program is free software; you can redistribute it and/or
8  * modify it under the terms of the GNU General Public License
9  * as published by the Free Software Foundation; either version 2
10  * of the License, or (at your option) any later version.
11  *
12  * This program is distributed in the hope that it will be useful,
13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
15  * GNU General Public License for more details.
16  *
17  * You should have received a copy of the GNU General Public License
18  * along with this program; if not, write to the Free Software
19  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
20  *
21  */
22 
23 /* VLC code
24  *
25  * Original copyright note: * Intel Indeo 4 (IV41, IV42, etc.) video decoder for ffmpeg
26  * written, produced, and directed by Alan Smithee
27  */
28 
29 #include "image/codecs/indeo/indeo_dsp.h"
30 
31 namespace Image {
32 namespace Indeo {
33 
34 /**
35  * butterfly operation for the inverse Haar transform
36  */
37 #define IVI_HAAR_BFLY(s1, s2, o1, o2, t) \
38 	t  = ((s1) - (s2)) >> 1;\
39 	o1 = ((s1) + (s2)) >> 1;\
40 	o2 = (t);\
41 
42 /**
43  * inverse 8-point Haar transform
44  */
45 #define INV_HAAR8(s1, s5, s3, s7, s2, s4, s6, s8,\
46 				  d1, d2, d3, d4, d5, d6, d7, d8,\
47 				  t0, t1, t2, t3, t4, t5, t6, t7, t8) {\
48 	t1 = (s1) << 1; t5 = (s5) << 1;\
49 	IVI_HAAR_BFLY(t1, t5, t1, t5, t0); IVI_HAAR_BFLY(t1, s3, t1, t3, t0);\
50 	IVI_HAAR_BFLY(t5, s7, t5, t7, t0); IVI_HAAR_BFLY(t1, s2, t1, t2, t0);\
51 	IVI_HAAR_BFLY(t3, s4, t3, t4, t0); IVI_HAAR_BFLY(t5, s6, t5, t6, t0);\
52 	IVI_HAAR_BFLY(t7, s8, t7, t8, t0);\
53 	d1 = COMPENSATE(t1);\
54 	d2 = COMPENSATE(t2);\
55 	d3 = COMPENSATE(t3);\
56 	d4 = COMPENSATE(t4);\
57 	d5 = COMPENSATE(t5);\
58 	d6 = COMPENSATE(t6);\
59 	d7 = COMPENSATE(t7);\
60 	d8 = COMPENSATE(t8); }
61 
62 /**
63  * inverse 4-point Haar transform
64  */
65 #define INV_HAAR4(s1, s3, s5, s7, d1, d2, d3, d4, t0, t1, t2, t3, t4) {\
66 	IVI_HAAR_BFLY(s1, s3, t0, t1, t4);\
67 	IVI_HAAR_BFLY(t0, s5, t2, t3, t4);\
68 	d1 = COMPENSATE(t2);\
69 	d2 = COMPENSATE(t3);\
70 	IVI_HAAR_BFLY(t1, s7, t2, t3, t4);\
71 	d3 = COMPENSATE(t2);\
72 	d4 = COMPENSATE(t3); }
73 
ffIviInverseHaar8x8(const int32 * in,int16 * out,uint32 pitch,const uint8 * flags)74 void IndeoDSP::ffIviInverseHaar8x8(const int32 *in, int16 *out, uint32 pitch,
75 							 const uint8 *flags) {
76 	int32 tmp[64];
77 	int t0, t1, t2, t3, t4, t5, t6, t7, t8;
78 
79 	// apply the InvHaar8 to all columns
80 #define COMPENSATE(x) (x)
81 	const int32 *src = in;
82 	int32 *dst = tmp;
83 	for (int i = 0; i < 8; i++) {
84 		if (flags[i]) {
85 			// pre-scaling
86 			int shift = !(i & 4);
87 			int sp1 = src[ 0] << shift;
88 			int sp2 = src[ 8] << shift;
89 			int sp3 = src[16] << shift;
90 			int sp4 = src[24] << shift;
91 			INV_HAAR8(    sp1,     sp2,     sp3,     sp4,
92 					  src[32], src[40], src[48], src[56],
