39 int32_t p0, p1, p2, p3, tmp0, tmp1, tmp2;
40 int32_t b0_1, b0_2, b1_1, b1_2, b1_3, b2_1, b2_2, b2_3, b2_4, b2_5, b2_6;
41 int32_t b3_1, b3_2, b3_3, b3_4, b3_5, b3_6, b3_7, b3_8, b3_9;
43 const IDWTELEM *b0_ptr, *b1_ptr, *b2_ptr, *b3_ptr;
44 const int num_bands = 4;
58 for (y = 0; y < plane->
height; y += 2) {
66 b1_1 = b1_ptr[back_pitch];
68 b1_3 = b1_1 - b1_2*6 + b1_ptr[pitch];
79 b3_2 = b3_ptr[back_pitch];
83 b3_8 = b3_2 - b3_5*6 + b3_ptr[pitch];
87 for (x = 0, indx = 0; x < plane->
width; x+=2, indx++) {
101 p0 = p1 = p2 = p3 = 0;
107 b0_1 = b0_ptr[indx+1];
108 b0_2 = b0_ptr[pitch+indx+1];
113 p2 = (tmp0 + tmp2) << 3;
114 p3 = (tmp1 + tmp2 + b0_2) << 2;
121 b1_2 = b1_ptr[indx+1];
122 b1_1 = b1_ptr[back_pitch+indx+1];
124 tmp2 = tmp1 - tmp0*6 + b1_3;
125 b1_3 = b1_1 - b1_2*6 + b1_ptr[pitch+indx+1];
127 p0 += (tmp0 + tmp1) << 3;
128 p1 += (tmp0 + tmp1 + b1_1 + b1_2) << 2;
130 p3 += (tmp2 + b1_3) << 1;
135 b2_3 = b2_ptr[indx+1];
136 b2_6 = b2_ptr[pitch+indx+1];
139 tmp1 = b2_1 - b2_2*6 + b2_3;
143 p2 += (tmp0 + b2_4 + b2_5) << 2;
144 p3 += (tmp1 + b2_4 - b2_5*6 + b2_6) << 1;
149 b3_6 = b3_ptr[indx+1];
150 b3_3 = b3_ptr[back_pitch+indx+1];
156 b3_9 = b3_3 - b3_6*6 + b3_ptr[pitch+indx+1];
158 p0 += (tmp0 + tmp1) << 2;
159 p1 += (tmp0 - tmp1*6 + tmp2) << 1;
160 p2 += (b3_7 + b3_8) << 1;
161 p3 += b3_7 - b3_8*6 + b3_9;
165 dst[x] = av_clip_uint8((p0 >> 6) + 128);
166 dst[x+1] = av_clip_uint8((p1 >> 6) + 128);
167 dst[dst_pitch+x] = av_clip_uint8((p2 >> 6) + 128);
168 dst[dst_pitch+x+1] = av_clip_uint8((p3 >> 6) + 128);
171 dst += dst_pitch << 1;
185 int x, y, indx, b0, b1, b2, b3, p0, p1, p2, p3;
186 const IDWTELEM *b0_ptr, *b1_ptr, *b2_ptr, *b3_ptr;
198 for (y = 0; y < plane->
height; y += 2) {
199 for (x = 0, indx = 0; x < plane->
width; x += 2, indx++) {
207 p0 = (b0 + b1 + b2 + b3 + 2) >> 2;
208 p1 = (b0 + b1 - b2 - b3 + 2) >> 2;
209 p2 = (b0 - b1 + b2 - b3 + 2) >> 2;
210 p3 = (b0 - b1 - b2 + b3 + 2) >> 2;
213 dst[x] = av_clip_uint8(p0 + 128);
214 dst[x + 1] = av_clip_uint8(p1 + 128);
215 dst[dst_pitch + x] = av_clip_uint8(p2 + 128);
216 dst[dst_pitch + x + 1] = av_clip_uint8(p3 + 128);
219 dst += dst_pitch << 1;
229 #define IVI_HAAR_BFLY(s1, s2, o1, o2, t) \
231 o1 = (s1 + s2) >> 1;\
235 #define INV_HAAR8(s1, s5, s3, s7, s2, s4, s6, s8,\
