/src/libavc/decoder/svc/isvcd_residual_resamp.c
Line | Count | Source |
1 | | /****************************************************************************** |
2 | | * |
3 | | * Copyright (C) 2022 The Android Open Source Project |
4 | | * |
5 | | * Licensed under the Apache License, Version 2.0 (the "License"); |
6 | | * you may not use this file except in compliance with the License. |
7 | | * You may obtain a copy of the License at: |
8 | | * |
9 | | * http://www.apache.org/licenses/LICENSE-2.0 |
10 | | * |
11 | | * Unless required by applicable law or agreed to in writing, software |
12 | | * distributed under the License is distributed on an "AS IS" BASIS, |
13 | | * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
14 | | * See the License for the specific language governing permissions and |
15 | | * limitations under the License. |
16 | | * |
17 | | ***************************************************************************** |
18 | | * Originally developed and contributed by Ittiam Systems Pvt. Ltd, Bangalore |
19 | | */ |
20 | | /** |
21 | | ******************************************************************************* |
22 | | * @file |
23 | | * isvcd_residual_resamp.c |
24 | | * |
25 | | * @brief |
26 | | * Contains routines that resample for SVC resampling |
27 | | * |
28 | | * @author |
29 | | * Kishore |
30 | | * |
31 | | * @par List of Functions: |
32 | | * - isvcd_residual_chroma_dyadic_alt() |
33 | | * - isvcd_residual_chroma_dyadic() |
34 | | * - isvcd_residual_luma_dyadic() |
35 | | * - isvcd_ref_layer_ptr_incr() |
36 | | * - isvcd_residual_reflayer_const_non_boundary_mb() |
37 | | * - isvcd_residual_reflayer_const_boundary_mb() |
38 | | * - isvcd_residual_reflayer_const() |
39 | | * - isvcd_interpolate_residual() |
40 | | * - isvcd_residual_samp_mb() |
41 | | * - isvcd_residual_samp_mb_dyadic() |
42 | | * - isvcd_residual_samp_populate_list() |
43 | | * - isvcd_residual_samp_res_init() |
44 | | * |
45 | | * @remarks |
46 | | * None |
47 | | * |
48 | | ******************************************************************************* |
49 | | */ |
50 | | |
51 | | #include <assert.h> |
52 | | #include <string.h> |
53 | | #include "ih264_typedefs.h" |
54 | | #include "ih264_macros.h" |
55 | | #include "ih264_platform_macros.h" |
56 | | #include "ih264d_bitstrm.h" |
57 | | #include "ih264d_defs.h" |
58 | | #include "ih264d_debug.h" |
59 | | #include "isvcd_structs.h" |
60 | | #include "ih264d_defs.h" |
61 | | #include "ih264d_parse_cavlc.h" |
62 | | #include "ih264d_mb_utils.h" |
63 | | #include "ih264d_deblocking.h" |
64 | | #include "ih264d_dpb_manager.h" |
65 | | #include "ih264d_mvpred.h" |
66 | | #include "ih264d_inter_pred.h" |
67 | | #include "ih264d_process_pslice.h" |
68 | | #include "ih264d_error_handler.h" |
69 | | #include "ih264d_cabac.h" |
70 | | #include "ih264d_debug.h" |
71 | | #include "ih264d_tables.h" |
72 | | #include "ih264d_parse_slice.h" |
73 | | #include "ih264d_utils.h" |
74 | | #include "ih264d_parse_islice.h" |
75 | | #include "ih264d_process_bslice.h" |
76 | | #include "ih264d_process_intra_mb.h" |
77 | | #include "ih264_debug.h" |
78 | | |
79 | | /*****************************************************************************/ |
80 | | /* */ |
81 | | /* Function Name : isvcd_residual_chroma_dyadic_alt */ |
82 | | /* */ |
83 | | /* Description : this fucntion does the upsampling of chroma residuals for*/ |
84 | | /* Dyadic cases and specific chroma phase cases */ |
85 | | /* */ |
86 | | /* Inputs : pv_residual_samp_ctxt : Residual upsampling context */ |
87 | | /* pu1_inp_data : input 8 bit data pointer */ |
88 | | /* i4_inp_data_stride : input buffer stride */ |
89 | | /* pi2_out_res : output 16 bit buffer pointer */ |
90 | | /* i4_out_res_stride : Output buffer stride */ |
91 | | /* pu1_inp_bitmap : input packed sign bit data pointer */ |
92 | | /* i4_inp_bitmap_stride : sign bit buffer stride */ |
93 | | /* i4_start_bit_pos : bit position in the byte of packed */ |
94 | | /* sign values */ |
95 | | /* Globals : none */ |
96 | | /* Processing : it does the upsampling with intial phase values */ |
97 | | /* */ |
98 | | /* Outputs : Upsampled residuals for chroma */ |
99 | | /* Returns : none */ |
100 | | /* */ |
101 | | /* Issues : none */ |
102 | | /* */ |
103 | | /* Revision History: */ |
104 | | /* */ |
105 | | /* DD MM YYYY Author(s) Changes (Describe the changes made) */ |
106 | | /* 25 09 2021 vijayakumar creation */ |
107 | | /* */ |
108 | | /*****************************************************************************/ |
109 | | void isvcd_residual_chroma_dyadic_alt(void *pv_residual_samp_ctxt, UWORD16 u2_mb_x, UWORD16 u2_mb_y, |
110 | | mem_element_t *ps_ref_mb_mode, WORD16 *pi2_inp_data, |
111 | | WORD32 i4_inp_data_stride, WORD16 *pi2_out_res, |
112 | | WORD32 i4_out_res_stride, WORD32 i4_cr_flag) |
113 | 0 | { |
114 | 0 | residual_sampling_ctxt_t *ps_ctxt; |
115 | 0 | res_lyr_ctxt *ps_lyr_ctxt; |
116 | 0 | ps_ctxt = (residual_sampling_ctxt_t *) pv_residual_samp_ctxt; |
117 | 0 | ps_lyr_ctxt = &ps_ctxt->as_res_lyrs[ps_ctxt->i4_res_lyr_id]; |
118 | | /* ----------------- Processing ------------------------------- */ |
119 | 0 | { |
120 | 0 | ref_pixel_map_t *ps_pos_phase; |
121 | 0 | residual_samp_map_ctxt_t *ps_chroma_map; |
122 | 0 | ref_mb_map_t *ps_x_off_len_chroma; |
123 | 0 | ref_mb_map_t *ps_y_off_len_chroma; |
124 | 0 | WORD32 i4_i; |
125 | 0 | WORD16 *pi2_ref_data_byte; |
126 | 0 | WORD32 *pi4_ref_array; |
127 | 0 | WORD32 i4_phase1, i4_phase2; |
128 | 0 | WORD32 i4_offset_x, i4_offset_y; |
129 | 0 | WORD32 i4_chrm_horz_int_mode, i4_chrm_vert_int_mode; |
130 | 0 | WORD32 i4_horz_intp_ctr = SUB_BLOCK_HEIGHT; |
131 | |
|
132 | 0 | ps_chroma_map = &ps_lyr_ctxt->s_chroma_map_ctxt; |
133 | 0 | ps_x_off_len_chroma = ps_chroma_map->ps_x_offset_length; |
134 | 0 | ps_y_off_len_chroma = ps_chroma_map->ps_y_offset_length; |
135 | | |
136 | | /* get the actual offset for the buffers */ |
137 | 0 | i4_offset_x = ps_x_off_len_chroma[u2_mb_x].i2_offset; |
138 | 0 | i4_offset_y = ps_y_off_len_chroma[u2_mb_y].i2_offset; |
139 | |
|
140 | 0 | { |
141 | 0 | UWORD8 u1_mask; |
142 | 0 | WORD32 i4_mb_x, i4_mb_y; |
143 | 0 | WORD32 i4_chrm_nnz; |
144 | 0 | WORD32 i4_num_element_stride; |
145 | 0 | inter_lyr_mb_prms_t *ps_inter_lyr_mb_prms, *ps_inter_lyr_mb_prms_curr; |
146 | |
|
147 | 0 | u1_mask = (SVCD_TRUE == i4_cr_flag) ? 0xF0 : 0x0F; |
148 | | |
149 | | /* Top Left */ |
150 | 0 | i4_mb_x = i4_offset_x >> 3; |
151 | 0 | i4_mb_y = i4_offset_y >> 3; |
152 | | |
153 | | /* get the location of the byte which has the current mb mode */ |
154 | 0 | ps_inter_lyr_mb_prms = ps_ref_mb_mode->pv_buffer; |
155 | 0 | i4_num_element_stride = ps_ref_mb_mode->i4_num_element_stride; |
156 | |
|
157 | 0 | ps_inter_lyr_mb_prms_curr = ps_inter_lyr_mb_prms + i4_mb_x; |
158 | 0 | ps_inter_lyr_mb_prms_curr += i4_mb_y * i4_num_element_stride; |
159 | 0 | i4_chrm_nnz = ps_inter_lyr_mb_prms_curr->u1_chroma_nnz & u1_mask; |
160 | | |
161 | | /* Top Right */ |
162 | 0 | i4_mb_x = (i4_offset_x + 4) >> 3; |
163 | 0 | ps_inter_lyr_mb_prms_curr = ps_inter_lyr_mb_prms + i4_mb_x; |
164 | 0 | ps_inter_lyr_mb_prms_curr += i4_mb_y * i4_num_element_stride; |
165 | 0 | i4_chrm_nnz |= ps_inter_lyr_mb_prms_curr->u1_chroma_nnz & u1_mask; |
166 | | |
167 | | /* Bottom Left */ |
168 | 0 | i4_mb_x = i4_offset_x >> 3; |
169 | 0 | i4_mb_y = (i4_offset_y + 4) >> 3; |
170 | 0 | ps_inter_lyr_mb_prms_curr = ps_inter_lyr_mb_prms + i4_mb_x; |
171 | 0 | ps_inter_lyr_mb_prms_curr += i4_mb_y * i4_num_element_stride; |
172 | 0 | i4_chrm_nnz |= ps_inter_lyr_mb_prms_curr->u1_chroma_nnz & u1_mask; |
173 | | |
174 | | /* Bottom Right */ |
175 | 0 | i4_mb_x = (i4_offset_x + 4) >> 3; |
176 | 0 | ps_inter_lyr_mb_prms_curr = ps_inter_lyr_mb_prms + i4_mb_x; |
177 | 0 | ps_inter_lyr_mb_prms_curr += i4_mb_y * i4_num_element_stride; |
178 | |
|
179 | 0 | i4_chrm_nnz |= ps_inter_lyr_mb_prms_curr->u1_chroma_nnz & u1_mask; |
180 | 0 | if(0 == i4_chrm_nnz) |
181 | 0 | { |
182 | 0 | return; |
183 | 0 | } |
184 | 0 | } |
185 | | |
186 | 0 | i4_chrm_horz_int_mode = ps_lyr_ctxt->i4_chrm_horz_int_mode; |
187 | 0 | i4_chrm_vert_int_mode = ps_lyr_ctxt->i4_chrm_vert_int_mode; |
188 | |
|
189 | 0 | if(0 == i4_chrm_horz_int_mode) |
190 | 0 | { |
191 | 0 | if(i4_offset_x >= 0) |
192 | 0 | { |
193 | 0 | pi2_inp_data++; |
194 | 0 | } |
195 | 0 | } |
196 | |
|
197 | 0 | if(0 == i4_chrm_vert_int_mode) |
198 | 0 | { |
199 | 0 | if(i4_offset_y >= 0) |
200 | 0 | { |
201 | 0 | pi2_inp_data += i4_inp_data_stride; |
202 | 0 | } |
203 | 0 | } |
204 | 0 | else |
205 | 0 | { |
206 | | /* extra additional row of interpolation required for this case */ |
207 | 0 | i4_horz_intp_ctr++; |
208 | 0 | } |
209 | | |
210 | | /* ----------- Horizontal Interpolation ---------------- */ |
211 | 0 | pi2_ref_data_byte = pi2_inp_data; |
212 | 0 | ps_pos_phase = ps_lyr_ctxt->s_chroma_map_ctxt.ps_x_pos_phase; |
213 | |
|
214 | 0 | pi4_ref_array = (WORD32 *) ps_ctxt->pi2_refarray_buffer; |
215 | 0 | i4_phase1 = ps_pos_phase[0].i2_phase; |
216 | 0 | i4_phase2 = (i4_phase1 + 8) & 0x0F; |
217 | | |
218 | | /* interchange the phase values for corner case */ |
219 | 0 | if(1 == i4_chrm_horz_int_mode) |
220 | 0 | { |
221 | 0 | WORD32 i4_temp; |
222 | 0 | i4_temp = i4_phase1; |
223 | 0 | i4_phase1 = i4_phase2; |
224 | 0 | i4_phase2 = i4_temp; |
225 | 0 | } |
226 | |
|
227 | 0 | for(i4_i = 0; i4_i < i4_horz_intp_ctr; i4_i++) |
228 | 0 | { |
229 | 0 | WORD16 i2_coeff1, i2_coeff2; |
230 | |
|
231 | 0 | i2_coeff1 = (WORD16) (pi2_ref_data_byte[0]); |
232 | |
|
233 | 0 | if(0 == i4_chrm_horz_int_mode) |
234 | 0 | { |
235 | | /* populate the first inter sample */ |
236 | 0 | *pi4_ref_array++ = i2_coeff1 << 4; |
237 | 0 | } |
238 | | |
239 | | /* unroll count 1 */ |
240 | 0 | i2_coeff2 = (WORD16) (pi2_ref_data_byte[2]); |