93 					  dst[ 0], dst[ 8], dst[16], dst[24],
94 					  dst[32], dst[40], dst[48], dst[56],
95 					  t0, t1, t2, t3, t4, t5, t6, t7, t8);
96 		} else {
97 			dst[ 0] = dst[ 8] = dst[16] = dst[24] =
98 			dst[32] = dst[40] = dst[48] = dst[56] = 0;
99 		}
100 
101 		src++;
102 		dst++;
103 	}
104 #undef  COMPENSATE
105 
106 	// apply the InvHaar8 to all rows
107 #define COMPENSATE(x) (x)
108 	src = tmp;
109 	for (int i = 0; i < 8; i++) {
110 		if (!src[0] && !src[1] && !src[2] && !src[3] &&
111 				!src[4] && !src[5] && !src[6] && !src[7]) {
112 			memset(out, 0, 8 * sizeof(out[0]));
113 		} else {
114 			INV_HAAR8(src[0], src[1], src[2], src[3],
115 					  src[4], src[5], src[6], src[7],
116 					  out[0], out[1], out[2], out[3],
117 					  out[4], out[5], out[6], out[7],
118 					  t0, t1, t2, t3, t4, t5, t6, t7, t8);
119 		}
120 		src += 8;
121 		out += pitch;
122 	}
123 #undef  COMPENSATE
124 }
125 
ffIviRowHaar8(const int32 * in,int16 * out,uint32 pitch,const uint8 * flags)126 void IndeoDSP::ffIviRowHaar8(const int32 *in, int16 *out, uint32 pitch,
127 					  const uint8 *flags) {
128 	int t0, t1, t2, t3, t4, t5, t6, t7, t8;
129 
130 	// apply the InvHaar8 to all rows
131 #define COMPENSATE(x) (x)
132 	for (int i = 0; i < 8; i++) {
133 		if (   !in[0] && !in[1] && !in[2] && !in[3]
134 			&& !in[4] && !in[5] && !in[6] && !in[7]) {
135 			memset(out, 0, 8 * sizeof(out[0]));
136 		} else {
137 			INV_HAAR8(in[0],  in[1],  in[2],  in[3],
138 					  in[4],  in[5],  in[6],  in[7],
139 					  out[0], out[1], out[2], out[3],
140 					  out[4], out[5], out[6], out[7],
141 					  t0, t1, t2, t3, t4, t5, t6, t7, t8);
142 		}
143 		in  += 8;
144 		out += pitch;
145 	}
146 #undef  COMPENSATE
147 }
148 
ffIviColHaar8(const int32 * in,int16 * out,uint32 pitch,const uint8 * flags)149 void IndeoDSP::ffIviColHaar8(const int32 *in, int16 *out, uint32 pitch,
150 					  const uint8 *flags) {
151 	int t0, t1, t2, t3, t4, t5, t6, t7, t8;
152 
153 	// apply the InvHaar8 to all columns
154 #define COMPENSATE(x) (x)
155 	for (int i = 0; i < 8; i++) {
156 		if (flags[i]) {
157 			INV_HAAR8(in[ 0], in[ 8], in[16], in[24],
158 					  in[32], in[40], in[48], in[56],
159 					  out[0 * pitch], out[1 * pitch],
160 					  out[2 * pitch], out[3 * pitch],
161 					  out[4 * pitch], out[5 * pitch],
162 					  out[6 * pitch], out[7 * pitch],
163 					  t0, t1, t2, t3, t4, t5, t6, t7, t8);
164 		} else {
165 			out[0 * pitch] = out[1 * pitch] =
166 			out[2 * pitch] = out[3 * pitch] =
167 			out[4 * pitch] = out[5 * pitch] =
168 			out[6 * pitch] = out[7 * pitch] = 0;
169 		}
170 
171 		in++;
172 		out++;
173 	}
174 #undef  COMPENSATE
175 }
176 
ffIviInverseHaar4x4(const int32 * in,int16 * out,uint32 pitch,const uint8 * flags)177 void IndeoDSP::ffIviInverseHaar4x4(const int32 *in, int16 *out, uint32 pitch,
178 							 const uint8 *flags) {
179 	int32 tmp[16];
180 	int t0, t1, t2, t3, t4;
181 