236 d1, d2, d3, d4, d5, d6, d7, d8,\
237 t0, t1, t2, t3, t4, t5, t6, t7, t8) {\
238 t1 = s1 << 1; t5 = s5 << 1;\
239 IVI_HAAR_BFLY(t1, t5, t1, t5, t0); IVI_HAAR_BFLY(t1, s3, t1, t3, t0);\
240 IVI_HAAR_BFLY(t5, s7, t5, t7, t0); IVI_HAAR_BFLY(t1, s2, t1, t2, t0);\
241 IVI_HAAR_BFLY(t3, s4, t3, t4, t0); IVI_HAAR_BFLY(t5, s6, t5, t6, t0);\
242 IVI_HAAR_BFLY(t7, s8, t7, t8, t0);\
243 d1 = COMPENSATE(t1);\
244 d2 = COMPENSATE(t2);\
245 d3 = COMPENSATE(t3);\
246 d4 = COMPENSATE(t4);\
247 d5 = COMPENSATE(t5);\
248 d6 = COMPENSATE(t6);\
249 d7 = COMPENSATE(t7);\
250 d8 = COMPENSATE(t8); }
253 #define INV_HAAR4(s1, s3, s5, s7) {\
254 HAAR_BFLY(s1, s5); HAAR_BFLY(s1, s3); HAAR_BFLY(s5, s7);\
255 s1 = COMPENSATE(s1);\
256 s3 = COMPENSATE(s3);\
257 s5 = COMPENSATE(s5);\
258 s7 = COMPENSATE(s7); }
263 int i, shift, sp1, sp2, sp3, sp4;
270 #define COMPENSATE(x) (x)
273 for (i = 0; i < 8; i++) {
277 sp1 = src[ 0] << shift;
278 sp2 = src[ 8] << shift;
279 sp3 = src[16] << shift;
280 sp4 = src[24] << shift;
282 src[32], src[40], src[48], src[56],
283 dst[ 0], dst[ 8], dst[16], dst[24],
284 dst[32], dst[40], dst[48], dst[56],
285 t0, t1, t2, t3, t4, t5, t6, t7, t8);
287 dst[ 0] = dst[ 8] = dst[16] = dst[24] =
288 dst[32] = dst[40] = dst[48] = dst[56] = 0;
296 #define COMPENSATE(x) (x)
298 for (i = 0; i < 8; i++) {
299 if ( !src[0] && !src[1] && !src[2] && !src[3]
300 && !src[4] && !src[5] && !src[6] && !src[7]) {
301 memset(out, 0, 8 *
sizeof(out[0]));
303 INV_HAAR8(src[0], src[1], src[2], src[3],
304 src[4], src[5], src[6], src[7],
305 out[0], out[1], out[2], out[3],
306 out[4], out[5], out[6], out[7],
307 t0, t1, t2, t3, t4, t5, t6, t7, t8);
321 dc_coeff = (*in + 0) >> 3;
323 for (y = 0; y < blk_size; out += pitch, y++) {
324 for (x = 0; x < blk_size; x++)
330 #define IVI_SLANT_BFLY(s1, s2, o1, o2, t) \
336 #define IVI_IREFLECT(s1, s2, o1, o2, t) \
337 t = ((s1 + s2*2 + 2) >> 2) + s1;\
338 o2 = ((s1*2 - s2 + 2) >> 2) - s2;\
342 #define IVI_SLANT_PART4(s1, s2, o1, o2, t) \
343 t = s2 + ((s1*4 - s2 + 4) >> 3);\
344 o2 = s1 + ((-s1 - s2*4 + 4) >> 3);\
348 #define IVI_INV_SLANT8(s1, s4, s8, s5, s2, s6, s3, s7,\
349 d1, d2, d3, d4, d5, d6, d7, d8,\
350 t0, t1, t2, t3, t4, t5, t6, t7, t8) {\
351 IVI_SLANT_PART4(s4, s5, t4, t5, t0);\