241 | | |
242 | | /* populate 2 samples based on current coeffs */ |
243 | 0 | *pi4_ref_array++ = ((16 - i4_phase2) * i2_coeff1 + i4_phase2 * i2_coeff2); |
244 | 0 | *pi4_ref_array++ = ((16 - i4_phase1) * i2_coeff1 + i4_phase1 * i2_coeff2); |
245 | | |
246 | | /* unroll count 2 */ |
247 | 0 | i2_coeff1 = (WORD16) (pi2_ref_data_byte[4]); |
248 | | |
249 | | /* populate 2 samples based on current coeffs */ |
250 | 0 | *pi4_ref_array++ = ((16 - i4_phase2) * i2_coeff2 + i4_phase2 * i2_coeff1); |
251 | 0 | *pi4_ref_array++ = ((16 - i4_phase1) * i2_coeff2 + i4_phase1 * i2_coeff1); |
252 | | |
253 | | /* unroll count 3 */ |
254 | 0 | i2_coeff2 = (WORD16) (pi2_ref_data_byte[6]); |
255 | | |
256 | | /* populate 2 samples based on current coeffs */ |
257 | 0 | *pi4_ref_array++ = ((16 - i4_phase2) * i2_coeff1 + i4_phase2 * i2_coeff2); |
258 | 0 | *pi4_ref_array++ = ((16 - i4_phase1) * i2_coeff1 + i4_phase1 * i2_coeff2); |
259 | | |
260 | | /* populate the last inter sample */ |
261 | 0 | *pi4_ref_array++ = i2_coeff2 << 4; |
262 | |
|
263 | 0 | if(1 == i4_chrm_horz_int_mode) |
264 | 0 | { |
265 | 0 | i2_coeff1 = (WORD16) (pi2_ref_data_byte[4]); |
266 | | |
267 | | /* populate the last inter sample */ |
268 | 0 | *pi4_ref_array++ = i2_coeff1 << 4; |
269 | 0 | } |
270 | | |
271 | | /* vertical loop updates */ |
272 | 0 | pi2_ref_data_byte = pi2_inp_data + ((i4_i + 1) * i4_inp_data_stride); |
273 | 0 | } |
274 | | |
275 | | /* ----------- Vertical Interpolation ---------------- */ |
276 | 0 | pi4_ref_array = (WORD32 *) ps_ctxt->pi2_refarray_buffer; |
277 | 0 | ps_pos_phase = ps_lyr_ctxt->s_chroma_map_ctxt.ps_y_pos_phase; |
278 | 0 | i4_phase1 = ps_pos_phase[0].i2_phase; |
279 | 0 | i4_phase2 = (i4_phase1 + 8) & 0x0F; |
280 | | |
281 | | /* interchange the phase values for corner case */ |
282 | 0 | if(0 != i4_chrm_vert_int_mode) |
283 | 0 | { |
284 | 0 | WORD32 i4_temp; |
285 | 0 | i4_temp = i4_phase1; |
286 | 0 | i4_phase1 = i4_phase2; |
287 | 0 | i4_phase2 = i4_temp; |
288 | 0 | } |
289 | |
|
290 | 0 | for(i4_i = 0; i4_i < BLOCK_WIDTH; i4_i++) |
291 | 0 | { |
292 | 0 | WORD16 *pi2_out; |
293 | 0 | WORD32 *pi4_ref_array_temp; |
294 | 0 | WORD32 i4_horz_samp_1, i4_horz_samp_2; |
295 | 0 | pi2_out = pi2_out_res; |
296 | 0 | pi4_ref_array_temp = pi4_ref_array; |
297 | | |
298 | | /* populate the first inter sample */ |
299 | 0 | i4_horz_samp_1 = *pi4_ref_array_temp; |
300 | 0 | pi4_ref_array_temp += BLOCK_WIDTH; |
301 | |
|
302 | 0 | if(1 != i4_chrm_vert_int_mode) |
303 | 0 | { |
304 | 0 | *pi2_out = (i4_horz_samp_1 + 8) >> 4; |
305 | 0 | pi2_out += i4_out_res_stride; |
306 | 0 | } |
307 | |
|
308 | 0 | if(2 == i4_chrm_vert_int_mode) |
309 | 0 | { |
310 | 0 | i4_horz_samp_1 = *pi4_ref_array_temp; |
311 | 0 | pi4_ref_array_temp += BLOCK_WIDTH; |
312 | 0 | *pi2_out = (i4_horz_samp_1 + 8) >> 4; |
313 | 0 | pi2_out += i4_out_res_stride; |
314 | 0 | } |
315 | | |
316 | | /* unroll count 1 */ |
317 | 0 | i4_horz_samp_2 = *pi4_ref_array_temp; |
318 | 0 | pi4_ref_array_temp += BLOCK_WIDTH; |
319 | | |
320 | | /* populate 2 samples based on current coeffs */ |
321 | 0 | *pi2_out = ((16 - i4_phase2) * i4_horz_samp_1 + i4_phase2 * i4_horz_samp_2 + 128) >> 8; |
322 | 0 | pi2_out += i4_out_res_stride; |
323 | | |
324 | | /* unroll count 2 */ |
325 | 0 | *pi2_out = ((16 - i4_phase1) * i4_horz_samp_1 + i4_phase1 * i4_horz_samp_2 + 128) >> 8; |
326 | 0 | pi2_out += i4_out_res_stride; |
327 | | |
328 | | /* unroll count 3 */ |
329 | 0 | i4_horz_samp_1 = *pi4_ref_array_temp; |
330 | 0 | pi4_ref_array_temp += BLOCK_WIDTH; |
331 | | |
332 | | /* populate 2 samples based on current coeffs */ |
333 | 0 | *pi2_out = ((16 - i4_phase2) * i4_horz_samp_2 + i4_phase2 * i4_horz_samp_1 + 128) >> 8; |
334 | 0 | pi2_out += i4_out_res_stride; |
335 | | |
336 | | /* unroll count 4 */ |
337 | 0 | *pi2_out = ((16 - i4_phase1) * i4_horz_samp_2 + i4_phase1 * i4_horz_samp_1 + 128) >> 8; |
338 | 0 | pi2_out += i4_out_res_stride; |
339 | | |
340 | | /* unroll count 5 */ |
341 | 0 | i4_horz_samp_2 = *pi4_ref_array_temp; |
342 | | |
343 | | /* populate 2 samples based on current coeffs */ |
344 | 0 | *pi2_out = ((16 - i4_phase2) * i4_horz_samp_1 + i4_phase2 * i4_horz_samp_2 + 128) >> 8; |
345 | 0 | pi2_out += i4_out_res_stride; |
346 | | |
347 | | /* unroll count 6 */ |
348 | 0 | *pi2_out = ((16 - i4_phase1) * i4_horz_samp_1 + i4_phase1 * i4_horz_samp_2 + 128) >> 8; |
349 | 0 | pi2_out += i4_out_res_stride; |
350 | |
|
351 | 0 | if(2 != i4_chrm_vert_int_mode) |
352 | 0 | { |
353 | | /* populate the last inter sample */ |
354 | 0 | *pi2_out = (i4_horz_samp_2 + 8) >> 4; |
355 | |
|
356 | 0 | if(1 == i4_chrm_vert_int_mode) |
357 | 0 | { |
358 | 0 | pi2_out += i4_out_res_stride; |
359 | 0 | pi4_ref_array_temp += BLOCK_WIDTH; |
360 | 0 | i4_horz_samp_1 = *pi4_ref_array_temp; |
361 | | |
362 | | /* populate the last inter sample */ |
363 | 0 | *pi2_out = (i4_horz_samp_1 + 8) >> 4; |
364 | 0 | } |
365 | 0 | } |
366 | | /* horizontal loop updates */ |
367 | 0 | pi4_ref_array++; |
368 | 0 | pi2_out_res += 2; |
369 | 0 | } |
370 | 0 | } |
371 | 0 | return; |
372 | 0 | } |
373 | | |
374 | | /*****************************************************************************/ |
375 | | /* */ |
376 | | /* Function Name : isvcd_residual_chroma_dyadic */ |
377 | | /* */ |
378 | | /* Description : this fucntion does the upsampling of chroma residuals for*/ |
379 | | /* Dyadic cases */ |
380 | | /* */ |
381 | | /* Inputs : pv_residual_samp_ctxt : Residual upsampling context */ |
382 | | /* pu1_inp_data : input 8 bit data pointer */ |
383 | | /* i4_inp_data_stride : input buffer stride */ |
384 | | /* pi2_out_res : output 16 bit buffer pointer */ |
385 | | /* i4_out_res_stride : Output buffer stride */ |
386 | | /* pu1_inp_bitmap : input packed sign bit data pointer */ |
387 | | /* i4_inp_bitmap_stride : sign bit buffer stride */ |
388 | | /* i4_start_bit_pos : bit position in the byte of packed */ |
389 | | /* sign values */ |
390 | | /* Globals : none */ |
391 | | /* Processing : it does the upsampling with intial phase values */ |
392 | | /* */ |
393 | | /* Outputs : Upsampled residuals for chroma */ |
394 | | /* Returns : none */ |
395 | | /* */ |
396 | | /* Issues : none */ |
397 | | /* */ |
398 | | /* Revision History: */ |
399 | | /* */ |
400 | | /* DD MM YYYY Author(s) Changes (Describe the changes made) */ |
401 | | /* 25 09 2021 vijayakumar creation */ |
402 | | /* */ |
403 | | /*****************************************************************************/ |
404 | | void isvcd_residual_chroma_dyadic(void *pv_residual_samp_ctxt, WORD16 *pi2_inp_data, |
405 | | WORD32 i4_inp_data_stride, WORD16 *pi2_out_res, |
406 | | WORD32 i4_out_res_stride) |
407 | 2.01k | { |
408 | 2.01k | residual_sampling_ctxt_t *ps_ctxt; |
409 | 2.01k | res_lyr_ctxt *ps_lyr_ctxt; |
410 | 2.01k | ps_ctxt = (residual_sampling_ctxt_t *) pv_residual_samp_ctxt; |
411 | 2.01k | ps_lyr_ctxt = &ps_ctxt->as_res_lyrs[ps_ctxt->i4_res_lyr_id]; |
412 | | /* ----------------- Processing ------------------------------- */ |
413 | 2.01k | { |
414 | 2.01k | WORD32 i4_i; |
415 | 2.01k | WORD16 *pi2_ref_data_byte; |
416 | 2.01k | WORD32 *pi4_ref_array; |
417 | 2.01k | ref_pixel_map_t *ps_pos_phase; |
418 | 2.01k | WORD32 i4_phase1, i4_phase2; |
419 | | |
420 | 2.01k | pi2_ref_data_byte = pi2_inp_data; |
421 | 2.01k | ps_pos_phase = ps_lyr_ctxt->s_chroma_map_ctxt.ps_x_pos_phase; |
422 | | |
423 | | /* ----------- Horizontal Interpolation ---------------- */ |
424 | 2.01k | pi4_ref_array = (WORD32 *) ps_ctxt->pi2_refarray_buffer; |
425 | 2.01k | i4_phase1 = ps_pos_phase[0].i2_phase; |
426 | 2.01k | i4_phase2 = (i4_phase1 + 8) & 0x0F; |
427 | | |
428 | 10.0k | for(i4_i = 0; i4_i < SUB_BLOCK_HEIGHT; i4_i++) |
429 | 8.06k | { |
430 | 8.06k | WORD16 i2_coeff1, i2_coeff2; |
431 | 8.06k | i2_coeff1 = (WORD16) (pi2_ref_data_byte[0]); |
432 | | |
433 | | /* populate the first inter sample */ |
434 | 8.06k | *pi4_ref_array++ = i2_coeff1 << 4; |
435 | | |
436 | | /* unroll count 1 */ |
437 | 8.06k | i2_coeff2 = (WORD16) (pi2_ref_data_byte[2]); |
438 | | |
439 | | /* populate 2 samples based on current coeffs */ |
440 | 8.06k | *pi4_ref_array++ = ((16 - i4_phase2) * i2_coeff1 + i4_phase2 * i2_coeff2); |
441 | | |
442 | | /* unroll count 2 */ |
443 | 8.06k | *pi4_ref_array++ = ((16 - i4_phase1) * i2_coeff1 + i4_phase1 * i2_coeff2); |
444 | | |
445 | | /* unroll count 3 */ |
446 | 8.06k | i2_coeff1 = (WORD16) (pi2_ref_data_byte[4]); |
447 | | |
448 | | /* populate 2 samples based on current coeffs */ |
449 | 8.06k | *pi4_ref_array++ = ((16 - i4_phase2) * i2_coeff2 + i4_phase2 * i2_coeff1); |
450 | | |
451 | | /* unroll count 4 */ |
452 | 8.06k | *pi4_ref_array++ = ((16 - i4_phase1) * i2_coeff2 + i4_phase1 * i2_coeff1); |
453 | | |
454 | | /* unroll count 5 */ |
455 | 8.06k | i2_coeff2 = (WORD16) (pi2_ref_data_byte[6]); |
456 | | |
457 | | /* populate 2 samples based on current coeffs */ |
458 | 8.06k | *pi4_ref_array++ = ((16 - i4_phase2) * i2_coeff1 + i4_phase2 * i2_coeff2); |
459 | | |
460 | | /* unroll count 6 */ |
461 | 8.06k | *pi4_ref_array++ = ((16 - i4_phase1) * i2_coeff1 + i4_phase1 * i2_coeff2); |
462 | | |
463 | | /* populate the last inter sample */ |
464 | 8.06k | *pi4_ref_array++ = i2_coeff2 << 4; |
465 | | |
466 | | /* vertical loop uopdates */ |
467 | 8.06k | pi2_ref_data_byte = pi2_inp_data + ((i4_i + 1) * i4_inp_data_stride); |
468 | 8.06k | } |
469 | | |
470 | | /* ----------- Vertical Interpolation ---------------- */ |
471 | 2.01k | pi4_ref_array = (WORD32 *) ps_ctxt->pi2_refarray_buffer; |
472 | 2.01k | ps_pos_phase = ps_lyr_ctxt->s_chroma_map_ctxt.ps_y_pos_phase; |
473 | 2.01k | i4_phase1 = ps_pos_phase[0].i2_phase; |
474 | 2.01k | i4_phase2 = (i4_phase1 + 8) & 0x0F; |
475 | | |
476 | 18.1k | for(i4_i = 0; i4_i < BLOCK_WIDTH; i4_i++) |
477 | 16.1k | { |