182 	// apply the InvHaar4 to all columns
183 #define COMPENSATE(x) (x)
184 	const int32 *src = in;
185 	int32 *dst = tmp;
186 	for (int i = 0; i < 4; i++) {
187 		if (flags[i]) {
188 			// pre-scaling
189 			int shift = !(i & 2);
190 			int sp1 = src[0] << shift;
191 			int sp2 = src[4] << shift;
192 			INV_HAAR4(   sp1,    sp2, src[8], src[12],
193 					  dst[0], dst[4], dst[8], dst[12],
194 					  t0, t1, t2, t3, t4);
195 		} else {
196 			dst[0] = dst[4] = dst[8] = dst[12] = 0;
197 		}
198 
199 		src++;
200 		dst++;
201 	}
202 #undef  COMPENSATE
203 
204 	// apply the InvHaar8 to all rows
205 #define COMPENSATE(x) (x)
206 	src = tmp;
207 	for (int i = 0; i < 4; i++) {
208 		if (!src[0] && !src[1] && !src[2] && !src[3]) {
209 			memset(out, 0, 4 * sizeof(out[0]));
210 		} else {
211 			INV_HAAR4(src[0], src[1], src[2], src[3],
212 					  out[0], out[1], out[2], out[3],
213 					  t0, t1, t2, t3, t4);
214 		}
215 		src += 4;
216 		out += pitch;
217 	}
218 #undef  COMPENSATE
219 }
220 
ffIviRowHaar4(const int32 * in,int16 * out,uint32 pitch,const uint8 * flags)221 void IndeoDSP::ffIviRowHaar4(const int32 *in, int16 *out, uint32 pitch,
222 					  const uint8 *flags) {
223 	int t0, t1, t2, t3, t4;
224 
225 	// apply the InvHaar4 to all rows
226 #define COMPENSATE(x) (x)
227 	for (int i = 0; i < 4; i++) {
228 		if (!in[0] && !in[1] && !in[2] && !in[3]) {
229 			memset(out, 0, 4 * sizeof(out[0]));
230 		} else {
231 			INV_HAAR4(in[0], in[1], in[2], in[3],
232 					  out[0], out[1], out[2], out[3],
233 					  t0, t1, t2, t3, t4);
234 		}
235 		in  += 4;
236 		out += pitch;
237 	}
238 #undef  COMPENSATE
239 }
240 
ffIviColHaar4(const int32 * in,int16 * out,uint32 pitch,const uint8 * flags)241 void IndeoDSP::ffIviColHaar4(const int32 *in, int16 *out, uint32 pitch,
242 					  const uint8 *flags) {
243 	int t0, t1, t2, t3, t4;
244 
245 	// apply the InvHaar8 to all columns
246 #define COMPENSATE(x) (x)
247 	for (int i = 0; i < 4; i++) {
248 		if (flags[i]) {
249 			INV_HAAR4(in[0], in[4], in[8], in[12],
250 					  out[0 * pitch], out[1 * pitch],
251 					  out[2 * pitch], out[3 * pitch],
252 					  t0, t1, t2, t3, t4);
253 		} else {
254 			out[0 * pitch] = out[1 * pitch] =
255 			out[2 * pitch] = out[3 * pitch] = 0;
256 		}
257 
258 		in++;
259 		out++;
260 	}
261 #undef  COMPENSATE
262 }
263 
ffIviDcHaar2d(const int32 * in,int16 * out,uint32 pitch,int blkSize)264 void IndeoDSP::ffIviDcHaar2d(const int32 *in, int16 *out, uint32 pitch,
265 					   int blkSize) {
266 	int16 dcCoeff = (*in + 0) >> 3;
267 
268 	for (int y = 0; y < blkSize; out += pitch, y++) {
269 		for (int x = 0; x < blkSize; x++)
270 			out[x] = dcCoeff;
271 	}
272 }
273 
274 //* butterfly operation for the inverse slant transform
275 #define IVI_SLANT_BFLY(s1, s2, o1, o2, t) \
276 	t  = (s1) - (s2);\
277 	o1 = (s1) + (s2);\
278 	o2 = (t);\
279 
280 //* This is a reflection a,b = 1/2, 5/4 for the inverse slant transform