353 IVI_SLANT_BFLY(s1, t5, t1, t5, t0); IVI_SLANT_BFLY(s2, s6, t2, t6, t0);\
354 IVI_SLANT_BFLY(s7, s3, t7, t3, t0); IVI_SLANT_BFLY(t4, s8, t4, t8, t0);\
356 IVI_SLANT_BFLY(t1, t2, t1, t2, t0); IVI_IREFLECT (t4, t3, t4, t3, t0);\
357 IVI_SLANT_BFLY(t5, t6, t5, t6, t0); IVI_IREFLECT (t8, t7, t8, t7, t0);\
358 IVI_SLANT_BFLY(t1, t4, t1, t4, t0); IVI_SLANT_BFLY(t2, t3, t2, t3, t0);\
359 IVI_SLANT_BFLY(t5, t8, t5, t8, t0); IVI_SLANT_BFLY(t6, t7, t6, t7, t0);\
360 d1 = COMPENSATE(t1);\
361 d2 = COMPENSATE(t2);\
362 d3 = COMPENSATE(t3);\
363 d4 = COMPENSATE(t4);\
364 d5 = COMPENSATE(t5);\
365 d6 = COMPENSATE(t6);\
366 d7 = COMPENSATE(t7);\
367 d8 = COMPENSATE(t8);}
370 #define IVI_INV_SLANT4(s1, s4, s2, s3, d1, d2, d3, d4, t0, t1, t2, t3, t4) {\
371 IVI_SLANT_BFLY(s1, s2, t1, t2, t0); IVI_IREFLECT (s4, s3, t4, t3, t0);\
373 IVI_SLANT_BFLY(t1, t4, t1, t4, t0); IVI_SLANT_BFLY(t2, t3, t2, t3, t0);\
374 d1 = COMPENSATE(t1);\
375 d2 = COMPENSATE(t2);\
376 d3 = COMPENSATE(t3);\
377 d4 = COMPENSATE(t4);}
387 #define COMPENSATE(x) (x)
390 for (i = 0; i < 8; i++) {
392 IVI_INV_SLANT8(src[0], src[8], src[16], src[24], src[32], src[40], src[48], src[56],
393 dst[0], dst[8], dst[16], dst[24], dst[32], dst[40], dst[48], dst[56],
394 t0, t1, t2, t3, t4, t5, t6, t7, t8);
396 dst[0] = dst[8] = dst[16] = dst[24] = dst[32] = dst[40] = dst[48] = dst[56] = 0;
403 #define COMPENSATE(x) ((x + 1)>>1)
405 for (i = 0; i < 8; i++) {
406 if (!src[0] && !src[1] && !src[2] && !src[3] && !src[4] && !src[5] && !src[6] && !src[7]) {
407 memset(out, 0, 8*
sizeof(out[0]));
409 IVI_INV_SLANT8(src[0], src[1], src[2], src[3], src[4], src[5], src[6], src[7],
410 out[0], out[1], out[2], out[3], out[4], out[5], out[6], out[7],
411 t0, t1, t2, t3, t4, t5, t6, t7, t8);
427 #define COMPENSATE(x) (x)
430 for (i = 0; i < 4; i++) {
433 dst[0], dst[4], dst[8], dst[12],
436 dst[0] = dst[4] = dst[8] = dst[12] = 0;
443 #define COMPENSATE(x) ((x + 1)>>1)
445 for (i = 0; i < 4; i++) {
446 if (!src[0] && !src[1] && !src[2] && !src[3]) {
447 out[0] = out[1] = out[2] = out[3] = 0;
450 out[0], out[1], out[2], out[3],
464 dc_coeff = (*in + 1) >> 1;
466 for (y = 0; y < blk_size; out += pitch, y++) {
467 for (x = 0; x < blk_size; x++)
477 #define COMPENSATE(x) ((x + 1)>>1)
478 for (i = 0; i < 8; i++) {