478 | 16.1k | WORD16 *pi2_out; |
479 | 16.1k | WORD32 *pi4_ref_array_temp; |
480 | 16.1k | WORD32 i4_horz_samp_1, i4_horz_samp_2; |
481 | 16.1k | pi2_out = pi2_out_res; |
482 | 16.1k | pi4_ref_array_temp = pi4_ref_array; |
483 | | |
484 | | /* populate the first inter sample */ |
485 | 16.1k | i4_horz_samp_1 = *pi4_ref_array_temp; |
486 | 16.1k | pi4_ref_array_temp += BLOCK_WIDTH; |
487 | 16.1k | *pi2_out = (i4_horz_samp_1 + 8) >> 4; |
488 | 16.1k | pi2_out += i4_out_res_stride; |
489 | | |
490 | | /* unroll count 1 */ |
491 | 16.1k | i4_horz_samp_2 = *pi4_ref_array_temp; |
492 | 16.1k | pi4_ref_array_temp += BLOCK_WIDTH; |
493 | | |
494 | | /* populate 2 samples based on current coeffs */ |
495 | 16.1k | *pi2_out = ((16 - i4_phase2) * i4_horz_samp_1 + i4_phase2 * i4_horz_samp_2 + 128) >> 8; |
496 | 16.1k | pi2_out += i4_out_res_stride; |
497 | | |
498 | | /* unroll count 2 */ |
499 | 16.1k | *pi2_out = ((16 - i4_phase1) * i4_horz_samp_1 + i4_phase1 * i4_horz_samp_2 + 128) >> 8; |
500 | 16.1k | pi2_out += i4_out_res_stride; |
501 | | |
502 | | /* unroll count 3 */ |
503 | 16.1k | i4_horz_samp_1 = *pi4_ref_array_temp; |
504 | 16.1k | pi4_ref_array_temp += BLOCK_WIDTH; |
505 | | |
506 | | /* populate 2 samples based on current coeffs */ |
507 | 16.1k | *pi2_out = ((16 - i4_phase2) * i4_horz_samp_2 + i4_phase2 * i4_horz_samp_1 + 128) >> 8; |
508 | 16.1k | pi2_out += i4_out_res_stride; |
509 | | |
510 | | /* unroll count 4 */ |
511 | 16.1k | *pi2_out = ((16 - i4_phase1) * i4_horz_samp_2 + i4_phase1 * i4_horz_samp_1 + 128) >> 8; |
512 | 16.1k | pi2_out += i4_out_res_stride; |
513 | | |
514 | | /* unroll count 5 */ |
515 | 16.1k | i4_horz_samp_2 = *pi4_ref_array_temp; |
516 | | |
517 | | /* populate 2 samples based on current coeffs */ |
518 | 16.1k | *pi2_out = ((16 - i4_phase2) * i4_horz_samp_1 + i4_phase2 * i4_horz_samp_2 + 128) >> 8; |
519 | 16.1k | pi2_out += i4_out_res_stride; |
520 | | |
521 | | /* unroll count 6 */ |
522 | 16.1k | *pi2_out = ((16 - i4_phase1) * i4_horz_samp_1 + i4_phase1 * i4_horz_samp_2 + 128) >> 8; |
523 | 16.1k | pi2_out += i4_out_res_stride; |
524 | | |
525 | | /* populate the last inter sample */ |
526 | 16.1k | *pi2_out = (i4_horz_samp_2 + 8) >> 4; |
527 | | |
528 | | /* horizontal loop updates */ |
529 | 16.1k | pi4_ref_array++; |
530 | 16.1k | pi2_out_res += 2; |
531 | 16.1k | } |
532 | 2.01k | } |
533 | 2.01k | return; |
534 | 2.01k | } |
535 | | |
536 | | /*****************************************************************************/ |
537 | | /* */ |
538 | | /* Function Name : isvcd_residual_luma_dyadic */ |
539 | | /* */ |
540 | | /* Description : this fucntion does the upsampling of luma residuals for */ |
541 | | /* Dyadic cases */ |
542 | | /* */ |
543 | | /* Inputs : pv_residual_samp_ctxt : Residual upsampling context */ |
544 | | /* pu1_inp_data : input 8 bit data pointer */ |
545 | | /* i4_inp_data_stride : input buffer stride */ |
546 | | /* pi2_out_res : output 16 bit buffer pointer */ |
547 | | /* i4_out_res_stride : Output buffer stride */ |
548 | | /* pu1_inp_bitmap : input packed sign bit data pointer */ |
549 | | /* i4_inp_bitmap_stride : sign bit buffer stride */ |
550 | | /* ps_ref_mb_mode : reference mb mode pointer of base layer */ |
551 | | /* ps_coord : mb co-ordinate pointer */ |
552 | | /* Globals : none */ |
553 | | /* Processing : it does the upsampling with fixed phase values and */ |
554 | | /* reference layer transform size */ |
555 | | /* Outputs : Upsampled residuals for luma */ |
556 | | /* Returns : none */ |
557 | | /* */ |
558 | | /* Issues : none */ |
559 | | /* */ |
560 | | /* Revision History: */ |
561 | | /* */ |
562 | | /* DD MM YYYY Author(s) Changes (Describe the changes made) */ |
563 | | /* 25 09 2021 vijayakumar creation */ |
564 | | /* */ |
565 | | /*****************************************************************************/ |
566 | | void isvcd_residual_luma_dyadic(void *pv_residual_samp_ctxt, WORD16 *pi2_inp_data, |
567 | | WORD32 i4_inp_data_stride, WORD16 *pi2_out_res, |
568 | | WORD32 i4_out_res_stride, mem_element_t *ps_ref_mb_mode, |
569 | | UWORD16 u2_mb_x, UWORD16 u2_mb_y, WORD32 i4_ref_nnz, |
570 | | WORD32 i4_ref_tx_size) |
571 | | |
572 | 23.1k | { |
573 | 23.1k | WORD16 *pi2_refarray_buffer; |
574 | 23.1k | WORD32 i4_blk_ctr; |
575 | 23.1k | residual_sampling_ctxt_t *ps_ctxt; |
576 | | |
577 | 23.1k | UNUSED(ps_ref_mb_mode); |
578 | 23.1k | UNUSED(u2_mb_x); |
579 | 23.1k | UNUSED(u2_mb_y); |
580 | | |
581 | 23.1k | ps_ctxt = (residual_sampling_ctxt_t *) pv_residual_samp_ctxt; |
582 | 23.1k | pi2_refarray_buffer = ps_ctxt->pi2_refarray_buffer; |
583 | | |
584 | | /* based on transform size the counter and interpolation width and */ |
585 | | /* height are intialised as follows */ |
586 | 23.1k | if((i4_ref_tx_size) && (0 != i4_ref_nnz)) |
587 | 1.67k | { |
588 | 1.67k | WORD16 *pi2_ref_data_byte; |
589 | 1.67k | WORD32 *pi4_ref_array; |
590 | 1.67k | WORD32 i4_i, i4_j; |
591 | | |
592 | 1.67k | pi2_ref_data_byte = pi2_inp_data; |
593 | | |
594 | | /* ----------- Horizontal Interpolation ---------------- */ |
595 | 1.67k | pi4_ref_array = (WORD32 *) pi2_refarray_buffer; |
596 | 15.0k | for(i4_i = 0; i4_i < BLOCK_HEIGHT; i4_i++) |
597 | 13.3k | { |
598 | 13.3k | WORD16 i2_coeff1, i2_coeff2; |
599 | 13.3k | i2_coeff1 = (WORD16) (*pi2_ref_data_byte++); |
600 | | |
601 | | /* populate the first inter sample */ |
602 | 13.3k | *pi4_ref_array++ = i2_coeff1 << 2; |
603 | | |
604 | 106k | for(i4_j = 0; i4_j < 14; i4_j += 2) |
605 | 93.5k | { |
606 | 93.5k | i2_coeff2 = (WORD16) (*pi2_ref_data_byte++); |
607 | | |
608 | | /* populate 2 samples based on current coeffs */ |
609 | 93.5k | *pi4_ref_array++ = ((i2_coeff1 << 1) + (i2_coeff1) + (i2_coeff2)); |
610 | 93.5k | *pi4_ref_array++ = ((i2_coeff2 << 1) + (i2_coeff2) + (i2_coeff1)); |
611 | | |
612 | | /* store the coeff 2 to coeff 1 */ |
613 | | /* (used in next iteration) */ |
614 | 93.5k | i2_coeff1 = i2_coeff2; |
615 | 93.5k | } |
616 | | |
617 | | /* populate the last inter sample */ |
618 | 13.3k | *pi4_ref_array++ = i2_coeff1 << 2; |
619 | | |
620 | | /* vertical loop uopdates */ |
621 | 13.3k | pi2_ref_data_byte = pi2_inp_data + ((i4_i + 1) * i4_inp_data_stride); |
622 | 13.3k | } |
623 | | |
624 | | /* ----------- Vertical Interpolation ---------------- */ |
625 | 1.67k | pi4_ref_array = (WORD32 *) pi2_refarray_buffer; |
626 | | |
627 | 28.3k | for(i4_i = 0; i4_i < MB_WIDTH; i4_i++) |
628 | 26.7k | { |
629 | 26.7k | WORD32 *pi4_ref_array_temp; |
630 | 26.7k | WORD16 *pi2_out; |
631 | 26.7k | WORD32 i4_horz_samp_1, i4_horz_samp_2; |
632 | | |
633 | 26.7k | pi4_ref_array_temp = pi4_ref_array; |
634 | 26.7k | pi2_out = pi2_out_res; |
635 | 26.7k | i4_horz_samp_1 = *pi4_ref_array_temp; |
636 | | |
637 | | /* populate the first inter sample */ |
638 | 26.7k | *pi2_out = (i4_horz_samp_1 + 2) >> 2; |
639 | 26.7k | pi2_out += i4_out_res_stride; |
640 | | |
641 | 213k | for(i4_j = 0; i4_j < 14; i4_j += 2) |
642 | 187k | { |
643 | 187k | pi4_ref_array_temp += MB_WIDTH; |
644 | 187k | i4_horz_samp_2 = *pi4_ref_array_temp; |
645 | | |
646 | | /* populate 2 samples based on current coeffs */ |
647 | 187k | *pi2_out = ((i4_horz_samp_1 << 1) + (i4_horz_samp_1) + (i4_horz_samp_2) + 8) >> 4; |
648 | 187k | pi2_out += i4_out_res_stride; |
649 | 187k | *pi2_out = ((i4_horz_samp_2 << 1) + (i4_horz_samp_2) + (i4_horz_samp_1) + 8) >> 4; |
650 | 187k | pi2_out += i4_out_res_stride; |
651 | | |
652 | | /* store the coeff 2 to coeff 1 */ |
653 | | /* (used in next iteration) */ |
654 | 187k | i4_horz_samp_1 = i4_horz_samp_2; |
655 | 187k | } |
656 | | |
657 | | /* populate the first inter sample */ |
658 | 26.7k | *pi2_out = (i4_horz_samp_1 + 2) >> 2; |
659 | | |
660 | | /* horizontal loop updates */ |
661 | 26.7k | pi4_ref_array++; |
662 | 26.7k | pi2_out_res++; |
663 | 26.7k | } |
664 | 1.67k | } |
665 | 21.5k | else |
666 | 21.5k | { |
667 | | /* ----------------------------------------------------------------- */ |
668 | | /* LOOP over number of blocks */ |
669 | | /* ----------------------------------------------------------------- */ |
670 | 107k | for(i4_blk_ctr = 0; i4_blk_ctr < 4; i4_blk_ctr++) |
671 | 86.0k | { |
672 | 86.0k | WORD16 *pi2_ref_data_byte; |
673 | 86.0k | WORD32 *pi4_ref_array; |
674 | 86.0k | WORD32 i4_i; |
675 | | |
676 | | /* if reference layer is not coded then no processing */ |
677 | 86.0k | if(0 != (i4_ref_nnz & 0x1)) |
678 | 7.75k | { |
679 | 7.75k | pi2_ref_data_byte = pi2_inp_data; |
680 | | |
681 | | /* ----------- Horizontal Interpolation ---------------- */ |
682 | 7.75k | pi4_ref_array = (WORD32 *) pi2_refarray_buffer; |
683 | | |
684 | 38.7k | for(i4_i = 0; i4_i < SUB_BLOCK_HEIGHT; i4_i++) |
685 | 31.0k | { |
686 | 31.0k | WORD16 i2_coeff1, i2_coeff2; |
687 | 31.0k | i2_coeff1 = (WORD16) (*pi2_ref_data_byte++); |
688 | | |
689 | | /* populate the first inter sample */ |
690 | 31.0k | *pi4_ref_array++ = i2_coeff1 << 2; |
691 | | |
692 | 31.0k | i2_coeff2 = (WORD16) (*pi2_ref_data_byte++); |
693 | | |
694 | | /* populate 2 samples based on current coeffs */ |
695 | 31.0k | *pi4_ref_array++ = ((i2_coeff1 << 1) + (i2_coeff1) + (i2_coeff2)); |
696 | 31.0k | *pi4_ref_array++ = ((i2_coeff2 << 1) + (i2_coeff2) + (i2_coeff1)); |
697 | | |
698 | 31.0k | i2_coeff1 = (WORD16) (*pi2_ref_data_byte++); |
699 | | |
700 | | /* populate 2 samples based on current coeffs */ |
701 | 31.0k | *pi4_ref_array++ = ((i2_coeff2 << 1) + (i2_coeff2) + (i2_coeff1)); |
702 | 31.0k | *pi4_ref_array++ = ((i2_coeff1 << 1) + (i2_coeff1) + (i2_coeff2)); |
703 | | |
704 | 31.0k | i2_coeff2 = (WORD16) (*pi2_ref_data_byte++); |
705 | | |
706 | | /* populate 2 samples based on current coeffs */ |
707 | 31.0k | *pi4_ref_array++ = ((i2_coeff1 << 1) + (i2_coeff1) + (i2_coeff2)); |
708 | 31.0k | *pi4_ref_array++ = ((i2_coeff2 << 1) + (i2_coeff2) + (i2_coeff1)); |