281 #define IVI_IREFLECT(s1, s2, o1, o2, t) \
282 	t  = (((s1) + (s2)*2 + 2) >> 2) + (s1);\
283 	o2 = (((s1)*2 - (s2) + 2) >> 2) - (s2);\
284 	o1 = (t);\
285 
286 //* This is a reflection a,b = 1/2, 7/8 for the inverse slant transform
287 #define IVI_SLANT_PART4(s1, s2, o1, o2, t) \
288 	t  = (s2) + (((s1)*4  - (s2) + 4) >> 3);\
289 	o2 = (s1) + ((-(s1) - (s2)*4 + 4) >> 3);\
290 	o1 = (t);\
291 
292 //* inverse slant8 transform
293 #define IVI_INV_SLANT8(s1, s4, s8, s5, s2, s6, s3, s7,\
294 					   d1, d2, d3, d4, d5, d6, d7, d8,\
295 					   t0, t1, t2, t3, t4, t5, t6, t7, t8) {\
296 	IVI_SLANT_PART4(s4, s5, t4, t5, t0);\
297 \
298 	IVI_SLANT_BFLY(s1, t5, t1, t5, t0); IVI_SLANT_BFLY(s2, s6, t2, t6, t0);\
299 	IVI_SLANT_BFLY(s7, s3, t7, t3, t0); IVI_SLANT_BFLY(t4, s8, t4, t8, t0);\
300 \
301 	IVI_SLANT_BFLY(t1, t2, t1, t2, t0); IVI_IREFLECT  (t4, t3, t4, t3, t0);\
302 	IVI_SLANT_BFLY(t5, t6, t5, t6, t0); IVI_IREFLECT  (t8, t7, t8, t7, t0);\
303 	IVI_SLANT_BFLY(t1, t4, t1, t4, t0); IVI_SLANT_BFLY(t2, t3, t2, t3, t0);\
304 	IVI_SLANT_BFLY(t5, t8, t5, t8, t0); IVI_SLANT_BFLY(t6, t7, t6, t7, t0);\
305 	d1 = COMPENSATE(t1);\
306 	d2 = COMPENSATE(t2);\
307 	d3 = COMPENSATE(t3);\
308 	d4 = COMPENSATE(t4);\
309 	d5 = COMPENSATE(t5);\
310 	d6 = COMPENSATE(t6);\
311 	d7 = COMPENSATE(t7);\
312 	d8 = COMPENSATE(t8);}
313 
314 //* inverse slant4 transform
315 #define IVI_INV_SLANT4(s1, s4, s2, s3, d1, d2, d3, d4, t0, t1, t2, t3, t4) {\
316 	IVI_SLANT_BFLY(s1, s2, t1, t2, t0); IVI_IREFLECT  (s4, s3, t4, t3, t0);\
317 \
318 	IVI_SLANT_BFLY(t1, t4, t1, t4, t0); IVI_SLANT_BFLY(t2, t3, t2, t3, t0);\
319 	d1 = COMPENSATE(t1);\
320 	d2 = COMPENSATE(t2);\
321 	d3 = COMPENSATE(t3);\
322 	d4 = COMPENSATE(t4);}
323 
ffIviInverseSlant8x8(const int32 * in,int16 * out,uint32 pitch,const uint8 * flags)324 void IndeoDSP::ffIviInverseSlant8x8(const int32 *in, int16 *out, uint32 pitch, const uint8 *flags) {
325 	int32 tmp[64];
326 	int t0, t1, t2, t3, t4, t5, t6, t7, t8;
327 
328 #define COMPENSATE(x) (x)
329 	const int32 *src = in;
330 	int32 *dst = tmp;
331 	for (int i = 0; i < 8; i++) {
332 		if (flags[i]) {
333 			IVI_INV_SLANT8(src[0], src[8], src[16], src[24], src[32], src[40], src[48], src[56],
334 						   dst[0], dst[8], dst[16], dst[24], dst[32], dst[40], dst[48], dst[56],
335 						   t0, t1, t2, t3, t4, t5, t6, t7, t8);
336 		} else {
337 			dst[0] = dst[8] = dst[16] = dst[24] = dst[32] = dst[40] = dst[48] = dst[56] = 0;
338 		}
339 
340 		src++;
341 		dst++;
342 	}
343 #undef COMPENSATE
344 
345 #define COMPENSATE(x) (((x) + 1)>>1)
346 	src = tmp;
347 	for (int i = 0; i < 8; i++) {
348 		if (!src[0] && !src[1] && !src[2] && !src[3] && !src[4] && !src[5] && !src[6] && !src[7]) {
349 			memset(out, 0, 8*sizeof(out[0]));
350 		} else {
351 			IVI_INV_SLANT8(src[0], src[1], src[2], src[3], src[4], src[5], src[6], src[7],
352 						   out[0], out[1], out[2], out[3], out[4], out[5], out[6], out[7],