479 if (!in[0] && !in[1] && !in[2] && !in[3] && !in[4] && !in[5] && !in[6] && !in[7]) {
480 memset(out, 0, 8*
sizeof(out[0]));
482 IVI_INV_SLANT8( in[0], in[1], in[2], in[3], in[4], in[5], in[6], in[7],
483 out[0], out[1], out[2], out[3], out[4], out[5], out[6], out[7],
484 t0, t1, t2, t3, t4, t5, t6, t7, t8);
497 dc_coeff = (*in + 1) >> 1;
499 for (x = 0; x < blk_size; x++)
504 for (y = 1; y < blk_size; out += pitch, y++) {
505 for (x = 0; x < blk_size; x++)
512 int i, row2, row4, row8;
519 #define COMPENSATE(x) ((x + 1)>>1)
520 for (i = 0; i < 8; i++) {
522 IVI_INV_SLANT8(in[0], in[8], in[16], in[24], in[32], in[40], in[48], in[56],
523 out[0], out[pitch], out[row2], out[row2 + pitch], out[row4],
524 out[row4 + pitch], out[row4 + row2], out[row8 - pitch],
525 t0, t1, t2, t3, t4, t5, t6, t7, t8);
527 out[0] = out[pitch] = out[row2] = out[row2 + pitch] = out[row4] =
528 out[row4 + pitch] = out[row4 + row2] = out[row8 - pitch] = 0;
542 dc_coeff = (*in + 1) >> 1;
544 for (y = 0; y < blk_size; out += pitch, y++) {
546 for (x = 1; x < blk_size; x++)
556 for (y = 0; y < 8; out += pitch, in += 8, y++)
557 for (x = 0; x < 8; x++)
567 memset(out + 1, 0, 7*
sizeof(out[0]));
570 for (y = 1; y < 8; out += pitch, y++)
571 memset(out, 0, 8*
sizeof(out[0]));
574 #define IVI_MC_TEMPLATE(size, suffix, OP) \
575 void ff_ivi_mc_ ## size ##x## size ## suffix (int16_t *buf, const int16_t *ref_buf, \
576 uint32_t pitch, int mc_type) \
579 const int16_t *wptr; \
583 for (i = 0; i < size; i++, buf += pitch, ref_buf += pitch) { \
584 for (j = 0; j < size; j++) {\
585 OP(buf[j], ref_buf[j]); \
590 for (i = 0; i < size; i++, buf += pitch, ref_buf += pitch) \
591 for (j = 0; j < size; j++) \
592 OP(buf[j], (ref_buf[j] + ref_buf[j+1]) >> 1); \
595 wptr = ref_buf + pitch; \
596 for (i = 0; i < size; i++, buf += pitch, wptr += pitch, ref_buf += pitch) \
597 for (j = 0; j < size; j++) \
598 OP(buf[j], (ref_buf[j] + wptr[j]) >> 1); \
601 wptr = ref_buf + pitch; \
602 for (i = 0; i < size; i++, buf += pitch, wptr += pitch, ref_buf += pitch) \
603 for (j = 0; j < size; j++) \
604 OP(buf[j], (ref_buf[j] + ref_buf[j+1] + wptr[j] + wptr[j+1]) >> 2); \
609 #define OP_PUT(a, b) (a) = (b)
610 #define OP_ADD(a, b) (a) += (b)
614 IVI_MC_TEMPLATE(4, _no_delta,
OP_PUT)
615 IVI_MC_TEMPLATE(4, _delta,
OP_ADD)