709 | | |
710 | | /* populate the last inter sample */ |
711 | 31.0k | *pi4_ref_array++ = i2_coeff2 << 2; |
712 | | |
713 | | /* vertical loop uopdates */ |
714 | 31.0k | pi2_ref_data_byte = pi2_inp_data + ((i4_i + 1) * i4_inp_data_stride); |
715 | 31.0k | } |
716 | | |
717 | | /* ----------- Vertical Interpolation ---------------- */ |
718 | 7.75k | pi4_ref_array = (WORD32 *) pi2_refarray_buffer; |
719 | | |
720 | 69.8k | for(i4_i = 0; i4_i < BLOCK_WIDTH; i4_i++) |
721 | 62.0k | { |
722 | 62.0k | WORD32 *pi4_ref_array_temp; |
723 | 62.0k | WORD16 *pi2_out; |
724 | 62.0k | WORD32 i4_horz_samp_1, i4_horz_samp_2; |
725 | | |
726 | 62.0k | pi4_ref_array_temp = pi4_ref_array; |
727 | 62.0k | pi2_out = pi2_out_res; |
728 | 62.0k | i4_horz_samp_1 = *pi4_ref_array_temp; |
729 | | |
730 | | /* populate the first inter sample */ |
731 | 62.0k | *pi2_out = (i4_horz_samp_1 + 2) >> 2; |
732 | 62.0k | pi2_out += i4_out_res_stride; |
733 | | |
734 | | /* unroll loop count 1 */ |
735 | 62.0k | pi4_ref_array_temp += BLOCK_WIDTH; |
736 | 62.0k | i4_horz_samp_2 = *pi4_ref_array_temp; |
737 | | |
738 | | /* populate 2 samples based on current coeffs */ |
739 | 62.0k | *pi2_out = |
740 | 62.0k | ((i4_horz_samp_1 << 1) + (i4_horz_samp_1) + (i4_horz_samp_2) + 8) >> 4; |
741 | 62.0k | pi2_out += i4_out_res_stride; |
742 | 62.0k | *pi2_out = |
743 | 62.0k | ((i4_horz_samp_2 << 1) + (i4_horz_samp_2) + (i4_horz_samp_1) + 8) >> 4; |
744 | 62.0k | pi2_out += i4_out_res_stride; |
745 | | |
746 | | /* unroll loop count 2 */ |
747 | 62.0k | pi4_ref_array_temp += BLOCK_WIDTH; |
748 | 62.0k | i4_horz_samp_1 = *pi4_ref_array_temp; |
749 | | |
750 | | /* populate 2 samples based on current coeffs */ |
751 | 62.0k | *pi2_out = |
752 | 62.0k | ((i4_horz_samp_2 << 1) + (i4_horz_samp_2) + (i4_horz_samp_1) + 8) >> 4; |
753 | 62.0k | pi2_out += i4_out_res_stride; |
754 | 62.0k | *pi2_out = |
755 | 62.0k | ((i4_horz_samp_1 << 1) + (i4_horz_samp_1) + (i4_horz_samp_2) + 8) >> 4; |
756 | 62.0k | pi2_out += i4_out_res_stride; |
757 | | |
758 | | /* unroll loop count 3 */ |
759 | 62.0k | pi4_ref_array_temp += BLOCK_WIDTH; |
760 | 62.0k | i4_horz_samp_2 = *pi4_ref_array_temp; |
761 | | |
762 | | /* populate 2 samples based on current coeffs */ |
763 | 62.0k | *pi2_out = |
764 | 62.0k | ((i4_horz_samp_1 << 1) + (i4_horz_samp_1) + (i4_horz_samp_2) + 8) >> 4; |
765 | 62.0k | pi2_out += i4_out_res_stride; |
766 | 62.0k | *pi2_out = |
767 | 62.0k | ((i4_horz_samp_2 << 1) + (i4_horz_samp_2) + (i4_horz_samp_1) + 8) >> 4; |
768 | 62.0k | pi2_out += i4_out_res_stride; |
769 | | |
770 | | /* populate the last inter sample */ |
771 | 62.0k | *pi2_out = (i4_horz_samp_2 + 2) >> 2; |
772 | | |
773 | | /* horizontal loop updates */ |
774 | 62.0k | pi4_ref_array++; |
775 | 62.0k | pi2_out_res++; |
776 | 62.0k | } |
777 | 7.75k | } |
778 | 78.3k | else |
779 | 78.3k | { |
780 | 78.3k | pi2_out_res += BLOCK_WIDTH; |
781 | 78.3k | } |
782 | | |
783 | | /* Block level loop updates */ |
784 | 86.0k | if(1 == i4_blk_ctr) |
785 | 21.5k | { |
786 | 21.5k | pi2_inp_data -= SUB_BLOCK_WIDTH; |
787 | 21.5k | pi2_inp_data += (i4_inp_data_stride * SUB_BLOCK_HEIGHT); |
788 | 21.5k | pi2_out_res -= MB_WIDTH; |
789 | 21.5k | pi2_out_res += (i4_out_res_stride * BLOCK_HEIGHT); |
790 | 21.5k | i4_ref_nnz >>= 2; |
791 | 21.5k | } |
792 | 64.5k | else |
793 | 64.5k | { |
794 | 64.5k | pi2_inp_data += SUB_BLOCK_WIDTH; |
795 | 64.5k | } |
796 | | |
797 | 86.0k | i4_ref_nnz >>= 1; |
798 | | |
799 | 86.0k | } /* end of loop over all the blocks */ |
800 | 21.5k | } |
801 | 23.1k | return; |
802 | 23.1k | } |
803 | | |
804 | | /*****************************************************************************/ |
805 | | /* */ |
806 | | /* Function Name : isvcd_ref_layer_ptr_incr */ |
807 | | /* */ |
808 | | /* Description : this function returns the pointer increments for */ |
809 | | /* the operand2 of the bilinear interpolation */ |
810 | | /* Inputs : pi1_ref_mb_modes : reference mb modes */ |
811 | | /* i4_ref_mode_stride : mb mode buffer stride */ |
812 | | /* i4_element_size : size of reference mb mode */ |
813 | | /* i4_x_offset : ref offset x */ |
814 | | /* i4_y_offset : ref offset y */ |
815 | | /* i4_refary_wd : reference array width */ |
816 | | /* i4_refary_ht : reference array height */ |
817 | | /* pu1_ref_x_ptr_incr : ptr increment buffer for x */ |
818 | | /* pu1_ref_y_ptr_incr : ptr increment buffer for y */ |
819 | | /* i4_chroma_flag : chroma processing flag */ |
820 | | /* Globals : none */ |
821 | | /* Processing : it calculates the increment as per the transform size */ |
822 | | /* */ |
823 | | /* Outputs : none */ |
824 | | /* Returns : none */ |
825 | | /* */ |
826 | | /* Issues : none */ |
827 | | /* */ |
828 | | /* Revision History: */ |
829 | | /* */ |
830 | | /* DD MM YYYY Author(s) Changes (Describe the changes made) */ |
831 | | /* 18 08 2021 Kishore creation */ |
832 | | /* */ |
833 | | /*****************************************************************************/ |
834 | | WORD32 isvcd_ref_layer_ptr_incr(WORD8 *pi1_ref_mb_modes, WORD32 i4_ref_mode_stride, |
835 | | WORD32 i4_element_size, WORD32 i4_x_offset, WORD32 i4_y_offset, |
836 | | WORD32 i4_refary_wd, WORD32 i4_refary_ht, |
837 | | UWORD8 *pu1_ref_x_ptr_incr, UWORD8 *pu1_ref_y_ptr_incr, |
838 | | WORD32 i4_chroma_flag) |
839 | 327k | { |
840 | 327k | WORD32 i4_x, i4_y; |
841 | 327k | WORD32 i4_x_idx, i4_y_idx; |
842 | 327k | WORD32 i4_prev_x, i4_prev_y; |
843 | 327k | WORD32 i4_const_val; |
844 | 327k | WORD32 i4_pos_x, i4_pos_y; |
845 | 327k | WORD32 i4_trans_size; |
846 | 327k | WORD32 i4_act_ary_wd, i4_act_ary_ht; |
847 | 327k | WORD32 i4_and_const; |
848 | 327k | UWORD8 *pu1_incr_x, *pu1_incr_y; |
849 | 327k | WORD32 i4_mb_sft; |
850 | 327k | WORD32 i4_mb_x, i4_mb_y; |
851 | 327k | WORD8 *pi1_ref_mb_modes_incr; |
852 | 327k | WORD8 *pi1_ref_mb_modes_incr_temp; |
853 | 327k | inter_lyr_mb_prms_t *ps_inter_lyr_mb_prms; |
854 | 327k | WORD32 i4_mb_x_strt, i4_mb_y_strt; |
855 | 327k | WORD32 i4_mb_quard1_part_x, i4_mb_quard1_part_y; |
856 | 327k | WORD32 i4_x_ref, i4_y_ref; |
857 | 327k | WORD32 i4_tx_size, i4_tx_size_q0, i4_tx_size_q1, i4_tx_size_q2, i4_tx_size_q3; |
858 | 327k | WORD8 i1_mb_mode_q0, i1_mb_mode_q1, i1_mb_mode_q2, i1_mb_mode_q3; |
859 | 327k | WORD32 i4_mb_wd; |
860 | 327k | WORD32 i4_mb_ht; |
861 | | |
862 | 327k | i4_mb_wd = MB_WIDTH >> i4_chroma_flag; |
863 | 327k | i4_mb_ht = MB_HEIGHT >> i4_chroma_flag; |
864 | | |
865 | | /* Memset to 1 the increment buffers */ |
866 | 327k | memset(pu1_ref_x_ptr_incr, 1, (i4_refary_wd * i4_refary_ht)); |
867 | 327k | memset(pu1_ref_y_ptr_incr, 1, (i4_refary_wd * i4_refary_ht)); |
868 | | |
869 | | /* Initialise actual width and height */ |
870 | 327k | i4_act_ary_wd = i4_refary_wd; |
871 | 327k | i4_act_ary_ht = i4_refary_ht; |
872 | | |
873 | | /* Initialize x and y */ |
874 | 327k | i4_x = 0; |
875 | 327k | i4_y = 0; |
876 | 327k | i4_prev_y = 0; |
877 | 327k | i4_mb_sft = (MB_WIDTH_SHIFT - i4_chroma_flag); |
878 | | |
879 | | /* Loop over all MBs in the reference array */ |
880 | 327k | if(0 == i4_chroma_flag) |
881 | 109k | { |
882 | 109k | i4_x_ref = i4_x_offset + 0; |
883 | 109k | i4_y_ref = i4_y_offset + 0; |
884 | 109k | i4_mb_x_strt = i4_x_ref % i4_mb_wd; |
885 | 109k | i4_mb_y_strt = i4_y_ref % i4_mb_ht; |
886 | 109k | i4_mb_quard1_part_x = i4_mb_wd - i4_mb_x_strt; |
887 | 109k | i4_mb_quard1_part_y = i4_mb_ht - i4_mb_y_strt; |
888 | | |
889 | 109k | if(!(i4_mb_quard1_part_x >= 0)) |
890 | 0 | { |
891 | 0 | return NOT_OK; |
892 | 0 | } |
893 | 109k | if(!(i4_mb_quard1_part_y >= 0)) |
894 | 0 | { |
895 | 0 | return NOT_OK; |
896 | 0 | } |
897 | | |
898 | | /* Take care of negative offsets */ |
899 | 109k | if(i4_x_ref > 0) |
900 | 72.1k | { |
901 | 72.1k | i4_mb_x = (i4_x_ref >> i4_mb_sft); |
902 | 72.1k | } |
903 | 36.9k | else |
904 | 36.9k | { |
905 | 36.9k | i4_mb_x = 0; |
906 | 36.9k | } |
907 | 109k | if(i4_y_ref > 0) |
908 | 92.3k | { |
909 | 92.3k | i4_mb_y = (i4_y_ref >> i4_mb_sft); |
910 | 92.3k | } |
911 | 16.7k | else |
912 | 16.7k | { |
913 | 16.7k | i4_mb_y = 0; |
914 | 16.7k | } |
915 | | |
916 | | /* get the location of the byte which has the current mb mode */ |
917 | 109k | pi1_ref_mb_modes_incr = pi1_ref_mb_modes + (i4_mb_y * i4_ref_mode_stride * i4_element_size); |
918 | 109k | pi1_ref_mb_modes_incr += (i4_mb_x * i4_element_size); |
919 | 109k | ps_inter_lyr_mb_prms = (inter_lyr_mb_prms_t *) pi1_ref_mb_modes_incr; |
920 | 109k | i1_mb_mode_q0 = ps_inter_lyr_mb_prms->i1_mb_mode; |
921 | 109k | i4_tx_size_q0 = |
922 | 109k | (i1_mb_mode_q0 <= SVC_INTER_MB) |
923 | 109k | ? ((ps_inter_lyr_mb_prms->i1_tx_size < 0) ? 1 : ps_inter_lyr_mb_prms->i1_tx_size) |
924 | 109k | : 1; |
925 | | |
926 | 109k | pi1_ref_mb_modes_incr_temp = pi1_ref_mb_modes_incr; |
927 | 109k | if(i4_mb_quard1_part_x > 0) |
928 | 109k | { |
929 | 109k | pi1_ref_mb_modes_incr_temp = pi1_ref_mb_modes_incr + i4_element_size; |
930 | 109k | ps_inter_lyr_mb_prms = (inter_lyr_mb_prms_t *) pi1_ref_mb_modes_incr_temp; |
931 | 109k | i1_mb_mode_q1 = ps_inter_lyr_mb_prms->i1_mb_mode; |
932 | 109k | i4_tx_size_q1 = |
933 | 109k | (i1_mb_mode_q1 <= SVC_INTER_MB) |
934 | 109k | ? ((ps_inter_lyr_mb_prms->i1_tx_size < 0) ? 1 |
935 | 101k | : ps_inter_lyr_mb_prms->i1_tx_size) |
936 | 109k | : 1; |
937 | 109k | } |
938 | | |
939 | 109k | if(i4_mb_quard1_part_y > 0) |
940 | 109k | { |
941 | 109k | pi1_ref_mb_modes_incr_temp = |
942 | 109k | pi1_ref_mb_modes_incr + (i4_ref_mode_stride * i4_element_size); |
943 | 109k | ps_inter_lyr_mb_prms = (inter_lyr_mb_prms_t *) pi1_ref_mb_modes_incr_temp; |
944 | 109k | i1_mb_mode_q2 = ps_inter_lyr_mb_prms->i1_mb_mode; |
945 | 109k | i4_tx_size_q2 = |
946 | 109k | (i1_mb_mode_q2 <= SVC_INTER_MB) |
947 | 109k | ? ((ps_inter_lyr_mb_prms->i1_tx_size < 0) ? 1 |