353 						   t0, t1, t2, t3, t4, t5, t6, t7, t8);
354 		}
355 		src += 8;
356 		out += pitch;
357 	}
358 #undef COMPENSATE
359 }
360 
ffIviInverseSlant4x4(const int32 * in,int16 * out,uint32 pitch,const uint8 * flags)361 void IndeoDSP::ffIviInverseSlant4x4(const int32 *in, int16 *out, uint32 pitch, const uint8 *flags) {
362 	int32 tmp[16];
363 	int t0, t1, t2, t3, t4;
364 
365 #define COMPENSATE(x) (x)
366 	const int32 *src = in;
367 	int32 *dst = tmp;
368 	for (int i = 0; i < 4; i++) {
369 		if (flags[i]) {
370 			IVI_INV_SLANT4(src[0], src[4], src[8], src[12],
371 						   dst[0], dst[4], dst[8], dst[12],
372 						   t0, t1, t2, t3, t4);
373 		} else {
374 			dst[0] = dst[4] = dst[8] = dst[12] = 0;
375 		}
376 		src++;
377 		dst++;
378 	}
379 #undef COMPENSATE
380 
381 #define COMPENSATE(x) (((x) + 1)>>1)
382 	src = tmp;
383 	for (int i = 0; i < 4; i++) {
384 		if (!src[0] && !src[1] && !src[2] && !src[3]) {
385 			out[0] = out[1] = out[2] = out[3] = 0;
386 		} else {
387 			IVI_INV_SLANT4(src[0], src[1], src[2], src[3],
388 						   out[0], out[1], out[2], out[3],
389 						   t0, t1, t2, t3, t4);
390 		}
391 		src += 4;
392 		out += pitch;
393 	}
394 #undef COMPENSATE
395 }
396 
ffIviDcSlant2d(const int32 * in,int16 * out,uint32 pitch,int blkSize)397 void IndeoDSP::ffIviDcSlant2d(const int32 *in, int16 *out, uint32 pitch,
398 		int blkSize) {
399 	int16 dcCoeff = (*in + 1) >> 1;
400 
401 	for (int y = 0; y < blkSize; out += pitch, y++) {
402 		for (int x = 0; x < blkSize; x++)
403 			out[x] = dcCoeff;
404 	}
405 }
406 
ffIviRowSlant8(const int32 * in,int16 * out,uint32 pitch,const uint8 * flags)407 void IndeoDSP::ffIviRowSlant8(const int32 *in, int16 *out, uint32 pitch,
408 		const uint8 *flags) {
409 	int t0, t1, t2, t3, t4, t5, t6, t7, t8;
410 
411 #define COMPENSATE(x) (((x) + 1)>>1)
412 	for (int i = 0; i < 8; i++) {
413 		if (!in[0] && !in[1] && !in[2] && !in[3] && !in[4] && !in[5] && !in[6] && !in[7]) {
414 			memset(out, 0, 8*sizeof(out[0]));
415 		} else {
416 			IVI_INV_SLANT8( in[0],  in[1],  in[2],  in[3],  in[4],  in[5],  in[6],  in[7],
417 						   out[0], out[1], out[2], out[3], out[4], out[5], out[6], out[7],
418 						   t0, t1, t2, t3, t4, t5, t6, t7, t8);
419 		}
420 		in += 8;
421 		out += pitch;
422 	}
423 #undef COMPENSATE
424 }
425 
ffIviDcRowSlant(const int32 * in,int16 * out,uint32 pitch,int blkSize)426 void IndeoDSP::ffIviDcRowSlant(const int32 *in, int16 *out, uint32 pitch, int blkSize) {
427 	int16 dcCoeff = (*in + 1) >> 1;
428 
429 	for (int x = 0; x < blkSize; x++)
430 		out[x] = dcCoeff;
431 
432 	out += pitch;
433 
434 	for (int y = 1; y < blkSize; out += pitch, y++) {
435 		for (int x = 0; x < blkSize; x++)
436 			out[x] = 0;
437 	}
438 }
439 
ffIviColSlant8(const int32 * in,int16 * out,uint32 pitch,const uint8 * flags)440 void IndeoDSP::ffIviColSlant8(const int32 *in, int16 *out, uint32 pitch, const uint8 *flags) {