948 | 103k | : ps_inter_lyr_mb_prms->i1_tx_size) |
949 | 109k | : 1; |
950 | 109k | } |
951 | | |
952 | 109k | if((i4_mb_quard1_part_x > 0) && (i4_mb_quard1_part_y > 0)) |
953 | 109k | { |
954 | 109k | pi1_ref_mb_modes_incr_temp = |
955 | 109k | pi1_ref_mb_modes_incr + (i4_ref_mode_stride * i4_element_size) + i4_element_size; |
956 | 109k | ps_inter_lyr_mb_prms = (inter_lyr_mb_prms_t *) pi1_ref_mb_modes_incr_temp; |
957 | 109k | i1_mb_mode_q3 = ps_inter_lyr_mb_prms->i1_mb_mode; |
958 | 109k | i4_tx_size_q3 = |
959 | 109k | (i1_mb_mode_q3 <= SVC_INTER_MB) |
960 | 109k | ? ((ps_inter_lyr_mb_prms->i1_tx_size < 0) ? 1 |
961 | 103k | : ps_inter_lyr_mb_prms->i1_tx_size) |
962 | 109k | : 1; |
963 | 109k | } |
964 | | |
965 | 109k | do |
966 | 436k | { |
967 | 436k | WORD32 i4_idx; |
968 | 436k | WORD32 i4_wd, i4_ht; |
969 | 436k | WORD32 i4_max_pos_x, i4_max_pos_y; |
970 | | |
971 | 436k | i4_prev_x = i4_x; |
972 | 436k | i4_x_ref = i4_x_offset + i4_x; |
973 | 436k | i4_y_ref = i4_y_offset + i4_y; |
974 | 436k | i4_tx_size = i4_tx_size_q0; |
975 | 436k | if(i4_x >= i4_mb_quard1_part_x) |
976 | 144k | { |
977 | 144k | if(i4_y < i4_mb_quard1_part_y) |
978 | 83.5k | { |
979 | 83.5k | i4_tx_size = i4_tx_size_q1; |
980 | 83.5k | } |
981 | 60.7k | else if(i4_y >= i4_mb_quard1_part_y) |
982 | 60.7k | { |
983 | 60.7k | i4_tx_size = i4_tx_size_q3; |
984 | 60.7k | } |
985 | 144k | } |
986 | 292k | else if(i4_x < i4_mb_quard1_part_x) |
987 | 292k | { |
988 | 292k | if(i4_y >= i4_mb_quard1_part_y) |
989 | 123k | { |
990 | 123k | i4_tx_size = i4_tx_size_q2; |
991 | 123k | } |
992 | 292k | } |
993 | | |
994 | | /* Get the transform size as 4 or 8 */ |
995 | 436k | i4_trans_size = ((i4_tx_size + 1) << 2); |
996 | 436k | i4_const_val = i4_trans_size - 1; |
997 | 436k | i4_and_const = i4_const_val; |
998 | | |
999 | | /* Fill horizontal tx block edges of current reference mb with 0 */ |
1000 | 436k | pu1_incr_x = pu1_ref_x_ptr_incr + i4_x; |
1001 | 436k | pu1_incr_x += (i4_y * i4_refary_wd); |
1002 | 436k | i4_ht = (16 - (i4_y_ref & 0xF)); |
1003 | 436k | i4_ht = MIN((i4_act_ary_ht - i4_y), i4_ht); |
1004 | | |
1005 | 436k | i4_x_idx = i4_x; |
1006 | 436k | i4_pos_x = i4_x_ref & 0xF; |
1007 | 436k | i4_max_pos_x = 16; |
1008 | 436k | i4_x += (16 - i4_pos_x); |
1009 | | |
1010 | | /* Get the transform block edge pos */ |
1011 | 436k | i4_idx = (i4_const_val - (i4_pos_x & i4_and_const)); |
1012 | 436k | i4_x_idx += i4_idx; |
1013 | | |
1014 | 1.06M | while((i4_pos_x < i4_max_pos_x) && (i4_x_idx < i4_act_ary_wd)) |
1015 | 628k | { |
1016 | 628k | WORD32 i4_i; |
1017 | 628k | UWORD8 *pu1_incr; |
1018 | | |
1019 | 628k | pu1_incr = pu1_incr_x + i4_idx; |
1020 | 4.45M | for(i4_i = 0; i4_i < i4_ht; i4_i++) |
1021 | 3.82M | { |
1022 | | /* Fill the block edge with 0s */ |
1023 | 3.82M | *pu1_incr = 0; |
1024 | 3.82M | pu1_incr += i4_refary_wd; |
1025 | 3.82M | } |
1026 | | |
1027 | | /* Updates */ |
1028 | 628k | i4_pos_x += i4_trans_size; |
1029 | 628k | pu1_incr_x += i4_trans_size; |
1030 | 628k | i4_x_idx += MIN(i4_trans_size, (i4_act_ary_wd - i4_x_idx)); |
1031 | 628k | } |
1032 | | |
1033 | | /* Fill vertical tx block edges of current reference mb with 0 */ |
1034 | 436k | pu1_incr_y = pu1_ref_y_ptr_incr + i4_prev_x; |
1035 | 436k | pu1_incr_y += (i4_y * i4_refary_wd); |
1036 | 436k | i4_wd = (16 - (i4_x_ref & 0xF)); |
1037 | 436k | i4_wd = MIN((i4_act_ary_wd - i4_prev_x), i4_wd); |
1038 | 436k | i4_y_idx = i4_y; |
1039 | 436k | i4_pos_y = i4_y_ref & 0xF; |
1040 | 436k | i4_max_pos_y = 16; |
1041 | 436k | i4_y += (16 - i4_pos_y); |
1042 | | |
1043 | | /* Get the transform block edge pos */ |
1044 | 436k | i4_idx = (i4_const_val - (i4_pos_y & i4_and_const)); |
1045 | 436k | i4_y_idx += i4_idx; |
1046 | | |
1047 | 1.03M | while((i4_pos_y < i4_max_pos_y) && (i4_y_idx < i4_act_ary_ht)) |
1048 | 595k | { |
1049 | 595k | WORD32 i4_i; |
1050 | 595k | UWORD8 *pu1_incr; |
1051 | | |
1052 | 595k | pu1_incr = pu1_incr_y + i4_idx * i4_refary_wd; |
1053 | 4.38M | for(i4_i = 0; i4_i < i4_wd; i4_i++) |
1054 | 3.79M | { |
1055 | | /* Fill the block edge with 0s */ |
1056 | 3.79M | *pu1_incr = 0; |
1057 | 3.79M | pu1_incr++; |
1058 | 3.79M | } |
1059 | | |
1060 | | /* Updates */ |
1061 | 595k | i4_pos_y += i4_trans_size; |
1062 | 595k | pu1_incr_y += i4_trans_size * i4_refary_wd; |
1063 | 595k | i4_y_idx += MIN(i4_trans_size, (i4_act_ary_ht - i4_y_idx)); |
1064 | 595k | } |
1065 | | |
1066 | | /* Loop updates */ |
1067 | 436k | if(i4_x < i4_act_ary_wd) |
1068 | 218k | { |
1069 | 218k | i4_y = i4_prev_y; |
1070 | 218k | } |
1071 | 218k | else if(i4_y < i4_act_ary_ht) |
1072 | 109k | { |
1073 | 109k | i4_prev_y = i4_y; |
1074 | 109k | i4_x = 0; |
1075 | 109k | } |
1076 | | |
1077 | 436k | } while((i4_y < i4_act_ary_ht) || (i4_x < i4_act_ary_wd)); |
1078 | | |
1079 | 109k | } /* End of if 0 == i4_chroma_flag */ |
1080 | 218k | else |
1081 | 218k | { |
1082 | | /* Set the transform size as 4 */ |
1083 | 218k | i4_trans_size = 4; |
1084 | 218k | i4_const_val = 3; |
1085 | | |
1086 | 218k | do |
1087 | 872k | { |
1088 | 872k | WORD32 i4_x_ref, i4_y_ref; |
1089 | 872k | WORD32 i4_idx; |
1090 | 872k | WORD32 i4_wd, i4_ht; |
1091 | 872k | WORD32 i4_max_pos_x, i4_max_pos_y; |
1092 | | |
1093 | 872k | i4_prev_x = i4_x; |
1094 | 872k | i4_x_ref = i4_x_offset + i4_x; |
1095 | 872k | i4_y_ref = i4_y_offset + i4_y; |
1096 | | |
1097 | | /* Fill horizontal tx block edges of current reference mb with 0 */ |
1098 | 872k | pu1_incr_x = pu1_ref_x_ptr_incr + i4_x; |
1099 | 872k | pu1_incr_x += (i4_y * i4_refary_wd); |
1100 | 872k | i4_ht = (8 - (i4_y_ref & 0x7)); |
1101 | 872k | i4_ht = MIN((i4_act_ary_ht - i4_y), i4_ht); |
1102 | 872k | i4_x_idx = i4_x; |
1103 | 872k | i4_pos_x = i4_x_ref & 0x7; |
1104 | 872k | i4_max_pos_x = 8; |
1105 | 872k | i4_x += (8 - i4_pos_x); |
1106 | | |
1107 | | /* Get the transform block edge pos */ |
1108 | 872k | i4_idx = (i4_const_val - (i4_pos_x & 0x3)); |
1109 | 872k | i4_x_idx += i4_idx; |
1110 | | |
1111 | 1.59M | while((i4_pos_x < i4_max_pos_x) && (i4_x_idx < i4_act_ary_wd)) |
1112 | 722k | { |
1113 | 722k | WORD32 i4_i; |
1114 | 722k | UWORD8 *pu1_incr; |
1115 | | |
1116 | 722k | pu1_incr = pu1_incr_x + i4_idx; |
1117 | 3.12M | for(i4_i = 0; i4_i < i4_ht; i4_i++) |
1118 | 2.40M | { |
1119 | | /* Fill the block edge with 0s */ |
1120 | 2.40M | *pu1_incr = 0; |
1121 | 2.40M | pu1_incr += i4_refary_wd; |
1122 | 2.40M | } |
1123 | | |
1124 | | /* Updates */ |
1125 | 722k | i4_pos_x += i4_trans_size; |
1126 | 722k | pu1_incr_x += i4_trans_size; |
1127 | 722k | i4_x_idx += MIN(i4_trans_size, (i4_act_ary_wd - i4_x_idx)); |
1128 | 722k | } |
1129 | | |
1130 | | /* Fill vertical tx block edges of current reference mb with 0 */ |
1131 | 872k | pu1_incr_y = pu1_ref_y_ptr_incr + i4_prev_x; |
1132 | 872k | pu1_incr_y += (i4_y * i4_refary_wd); |
1133 | 872k | i4_wd = (8 - (i4_x_ref & 0x7)); |
1134 | 872k | i4_wd = MIN((i4_act_ary_wd - i4_prev_x), i4_wd); |
1135 | 872k | i4_y_idx = i4_y; |
1136 | 872k | i4_pos_y = i4_y_ref & 0x7; |
1137 | 872k | i4_max_pos_y = 8; |
1138 | 872k | i4_y += (8 - i4_pos_y); |
1139 | | |
1140 | | /* Get the transform block edge pos */ |
1141 | 872k | i4_idx = (i4_const_val - (i4_pos_y & 0x3)); |
1142 | 872k | i4_y_idx += i4_idx; |
1143 | | |
1144 | 1.59M | while((i4_pos_y < i4_max_pos_y) && (i4_y_idx < i4_act_ary_ht)) |
1145 | 724k | { |
1146 | 724k | WORD32 i4_i; |
1147 | 724k | UWORD8 *pu1_incr; |
1148 | | |
1149 | 724k | pu1_incr = pu1_incr_y + i4_idx * i4_refary_wd; |
1150 | 3.13M | for(i4_i = 0; i4_i < i4_wd; i4_i++) |
1151 | 2.40M | { |
1152 | | /* Fill the block edge with 0s */ |
1153 | 2.40M | *pu1_incr = 0; |
1154 | 2.40M | pu1_incr++; |
1155 | 2.40M | } |
1156 | | |
1157 | | /* Updates */ |
1158 | 724k | i4_pos_y += i4_trans_size; |
1159 | 724k | pu1_incr_y += i4_trans_size * i4_refary_wd; |
1160 | 724k | i4_y_idx += MIN(i4_trans_size, (i4_act_ary_ht - i4_y_idx)); |
1161 | 724k | } |
1162 | | |
1163 | | /* Loop updates */ |
1164 | 872k | if(i4_x < i4_act_ary_wd) |
1165 | 436k | { |
1166 | 436k | i4_y = i4_prev_y; |
1167 | 436k | } |
1168 | 436k | else if(i4_y < i4_act_ary_ht) |
1169 | 218k | { |
1170 | 218k | i4_prev_y = i4_y; |
1171 | 218k | i4_x = 0; |
1172 | 218k | } |
1173 | | |
1174 | 872k | } while((i4_y < i4_act_ary_ht) || (i4_x < i4_act_ary_wd)); |
1175 | | |
1176 | 218k | } /* End of chroma */ |
1177 | 327k | return OK; |
1178 | | |
1179 | 327k | } /* End of "isvcd_ref_layer_ptr_incr" */ |
1180 | | |
1181 | | void isvcd_residual_reflayer_const_non_boundary_mb( |
1182 | | WORD16 *pi2_inp_data, WORD32 i4_inp_data_stride, WORD16 *pi2_ref_array, WORD32 i4_refarray_wd, |
1183 | | WORD32 i4_refarray_ht, WORD32 i4_ref_mb_type_q0, WORD32 i4_ref_mb_type_q1, |
1184 | | WORD32 i4_ref_mb_type_q2, WORD32 i4_ref_mb_type_q3, WORD32 i4_mb_quard1_part_x, |
1185 | | WORD32 i4_mb_quard1_part_y, WORD32 i4_chroma_flag) |
1186 | 49.2k | { |
1187 | 49.2k | WORD32 i4_x_ref, i4_y_ref; |
1188 | 49.2k | WORD32 i4_x, i4_y; |
1189 | 49.2k | WORD32 i4_ref_mb_type; |
1190 | 49.2k | WORD16 *pi2_ref_data_byte; |
1191 | 49.2k | WORD16 *pi2_ref_array_temp; |
1192 | | |
1193 | 461k | for(i4_y = 0; i4_y < i4_refarray_ht; i4_y++) |
1194 | 412k | { |
1195 | 4.06M | for(i4_x = 0; i4_x < i4_refarray_wd; i4_x++) |
1196 | 3.65M | { |
1197 | 3.65M | i4_y_ref = i4_y; |
1198 | 3.65M | i4_x_ref = i4_x; |
1199 | | |
1200 | 3.65M | i4_ref_mb_type = i4_ref_mb_type_q0; |
1201 | 3.65M | if(i4_x >= i4_mb_quard1_part_x) |
1202 | 1.82M | { |
1203 | 1.82M | if(i4_y < i4_mb_quard1_part_y) |
1204 | 931k | { |
1205 | 931k | i4_ref_mb_type = i4_ref_mb_type_q1; |
1206 | 931k | } |
1207 | 896k | else if(i4_y >= i4_mb_quard1_part_y) |
1208 | 896k | { |
1209 | 896k | i4_ref_mb_type = i4_ref_mb_type_q3; |
1210 | 896k | } |
1211 | 1.82M | } |
1212 | 1.82M | else if(i4_x < i4_mb_quard1_part_x) |