441 	int t0, t1, t2, t3, t4, t5, t6, t7, t8;
442 
443 	int row2 = pitch << 1;
444 	int row4 = pitch << 2;
445 	int row8 = pitch << 3;
446 
447 #define COMPENSATE(x) (((x) + 1)>>1)
448 	for (int i = 0; i < 8; i++) {
449 		if (flags[i]) {
450 			IVI_INV_SLANT8(in[0], in[8], in[16], in[24], in[32], in[40], in[48], in[56],
451 						   out[0], out[pitch], out[row2], out[row2 + pitch], out[row4],
452 						   out[row4 + pitch],  out[row4 + row2], out[row8 - pitch],
453 						   t0, t1, t2, t3, t4, t5, t6, t7, t8);
454 		} else {
455 			out[0] = out[pitch] = out[row2] = out[row2 + pitch] = out[row4] =
456 			out[row4 + pitch] =  out[row4 + row2] = out[row8 - pitch] = 0;
457 		}
458 
459 		in++;
460 		out++;
461 	}
462 #undef COMPENSATE
463 }
464 
ffIviDcColSlant(const int32 * in,int16 * out,uint32 pitch,int blkSize)465 void IndeoDSP::ffIviDcColSlant(const int32 *in, int16 *out, uint32 pitch, int blkSize) {
466 	int16 dcCoeff = (*in + 1) >> 1;
467 
468 	for (int y = 0; y < blkSize; out += pitch, y++) {
469 		out[0] = dcCoeff;
470 		for (int x = 1; x < blkSize; x++)
471 			out[x] = 0;
472 	}
473 }
474 
ffIviRowSlant4(const int32 * in,int16 * out,uint32 pitch,const uint8 * flags)475 void IndeoDSP::ffIviRowSlant4(const int32 *in, int16 *out,
476 		uint32 pitch, const uint8 *flags) {
477 	int t0, t1, t2, t3, t4;
478 
479 #define COMPENSATE(x) (((x) + 1)>>1)
480 	for (int i = 0; i < 4; i++) {
481 		if (!in[0] && !in[1] && !in[2] && !in[3]) {
482 			memset(out, 0, 4*sizeof(out[0]));
483 		} else {
484 			IVI_INV_SLANT4( in[0],  in[1],  in[2],  in[3],
485 						   out[0], out[1], out[2], out[3],
486 						   t0, t1, t2, t3, t4);
487 		}
488 		in  += 4;
489 		out += pitch;
490 	}
491 #undef COMPENSATE
492 }
493 
ffIviColSlant4(const int32 * in,int16 * out,uint32 pitch,const uint8 * flags)494 void IndeoDSP::ffIviColSlant4(const int32 *in, int16 *out, uint32 pitch,
495 		const uint8 *flags) {
496 	int t0, t1, t2, t3, t4;
497 
498 	int row2 = pitch << 1;
499 
500 #define COMPENSATE(x) (((x) + 1)>>1)
501 	for (int i = 0; i < 4; i++) {
502 		if (flags[i]) {
503 			IVI_INV_SLANT4(in[0], in[4], in[8], in[12],
504 						   out[0], out[pitch], out[row2], out[row2 + pitch],
505 						   t0, t1, t2, t3, t4);
506 		} else {
507 			out[0] = out[pitch] = out[row2] = out[row2 + pitch] = 0;
508 		}
509 
510 		in++;
511 		out++;
512 	}
513 #undef COMPENSATE
514 }
515 
ffIviPutPixels8x8(const int32 * in,int16 * out,uint32 pitch,const uint8 * flags)516 void IndeoDSP::ffIviPutPixels8x8(const int32 *in, int16 *out, uint32 pitch,
517 		const uint8 *flags) {
518 	for (int y = 0; y < 8; out += pitch, in += 8, y++)
519 		for (int x = 0; x < 8; x++)
520 			out[x] = in[x];
521 }
522 
ffIviPutDcPixel8x8(const int32 * in,int16 * out,uint32 pitch,int blkSize)523 void IndeoDSP::ffIviPutDcPixel8x8(const int32 *in, int16 *out, uint32 pitch,
524 		int blkSize) {
525 	out[0] = in[0];