1213 | 1.82M | { |
1214 | 1.82M | if(i4_y >= i4_mb_quard1_part_y) |
1215 | 896k | { |
1216 | 896k | i4_ref_mb_type = i4_ref_mb_type_q2; |
1217 | 896k | } |
1218 | 1.82M | } |
1219 | | |
1220 | | /****************************************************************/ |
1221 | | /* Reference layer Residual Buffer is maintained as 8-bit data */ |
1222 | | /* Buffer and 1-bit sign bit packed buffer. Sign Byte is read */ |
1223 | | /* and sign of the data sample is extracted depending upon bit */ |
1224 | | /* postition. */ |
1225 | | /****************************************************************/ |
1226 | | |
1227 | | /* update the buffer pointers to appropriate locations */ |
1228 | 3.65M | pi2_ref_array_temp = pi2_ref_array + i4_x; |
1229 | 3.65M | pi2_ref_array_temp += i4_y * i4_refarray_wd; |
1230 | | |
1231 | | /* extract the residual value and fill the buffer */ |
1232 | 3.65M | if(SVC_INTER_MB == i4_ref_mb_type) |
1233 | 3.27M | { |
1234 | | /* derive the reference data pointers */ |
1235 | 3.27M | pi2_ref_data_byte = pi2_inp_data + (i4_x_ref << i4_chroma_flag); |
1236 | 3.27M | pi2_ref_data_byte += i4_y_ref * i4_inp_data_stride; |
1237 | | |
1238 | | /* store the residual value */ |
1239 | 3.27M | *pi2_ref_array_temp = (WORD16) (*pi2_ref_data_byte); |
1240 | 3.27M | } |
1241 | 378k | else |
1242 | 378k | { |
1243 | | /* if non inter MB then store the 0 */ |
1244 | 378k | *pi2_ref_array_temp = 0; |
1245 | 378k | } |
1246 | 3.65M | } |
1247 | 412k | } |
1248 | 49.2k | } |
1249 | | |
1250 | | void isvcd_residual_reflayer_const_boundary_mb(WORD16 *pi2_inp_data, WORD32 i4_inp_data_stride, |
1251 | | WORD16 *pi2_ref_array, WORD32 i4_refarray_wd, |
1252 | | WORD32 i4_refarray_ht, WORD32 i4_ref_wd, |
1253 | | WORD32 i4_ref_ht, WORD32 i4_x_offset, |
1254 | | WORD32 i4_y_offset, WORD32 i4_ref_mb_type_q0, |
1255 | | WORD32 i4_ref_mb_type_q1, WORD32 i4_ref_mb_type_q2, |
1256 | | WORD32 i4_ref_mb_type_q3, WORD32 i4_mb_quard1_part_x, |
1257 | | WORD32 i4_mb_quard1_part_y, WORD32 i4_chroma_flag) |
1258 | 238k | { |
1259 | 238k | WORD32 i4_x_ref, i4_y_ref; |
1260 | 238k | WORD32 i4_x, i4_y; |
1261 | 238k | WORD16 *pi2_ref_data_byte; |
1262 | 238k | WORD16 *pi2_ref_array_temp; |
1263 | | |
1264 | | /*Quard 0*/ |
1265 | 1.63M | for(i4_y = 0; i4_y < i4_mb_quard1_part_y; i4_y++) |
1266 | 1.39M | { |
1267 | 14.3M | for(i4_x = 0; i4_x < i4_mb_quard1_part_x; i4_x++) |
1268 | 12.9M | { |
1269 | 12.9M | i4_y_ref = MAX(0, MIN(i4_ref_ht - 1, i4_y + i4_y_offset)); |
1270 | 12.9M | i4_x_ref = MAX(0, MIN(i4_ref_wd - 1, i4_x + i4_x_offset)); |
1271 | | |
1272 | | /****************************************************************/ |
1273 | | /* Reference layer Residual Buffer is maintained as 8-bit data */ |
1274 | | /* Buffer and 1-bit sign bit packed buffer. Sign Byte is read */ |
1275 | | /* and sign of the data sample is extracted depending upon bit */ |
1276 | | /* postition. */ |
1277 | | /****************************************************************/ |
1278 | | |
1279 | | /* update the buffer pointers to appropriate locations */ |
1280 | 12.9M | pi2_ref_array_temp = pi2_ref_array + i4_x; |
1281 | 12.9M | pi2_ref_array_temp += i4_y * i4_refarray_wd; |
1282 | | |
1283 | | /* extract the residual value and fill the buffer */ |
1284 | 12.9M | if(SVC_INTER_MB == i4_ref_mb_type_q0) |
1285 | 12.5M | { |
1286 | | /* input pointer will be pointing to (xoffset,yoffset) */ |
1287 | | /* So subtract the correction to reference location */ |
1288 | 12.5M | if(0 <= i4_x_offset) |
1289 | 5.58M | { |
1290 | | /* if only inside frame dimension */ |
1291 | 5.58M | i4_x_ref = i4_x_ref - i4_x_offset; |
1292 | 5.58M | } |
1293 | | |
1294 | 12.5M | if(0 <= i4_y_offset) |
1295 | 8.19M | { |
1296 | | /* if only inside frame dimension */ |
1297 | 8.19M | i4_y_ref = i4_y_ref - i4_y_offset; |
1298 | 8.19M | } |
1299 | | /* derive the reference data pointers */ |
1300 | | |
1301 | 12.5M | pi2_ref_data_byte = pi2_inp_data + (i4_x_ref << i4_chroma_flag); |
1302 | 12.5M | pi2_ref_data_byte += i4_y_ref * i4_inp_data_stride; |
1303 | | |
1304 | | /* store the residual value */ |
1305 | 12.5M | *pi2_ref_array_temp = (WORD16) (*pi2_ref_data_byte); |
1306 | 12.5M | } |
1307 | 362k | else |
1308 | 362k | { |
1309 | | /* if non inter MB then store the 0 */ |
1310 | 362k | *pi2_ref_array_temp = 0; |
1311 | 362k | } |
1312 | 12.9M | } |
1313 | 1.39M | } |
1314 | | |
1315 | | /*Quard 1*/ |
1316 | 1.63M | for(i4_y = 0; i4_y < i4_mb_quard1_part_y; i4_y++) |
1317 | 1.39M | { |
1318 | 3.02M | for(i4_x = i4_mb_quard1_part_x; i4_x < i4_refarray_wd; i4_x++) |
1319 | 1.63M | { |
1320 | 1.63M | i4_y_ref = MAX(0, MIN(i4_ref_ht - 1, i4_y + i4_y_offset)); |
1321 | 1.63M | i4_x_ref = MAX(0, MIN(i4_ref_wd - 1, i4_x + i4_x_offset)); |
1322 | | |
1323 | | /****************************************************************/ |
1324 | | /* Reference layer Residual Buffer is maintained as 8-bit data */ |
1325 | | /* Buffer and 1-bit sign bit packed buffer. Sign Byte is read */ |
1326 | | /* and sign of the data sample is extracted depending upon bit */ |
1327 | | /* postition. */ |
1328 | | /****************************************************************/ |
1329 | | |
1330 | | /* update the buffer pointers to appropriate locations */ |
1331 | 1.63M | pi2_ref_array_temp = pi2_ref_array + i4_x; |
1332 | 1.63M | pi2_ref_array_temp += i4_y * i4_refarray_wd; |
1333 | | |
1334 | | /* extract the residual value and fill the buffer */ |
1335 | 1.63M | if(SVC_INTER_MB == i4_ref_mb_type_q1) |
1336 | 1.51M | { |
1337 | | /* input pointer will be pointing to (xoffset,yoffset) */ |
1338 | | /* So subtract the correction to reference location */ |
1339 | 1.51M | if(0 <= i4_x_offset) |
1340 | 1.51M | { |
1341 | | /* if only inside frame dimension */ |
1342 | 1.51M | i4_x_ref = i4_x_ref - i4_x_offset; |
1343 | 1.51M | } |
1344 | 1.51M | if(0 <= i4_y_offset) |
1345 | 582k | { |
1346 | | /* if only inside frame dimension */ |
1347 | 582k | i4_y_ref = i4_y_ref - i4_y_offset; |
1348 | 582k | } |
1349 | | /* derive the reference data pointers */ |
1350 | | |
1351 | 1.51M | pi2_ref_data_byte = pi2_inp_data + (i4_x_ref << i4_chroma_flag); |
1352 | 1.51M | pi2_ref_data_byte += i4_y_ref * i4_inp_data_stride; |
1353 | | |
1354 | | /* store the residual value */ |
1355 | 1.51M | *pi2_ref_array_temp = (WORD16) (*pi2_ref_data_byte); |
1356 | 1.51M | } |
1357 | 110k | else |
1358 | 110k | { |
1359 | | /* if non inter MB then store the 0 */ |
1360 | 110k | *pi2_ref_array_temp = 0; |
1361 | 110k | } |
1362 | 1.63M | } |
1363 | 1.39M | } |
1364 | | |
1365 | | /*Quard 2*/ |
1366 | 961k | for(i4_y = i4_mb_quard1_part_y; i4_y < i4_refarray_ht; i4_y++) |
1367 | 722k | { |
1368 | 7.99M | for(i4_x = 0; i4_x < i4_mb_quard1_part_x; i4_x++) |
1369 | 7.27M | { |
1370 | 7.27M | i4_y_ref = MAX(0, MIN(i4_ref_ht - 1, i4_y + i4_y_offset)); |
1371 | 7.27M | i4_x_ref = MAX(0, MIN(i4_ref_wd - 1, i4_x + i4_x_offset)); |
1372 | | |
1373 | | /****************************************************************/ |
1374 | | /* Reference layer Residual Buffer is maintained as 8-bit data */ |
1375 | | /* Buffer and 1-bit sign bit packed buffer. Sign Byte is read */ |
1376 | | /* and sign of the data sample is extracted depending upon bit */ |
1377 | | /* postition. */ |
1378 | | /****************************************************************/ |
1379 | | |
1380 | | /* update the buffer pointers to appropriate locations */ |
1381 | 7.27M | pi2_ref_array_temp = pi2_ref_array + i4_x; |
1382 | 7.27M | pi2_ref_array_temp += i4_y * i4_refarray_wd; |
1383 | | |
1384 | | /* extract the residual value and fill the buffer */ |
1385 | 7.27M | if(SVC_INTER_MB == i4_ref_mb_type_q2) |
1386 | 7.06M | { |
1387 | | /* input pointer will be pointing to (xoffset,yoffset) */ |
1388 | | /* So subtract the correction to reference location */ |
1389 | 7.06M | if(0 <= i4_x_offset) |
1390 | 2.85M | { |
1391 | | /* if only inside frame dimension */ |
1392 | 2.85M | i4_x_ref = i4_x_ref - i4_x_offset; |
1393 | 2.85M | } |
1394 | 7.06M | if(0 <= i4_y_offset) |
1395 | 7.06M | { |
1396 | | /* if only inside frame dimension */ |
1397 | 7.06M | i4_y_ref = i4_y_ref - i4_y_offset; |
1398 | 7.06M | } |
1399 | | /* derive the reference data pointers */ |
1400 | 7.06M | pi2_ref_data_byte = pi2_inp_data + (i4_x_ref << i4_chroma_flag); |
1401 | 7.06M | pi2_ref_data_byte += i4_y_ref * i4_inp_data_stride; |
1402 | | |
1403 | | /* store the residual value */ |
1404 | 7.06M | *pi2_ref_array_temp = (WORD16) (*pi2_ref_data_byte); |
1405 | 7.06M | } |
1406 | 208k | else |
1407 | 208k | { |
1408 | | /* if non inter MB then store the 0 */ |
1409 | 208k | *pi2_ref_array_temp = 0; |
1410 | 208k | } |
1411 | 7.27M | } |
1412 | 722k | } |
1413 | | |
1414 | | /*Quard 3*/ |
1415 | 961k | for(i4_y = i4_mb_quard1_part_y; i4_y < i4_refarray_ht; i4_y++) |
1416 | 722k | { |
1417 | 1.10M | for(i4_x = i4_mb_quard1_part_x; i4_x < i4_refarray_wd; i4_x++) |
1418 | 379k | { |
1419 | 379k | i4_y_ref = MAX(0, MIN(i4_ref_ht - 1, i4_y + i4_y_offset)); |
1420 | 379k | i4_x_ref = MAX(0, MIN(i4_ref_wd - 1, i4_x + i4_x_offset)); |
1421 | | |
1422 | | /****************************************************************/ |
1423 | | /* Reference layer Residual Buffer is maintained as 8-bit data */ |
1424 | | /* Buffer and 1-bit sign bit packed buffer. Sign Byte is read */ |
1425 | | /* and sign of the data sample is extracted depending upon bit */ |
1426 | | /* postition. */ |
1427 | | /****************************************************************/ |
1428 | | |
1429 | | /* update the buffer pointers to appropriate locations */ |
1430 | 379k | pi2_ref_array_temp = pi2_ref_array + i4_x; |
1431 | 379k | pi2_ref_array_temp += i4_y * i4_refarray_wd; |
1432 | | |
1433 | | /* extract the residual value and fill the buffer */ |
1434 | 379k | if(SVC_INTER_MB == i4_ref_mb_type_q3) |
1435 | 379k | { |
1436 | | /* input pointer will be pointing to (xoffset,yoffset) */ |
1437 | | /* So subtract the correction to reference location */ |
1438 | 379k | if(0 <= i4_x_offset) |
1439 | 379k | { |
1440 | | /* if only inside frame dimension */ |
1441 | 379k | i4_x_ref = i4_x_ref - i4_x_offset; |
1442 | 379k | } |
1443 | 379k | if(0 <= i4_y_offset) |
1444 | 379k | { |
1445 | | /* if only inside frame dimension */ |
1446 | 379k | i4_y_ref = i4_y_ref - i4_y_offset; |
1447 | 379k | } |
1448 | | /* derive the reference data pointers */ |
1449 | 379k | pi2_ref_data_byte = pi2_inp_data + (i4_x_ref << i4_chroma_flag); |
1450 | 379k | pi2_ref_data_byte += i4_y_ref * i4_inp_data_stride; |
1451 | | |
1452 | | /* store the residual value */ |
1453 | 379k | *pi2_ref_array_temp = (WORD16) (*pi2_ref_data_byte); |
1454 | 379k | } |
1455 | 0 | else |
1456 | 0 | { |
1457 | | /* if non inter MB then store the 0 */ |
1458 | 0 | *pi2_ref_array_temp = 0; |
1459 | 0 | } |
1460 | 379k | } |
1461 | 722k | } |
1462 | 238k | } |
1463 | | |
1464 | | /*****************************************************************************/ |
1465 | | /* */ |
1466 | | /* Function Name : isvcd_residual_reflayer_const */ |
1467 | | /* */ |
1468 | | /* Description : This function constructs the reference array buffer */ |
1469 | | /* used for residual resampling of a component in an MB */ |
1470 | | /* */ |
1471 | | /* Inputs : pv_residual_samp_ctxt: intra sampling context */ |
1472 | | /* pu1_inp : input (reference layer data) */ |
1473 | | /* i4_inp_stride : input buffer stride */ |
1474 | | /* pu1_inp_bitmap : input (reference layer sign bits) */ |
1475 | | /* i4_inp_stride : input buffer stride */ |
1476 | | /* ps_ref_mb_mode : ref layer mb mode buffer desc */ |
1477 | | /* pi4_refarr_wd : pointer to store the reference array WD */ |
1478 | | /* pi4_refarr_ht : pointer to store the reference array HT */ |
1479 | | /* pi4_x_offset : pointer to store the reference X offset */ |
1480 | | /* pi4_y_offset : pointer to store the reference Y offset */ |
1481 | | /* ps_coord : mb co-ordinate structure */ |
1482 | | /* i4_chroma_flag : chroma processing flag */ |
1483 | | /* Globals : none */ |
1484 | | /* Processing : it fills the reference layer data if they are falling in */ |
1485 | | /* INTRA MB region. If all the pixels are not filled it */ |
1486 | | /* calls the border extension algorithm to fill them */ |
1487 | | /* Outputs : none */ |
1488 | | /* Returns : none */ |
1489 | | /* */ |
1490 | | /* Issues : none */ |
1491 | | /* */ |
1492 | | /* Revision History: */ |
1493 | | /* */ |
1494 | | /* DD MM YYYY Author(s) Changes (Describe the changes made) */ |
1495 | | /* 06 07 2021 vijayakumar creation */ |
1496 | | /* */ |
1497 | | /*****************************************************************************/ |
1498 | | WORD32 isvcd_residual_reflayer_const(void *pv_residual_samp_ctxt, WORD16 *pi2_inp_data, |
1499 | | WORD32 i4_inp_data_stride, mem_element_t *ps_ref_mb_mode, |
1500 | | WORD32 *pi4_refarr_wd, mb_coord_t *ps_coord, |
1501 | | WORD32 i4_chroma_flag) |
1502 | 327k | { |
1503 | 327k | residual_sampling_ctxt_t *ps_ctxt; |
1504 | 327k | res_lyr_ctxt *ps_lyr_ctxt; |
1505 | 327k | WORD8 *pi1_ref_mb_modes; |
1506 | 327k | WORD32 i4_ref_mode_stride; |
1507 | 327k | WORD32 i4_element_size; |
1508 | 327k | WORD32 i4_ref_wd; |
1509 | 327k | WORD32 i4_ref_ht; |
1510 | 327k | WORD32 i4_x_offset; |
1511 | 327k | WORD32 i4_y_offset; |
1512 | 327k | WORD32 i4_refarray_wd; |
1513 | 327k | WORD32 i4_refarray_ht; |
1514 | 327k | WORD8 i1_edge_mb; |
1515 | 327k | WORD16 *pi2_ref_array; |
1516 | 327k | WORD32 i4_mb_sft; |
1517 | 327k | WORD32 i4_mb_x, i4_mb_y; |
1518 | 327k | WORD32 i4_mb_x_strt, i4_mb_y_strt; |
1519 | 327k | WORD32 i4_mb_quard1_part_x, i4_mb_quard1_part_y; |
1520 | 327k | WORD8 *pi1_ref_mb_modes_incr; |
1521 | 327k | WORD8 *pi1_ref_mb_modes_incr_temp; |
1522 | 327k | inter_lyr_mb_prms_t *ps_inter_lyr_mb_prms; |
1523 | | |
1524 | 327k | WORD32 i4_mb_wd; |
1525 | 327k | WORD32 i4_mb_ht; |
1526 | 327k | WORD32 i4_x_ref, i4_y_ref; |
1527 | 327k | WORD32 i4_ref_mb_type_q0, i4_ref_mb_type_q1, i4_ref_mb_type_q2, i4_ref_mb_type_q3; |
1528 | 327k | WORD8 i1_mb_mode_q0, i1_mb_mode_q1, i1_mb_mode_q2, i1_mb_mode_q3; |
1529 | 327k | WORD32 ret; |
1530 | | |
1531 | 327k | ps_ctxt = (residual_sampling_ctxt_t *) pv_residual_samp_ctxt; |
1532 | 327k | ps_lyr_ctxt = &ps_ctxt->as_res_lyrs[ps_ctxt->i4_res_lyr_id]; |
1533 | 327k | pi2_ref_array = ps_ctxt->pi2_refarray_buffer; |
1534 | | |
1535 | 327k | pi1_ref_mb_modes = (WORD8 *) ps_ref_mb_mode->pv_buffer; |
1536 | 327k | i4_ref_mode_stride = ps_ref_mb_mode->i4_num_element_stride; |
1537 | 327k | i4_element_size = ps_ref_mb_mode->i4_element_size; |
1538 | | |
1539 | 327k | if(NULL == pi1_ref_mb_modes) |
1540 | 0 | { |
1541 | 0 | return NOT_OK; |
1542 | 0 | } |
1543 | | |
1544 | 327k | i4_mb_wd = MB_WIDTH >> i4_chroma_flag; |
1545 | 327k | i4_mb_ht = MB_HEIGHT >> i4_chroma_flag; |
1546 | | |
1547 | | /* ----------------------------------------------------------------- */ |
1548 | | /* Deriving the parameters required for further processing */ |
1549 | | /* ----------------------------------------------------------------- */ |
1550 | 327k | { |
1551 | 327k | ref_mb_map_t *ps_x_off_len; |
1552 | 327k | ref_mb_map_t *ps_y_off_len; |
1553 | 327k | WORD32 i4_mbaddr_x; |
1554 | 327k | WORD32 i4_mbaddr_y; |
1555 | 327k | WORD32 i4_base_width; |
1556 | 327k | WORD32 i4_base_height; |
1557 | 327k | residual_samp_map_ctxt_t *ps_map_ctxt; |
1558 | | |
1559 | 327k | if(1 == i4_chroma_flag) |
1560 | 218k | ps_map_ctxt = &ps_lyr_ctxt->s_chroma_map_ctxt; |
1561 | 109k | else |
1562 | 109k | ps_map_ctxt = &ps_lyr_ctxt->s_luma_map_ctxt; |
1563 | | |
1564 | 327k | i4_mbaddr_y = ps_coord->u2_mb_y; |
1565 | 327k | i4_mbaddr_x = ps_coord->u2_mb_x; |
1566 | 327k | i4_base_width = ps_lyr_ctxt->i4_ref_width; |
1567 | 327k | i4_base_height = ps_lyr_ctxt->i4_ref_height; |
1568 | 327k | i4_ref_wd = i4_base_width >> i4_chroma_flag; |
1569 | 327k | i4_ref_ht = i4_base_height >> i4_chroma_flag; |
1570 | | |
1571 | | /* --------------------------------------------------------------------- */ |
1572 | | /* Extracting information from the mapping context */ |
1573 | | /* --------------------------------------------------------------------- */ |
1574 | 327k | ps_x_off_len = ps_map_ctxt->ps_x_offset_length; |
1575 | 327k | ps_y_off_len = ps_map_ctxt->ps_y_offset_length; |
1576 | 327k | i4_x_offset = ps_x_off_len[i4_mbaddr_x].i2_offset; |
1577 | 327k | i4_y_offset = ps_y_off_len[i4_mbaddr_y].i2_offset; |
1578 | 327k | i4_refarray_wd = ps_x_off_len[i4_mbaddr_x].i2_length; |
1579 | 327k | i4_refarray_ht = ps_y_off_len[i4_mbaddr_y].i2_length; |
1580 | 327k | } |
1581 | | |
1582 | | /* Call the module to fill the increments based on transform blocks */ |
1583 | 327k | ret = isvcd_ref_layer_ptr_incr(pi1_ref_mb_modes, i4_ref_mode_stride, i4_element_size, |
1584 | 327k | i4_x_offset, i4_y_offset, i4_refarray_wd, i4_refarray_ht, |
1585 | 327k | ps_ctxt->pu1_ref_x_ptr_incr, ps_ctxt->pu1_ref_y_ptr_incr, |
1586 | 327k | i4_chroma_flag); |
1587 | | |
1588 | 327k | if(ret != OK) |
1589 | 0 | { |
1590 | 0 | return ret; |
1591 | 0 | } |
1592 | 327k | i4_mb_sft = (MB_WIDTH_SHIFT - i4_chroma_flag); |
1593 | | |
1594 | | /* --------------------------------------------------------------------- */ |
1595 | | /* MB Level Resampling for the MB - Pointers sent for MB in both layers */ |
1596 | | /* This has been written according to the dyadic case */ |
1597 | | /* --------------------------------------------------------------------- */ |
1598 | 327k | i4_y_ref = MAX(0, MIN(i4_ref_ht - 1, 0 + i4_y_offset)); |
1599 | 327k | i4_x_ref = MAX(0, MIN(i4_ref_wd - 1, 0 + i4_x_offset)); |
1600 | 327k | i4_mb_x_strt = i4_x_ref % i4_mb_wd; |
1601 | 327k | i4_mb_y_strt = i4_y_ref % i4_mb_ht; |
1602 | | |
1603 | 327k | i4_mb_quard1_part_x = i4_mb_wd - i4_mb_x_strt; |
1604 | 327k | i4_mb_quard1_part_y = i4_mb_ht - i4_mb_y_strt; |
1605 | 327k | if(!(i4_mb_quard1_part_x >= 0)) |
1606 | 0 | { |
1607 | 0 | return NOT_OK; |
1608 | 0 | } |
1609 | 327k | if(!(i4_mb_quard1_part_y >= 0)) |
1610 | 0 | { |
1611 | 0 | return NOT_OK; |
1612 | 0 | } |
1613 | | |
1614 | 327k | i4_mb_x = (i4_x_ref >> i4_mb_sft); |
1615 | 327k | i4_mb_y = (i4_y_ref >> i4_mb_sft); |
1616 | | |
1617 | | /* get the location of the byte which has the current mb mode */ |
1618 | 327k | pi1_ref_mb_modes_incr = pi1_ref_mb_modes + (i4_mb_y * i4_ref_mode_stride * i4_element_size); |
1619 | 327k | pi1_ref_mb_modes_incr += (i4_mb_x * i4_element_size); |
1620 | 327k | ps_inter_lyr_mb_prms = (inter_lyr_mb_prms_t *) pi1_ref_mb_modes_incr; |
1621 | 327k | i1_mb_mode_q0 = ps_inter_lyr_mb_prms->i1_mb_mode; |
1622 | 327k | i1_mb_mode_q1 = i1_mb_mode_q0; |
1623 | 327k | i1_mb_mode_q2 = i1_mb_mode_q0; |
1624 | 327k | i1_mb_mode_q3 = i1_mb_mode_q0; |
1625 | | |
1626 | 327k | pi1_ref_mb_modes_incr_temp = pi1_ref_mb_modes_incr; |
1627 | 327k | if(i4_mb_quard1_part_x > 0) |
1628 | 327k | { |
1629 | 327k | pi1_ref_mb_modes_incr_temp = pi1_ref_mb_modes_incr + i4_element_size; |
1630 | 327k | ps_inter_lyr_mb_prms = (inter_lyr_mb_prms_t *) pi1_ref_mb_modes_incr_temp; |
1631 | 327k | i1_mb_mode_q1 = ps_inter_lyr_mb_prms->i1_mb_mode; |
1632 | 327k | } |
1633 | | |
1634 | 327k | if(i4_mb_quard1_part_y > 0) |
1635 | 327k | { |
1636 | 327k | pi1_ref_mb_modes_incr_temp = pi1_ref_mb_modes_incr + (i4_ref_mode_stride * i4_element_size); |
1637 | 327k | ps_inter_lyr_mb_prms = (inter_lyr_mb_prms_t *) pi1_ref_mb_modes_incr_temp; |
1638 | 327k | i1_mb_mode_q2 = ps_inter_lyr_mb_prms->i1_mb_mode; |
1639 | 327k | } |
1640 | | |
1641 | 327k | if((i4_mb_quard1_part_x > 0) && (i4_mb_quard1_part_y > 0)) |
1642 | 327k | { |