526 	memset(out + 1, 0, 7 * sizeof(out[0]));
527 	out += pitch;
528 
529 	for (int y = 1; y < 8; out += pitch, y++)
530 		memset(out, 0, 8 * sizeof(out[0]));
531 }
532 
533 #define IVI_MC_TEMPLATE(size, suffix, OP) \
534 static void iviMc ## size ##x## size ## suffix(int16 *buf, \
535 												 uint32 dpitch, \
536 												 const int16 *refBuf, \
537 												 uint32 pitch, int mcType) \
538 { \
539 	const int16 *wptr; \
540 \
541 	switch (mcType) { \
542 	case 0: /* fullpel (no interpolation) */ \
543 		for (int i = 0; i < size; i++, buf += dpitch, refBuf += pitch) { \
544 			for (int j = 0; j < size; j++) {\
545 				OP(buf[j], refBuf[j]); \
546 			} \
547 		} \
548 		break; \
549 	case 1: /* horizontal halfpel interpolation */ \
550 		for (int i = 0; i < size; i++, buf += dpitch, refBuf += pitch) \
551 			for (int j = 0; j < size; j++) \
552 				OP(buf[j], (refBuf[j] + refBuf[j+1]) >> 1); \
553 		break; \
554 	case 2: /* vertical halfpel interpolation */ \
555 		wptr = refBuf + pitch; \
556 		for (int i = 0; i < size; i++, buf += dpitch, wptr += pitch, refBuf += pitch) \
557 			for (int j = 0; j < size; j++) \
558 				OP(buf[j], (refBuf[j] + wptr[j]) >> 1); \
559 		break; \
560 	case 3: /* vertical and horizontal halfpel interpolation */ \
561 		wptr = refBuf + pitch; \
562 		for (int i = 0; i < size; i++, buf += dpitch, wptr += pitch, refBuf += pitch) \
563 			for (int j = 0; j < size; j++) \
564 				OP(buf[j], (refBuf[j] + refBuf[j+1] + wptr[j] + wptr[j+1]) >> 2); \
565 		break; \
566 	} \
567 } \
568 \
569 void IndeoDSP::ffIviMc ## size ##x## size ## suffix(int16 *buf, const int16 *refBuf, \
570 											 uint32 pitch, int mcType) \
571 { \
572 	iviMc ## size ##x## size ## suffix(buf, pitch, refBuf, pitch, mcType); \
573 }
574 
575 #define IVI_MC_AVG_TEMPLATE(size, suffix, OP) \
576 void IndeoDSP::ffIviMcAvg ## size ##x## size ## suffix(int16 *buf, \
577 												 const int16 *refBuf, \
578 												 const int16 *refBuf2, \
579 												 uint32 pitch, \
580 											   int mcType, int mcType2) \
581 { \
582 	int16 tmp[size * size]; \
583 \
584 	iviMc ## size ##x## size ## NoDelta(tmp, size, refBuf, pitch, mcType); \
585 	iviMc ## size ##x## size ## Delta(tmp, size, refBuf2, pitch, mcType2); \
586 	for (int i = 0; i < size; i++, buf += pitch) { \
587 		for (int j = 0; j < size; j++) {\
588 			OP(buf[j], tmp[i * size + j] >> 1); \
589 		} \
590 	} \
591 }
592 
593 #define OP_PUT(a, b)  (a) = (b)
594 #define OP_ADD(a, b)  (a) += (b)
595 
596 IVI_MC_TEMPLATE(8, NoDelta, OP_PUT)
597 IVI_MC_TEMPLATE(8, Delta,   OP_ADD)
598 IVI_MC_TEMPLATE(4, NoDelta, OP_PUT)
599 IVI_MC_TEMPLATE(4, Delta,   OP_ADD)
600 IVI_MC_AVG_TEMPLATE(8, NoDelta, OP_PUT)
601 IVI_MC_AVG_TEMPLATE(8, Delta,   OP_ADD)
602 IVI_MC_AVG_TEMPLATE(4, NoDelta, OP_PUT)
603 IVI_MC_AVG_TEMPLATE(4, Delta,   OP_ADD)
604 
605 } // End of namespace Indeo
606 } // End of namespace Image
607