1643 | 327k | pi1_ref_mb_modes_incr_temp = |
1644 | 327k | pi1_ref_mb_modes_incr + (i4_ref_mode_stride * i4_element_size) + i4_element_size; |
1645 | 327k | ps_inter_lyr_mb_prms = (inter_lyr_mb_prms_t *) pi1_ref_mb_modes_incr_temp; |
1646 | 327k | i1_mb_mode_q3 = ps_inter_lyr_mb_prms->i1_mb_mode; |
1647 | 327k | } |
1648 | | |
1649 | 327k | i4_ref_mb_type_q0 = (i1_mb_mode_q0 <= SVC_INTER_MB) ? SVC_INTER_MB : SVC_INTRA_MB; |
1650 | 327k | i4_ref_mb_type_q1 = (i1_mb_mode_q1 <= SVC_INTER_MB) ? SVC_INTER_MB : SVC_INTRA_MB; |
1651 | 327k | i4_ref_mb_type_q2 = (i1_mb_mode_q2 <= SVC_INTER_MB) ? SVC_INTER_MB : SVC_INTRA_MB; |
1652 | 327k | i4_ref_mb_type_q3 = (i1_mb_mode_q3 <= SVC_INTER_MB) ? SVC_INTER_MB : SVC_INTRA_MB; |
1653 | | |
1654 | 327k | i1_edge_mb = (ps_coord->u2_mb_x == 0 || ps_coord->u2_mb_y == 0 || |
1655 | 182k | (ps_coord->u2_mb_x == ((ps_lyr_ctxt->i4_curr_width >> MB_WIDTH_SHIFT) - 1)) || |
1656 | 95.6k | (ps_coord->u2_mb_y == ((ps_lyr_ctxt->i4_curr_height >> MB_HEIGHT_SHIFT) - 1))); |
1657 | 327k | if(i1_edge_mb) |
1658 | 238k | { |
1659 | 238k | ps_ctxt->pf_residual_reflayer_const_boundary_mb( |
1660 | 238k | pi2_inp_data, i4_inp_data_stride, pi2_ref_array, i4_refarray_wd, i4_refarray_ht, |
1661 | 238k | i4_ref_wd, i4_ref_ht, i4_x_offset, i4_y_offset, i4_ref_mb_type_q0, i4_ref_mb_type_q1, |
1662 | 238k | i4_ref_mb_type_q2, i4_ref_mb_type_q3, i4_mb_quard1_part_x, i4_mb_quard1_part_y, |
1663 | 238k | i4_chroma_flag); |
1664 | 238k | } |
1665 | 88.4k | else |
1666 | 88.4k | { |
1667 | 88.4k | ps_ctxt->pf_residual_reflayer_const_non_boundary_mb( |
1668 | 88.4k | pi2_inp_data, i4_inp_data_stride, pi2_ref_array, i4_refarray_wd, i4_refarray_ht, |
1669 | 88.4k | i4_ref_mb_type_q0, i4_ref_mb_type_q1, i4_ref_mb_type_q2, i4_ref_mb_type_q3, |
1670 | 88.4k | i4_mb_quard1_part_x, i4_mb_quard1_part_y, i4_chroma_flag); |
1671 | 88.4k | } |
1672 | | /* store the values into the place holders */ |
1673 | 327k | *pi4_refarr_wd = i4_refarray_wd; |
1674 | | |
1675 | 327k | return OK; |
1676 | 327k | } |
1677 | | |
1678 | | /*****************************************************************************/ |
1679 | | /* */ |
1680 | | /* Function Name : isvcd_interpolate_residual */ |
1681 | | /* */ |
1682 | | /* Description : This function takes the refernce array buffer and perform*/ |
1683 | | /* interpolation of a component to find the residual */ |
1684 | | /* resampled value */ |
1685 | | /* Inputs : pv_residual_samp_ctxt : residual sampling context */ |
1686 | | /* pi2_out : output buffer pointer */ |
1687 | | /* i4_out_stride : output buffer stride */ |
1688 | | /* i4_refarray_wd : reference array width */ |
1689 | | /* i4_x_offset : offset in reference layer in horz direction*/ |
1690 | | /* ps_coord : current mb co-ordinate */ |
1691 | | /* i4_chroma_flag : chroma processing flag */ |
1692 | | /* Globals : none */ |
1693 | | /* Processing : it does the interpolation in vertical direction followed */ |
1694 | | /* by horizontal direction */ |
1695 | | /* Outputs : resampled pixels */ |
1696 | | /* Returns : none */ |
1697 | | /* */ |
1698 | | /* Issues : none */ |
1699 | | /* */ |
1700 | | /* Revision History: */ |
1701 | | /* */ |
1702 | | /* DD MM YYYY Author(s) Changes (Describe the changes made) */ |
1703 | | /* 06 07 2021 vijayakumar creation */ |
1704 | | /* */ |
1705 | | /*****************************************************************************/ |
1706 | | void isvcd_interpolate_residual(void *pv_residual_samp_ctxt, WORD16 *pi2_out, WORD32 i4_out_stride, |
1707 | | WORD32 i4_refarray_wd, UWORD16 u2_mb_x, UWORD16 u2_mb_y, |
1708 | | WORD32 i4_chroma_flag) |
1709 | 180k | { |
1710 | 180k | residual_sampling_ctxt_t *ps_ctxt; |
1711 | 180k | residual_samp_map_ctxt_t *ps_map_ctxt; |
1712 | 180k | res_lyr_ctxt *ps_lyr_ctxt; |
1713 | 180k | ref_pixel_map_t *ps_x_pos_phase; |
1714 | 180k | ref_pixel_map_t *ps_y_pos_phase; |
1715 | | |
1716 | 180k | WORD32 i4_x, i4_y; |
1717 | 180k | WORD32 i4_frm_mb_x, i4_frm_mb_y; |
1718 | 180k | WORD32 i4_temp_array_ht; |
1719 | 180k | WORD32 i4_mb_wd; |
1720 | 180k | WORD32 i4_mb_ht; |
1721 | 180k | WORD16 *pi2_ref_array; |
1722 | 180k | UWORD8 *pu1_ref_x_ptr_incr, *pu1_ref_y_ptr_incr; |
1723 | | |
1724 | 180k | ps_ctxt = (residual_sampling_ctxt_t *) pv_residual_samp_ctxt; |
1725 | 180k | ps_lyr_ctxt = &ps_ctxt->as_res_lyrs[ps_ctxt->i4_res_lyr_id]; |
1726 | 180k | pi2_ref_array = ps_ctxt->pi2_refarray_buffer; |
1727 | 180k | pu1_ref_x_ptr_incr = ps_ctxt->pu1_ref_x_ptr_incr; |
1728 | 180k | pu1_ref_y_ptr_incr = ps_ctxt->pu1_ref_y_ptr_incr; |
1729 | | |
1730 | | /* --------------------------------------------------------------------- */ |
1731 | | /* Extracting information from the mapping context */ |
1732 | | /* --------------------------------------------------------------------- */ |
1733 | 180k | if(1 == i4_chroma_flag) |
1734 | 120k | ps_map_ctxt = &ps_lyr_ctxt->s_chroma_map_ctxt; |
1735 | 60.1k | else |
1736 | 60.1k | ps_map_ctxt = &ps_lyr_ctxt->s_luma_map_ctxt; |
1737 | | |
1738 | 180k | i4_mb_wd = MB_WIDTH >> i4_chroma_flag; |
1739 | 180k | i4_mb_ht = MB_HEIGHT >> i4_chroma_flag; |
1740 | | |
1741 | 180k | ps_x_pos_phase = ps_map_ctxt->ps_x_pos_phase; |
1742 | 180k | ps_y_pos_phase = ps_map_ctxt->ps_y_pos_phase; |
1743 | 180k | i4_temp_array_ht = i4_mb_ht; |
1744 | 180k | i4_frm_mb_y = u2_mb_y * i4_mb_ht; |
1745 | 180k | i4_frm_mb_x = u2_mb_x * i4_mb_wd; |
1746 | | |
1747 | | /* --------------------------------------------------------------------- */ |
1748 | | /* Loop for interpolation */ |
1749 | | /* --------------------------------------------------------------------- */ |
1750 | 2.10M | for(i4_y = 0; i4_y < (i4_temp_array_ht); i4_y++) |
1751 | 1.92M | { |
1752 | 25.0M | for(i4_x = 0; i4_x < (i4_mb_wd); i4_x++) |
1753 | 23.1M | { |
1754 | 23.1M | WORD32 i4_i; |
1755 | 23.1M | WORD32 i4_y_ref; |
1756 | 23.1M | WORD32 i4_y_phase; |
1757 | 23.1M | WORD32 i4_x_ref; |
1758 | 23.1M | WORD32 i4_x_phase; |
1759 | 23.1M | WORD32 i4_x_ref_round; |
1760 | 23.1M | WORD16 *pi2_out_curr; |
1761 | 23.1M | WORD32 ai4_temp_pred[2] = {0}; |
1762 | 23.1M | UWORD8 *pu1_ref_y_ptr_incr_temp; |
1763 | 23.1M | WORD32 *pi4_temp_pred; |
1764 | 23.1M | UWORD8 u1_incr_y; |
1765 | 23.1M | WORD16 i2_res; |
1766 | | |
1767 | | /* derive the current output pointer */ |
1768 | 23.1M | pi2_out_curr = pi2_out + (i4_x << i4_chroma_flag) + (i4_y * i4_out_stride); |
1769 | | |
1770 | | /* -------------------------------------------------------------- */ |
1771 | | /* Finding the offset */ |
1772 | | /* -------------------------------------------------------------- */ |
1773 | 23.1M | i4_y_ref = ps_y_pos_phase[i4_y + i4_frm_mb_y].i2_ref_pos; |
1774 | 23.1M | i4_y_phase = ps_y_pos_phase[i4_y + i4_frm_mb_y].i2_phase; |
1775 | 23.1M | i4_x_ref = ps_x_pos_phase[i4_x + i4_frm_mb_x].i2_ref_pos; |
1776 | 23.1M | i4_x_phase = ps_x_pos_phase[i4_x + i4_frm_mb_x].i2_phase; |
1777 | | |
1778 | | /* horizontal processing*/ |
1779 | 69.3M | for(i4_i = 0; i4_i < 2; i4_i++) |
1780 | 46.2M | { |
1781 | 46.2M | UWORD8 *pu1_ref_x_ptr_incr_temp; |
1782 | 46.2M | UWORD8 u1_incr; |
1783 | 46.2M | WORD16 *pi2_ref_array_1, *pi2_ref_array_2; |
1784 | | |
1785 | | /* derive appropriate pointers */ |
1786 | 46.2M | pu1_ref_x_ptr_incr_temp = pu1_ref_x_ptr_incr + i4_x_ref; |
1787 | 46.2M | pu1_ref_x_ptr_incr_temp += ((i4_y_ref + i4_i) * i4_refarray_wd); |
1788 | 46.2M | u1_incr = *pu1_ref_x_ptr_incr_temp; |
1789 | 46.2M | pi2_ref_array_1 = pi2_ref_array + i4_x_ref; |
1790 | 46.2M | pi2_ref_array_1 += ((i4_y_ref + i4_i) * i4_refarray_wd); |
1791 | | |
1792 | 46.2M | if(!u1_incr) |
1793 | 15.0M | { |
1794 | 15.0M | pi2_ref_array_1 += (i4_x_phase >> 3); |
1795 | 15.0M | } |
1796 | | |
1797 | 46.2M | pi2_ref_array_2 = pi2_ref_array_1 + u1_incr; |
1798 | 46.2M | ai4_temp_pred[i4_i] = |
1799 | 46.2M | (16 - i4_x_phase) * (*pi2_ref_array_1) + i4_x_phase * (*pi2_ref_array_2); |
1800 | 46.2M | } |
1801 | | |
1802 | | /* vertical processing */ |
1803 | 23.1M | i4_x_ref_round = (i4_x_ref + (i4_x_phase >> 3)); |
1804 | 23.1M | pu1_ref_y_ptr_incr_temp = |
1805 | 23.1M | pu1_ref_y_ptr_incr + i4_x_ref_round + (i4_y_ref * i4_refarray_wd); |
1806 | 23.1M | u1_incr_y = *pu1_ref_y_ptr_incr_temp; |
1807 | | |
1808 | 23.1M | pi4_temp_pred = &ai4_temp_pred[0]; |
1809 | 23.1M | if(!u1_incr_y) |
1810 | 7.51M | { |
1811 | 7.51M | pi4_temp_pred += (i4_y_phase >> 3); |
1812 | 7.51M | } |
1813 | | |
1814 | 23.1M | i2_res = (((16 - i4_y_phase) * pi4_temp_pred[0] + |
1815 | 23.1M | i4_y_phase * pi4_temp_pred[u1_incr_y] + 128) >> |
1816 | 23.1M | 8); |
1817 | | |
1818 | | /* store back the final residual */ |
1819 | 23.1M | *pi2_out_curr = i2_res; |
1820 | 23.1M | } /* end of loop over width */ |
1821 | 1.92M | } /* end of loop over height */ |
1822 | | |
1823 | 180k | return; |
1824 | 180k | } /* End of Interpolation Function */ |
1825 | | /*****************************************************************************/ |
1826 | | /* */ |
1827 | | /* Function Name : isvcd_residual_samp_mb */ |
1828 | | /* */ |
1829 | | /* Description : MB level function whcih perform the residual resampling */ |
1830 | | /* of data of an MB (luma and chroma insclusive) */ |
1831 | | /* */ |
1832 | | /* Inputs : pv_residual_samp_ctxt : residual sampling context */ |
1833 | | /* ps_ref_luma : reference layer luma data buffer desc */ |
1834 | | /* ps_ref_chroma : reference layer chroma data buffer desc */ |
1835 | | /* ps_ref_luma_bitmap : ref layer luma bit map buffer desc */ |
1836 | | /* ps_ref_chroma_bitmap : ref layer chroma bit map buff des */ |
1837 | | /* ps_ref_mb_mode : ref layer mb mode map buff desc */ |
1838 | | /* ps_out_luma : current layer out luma buffer desc */ |
1839 | | /* ps_out_chroma : current layer out chroma buffer desc */ |
1840 | | /* ps_mb_coord : current mb coorinate */ |
1841 | | /* Globals : none */ |
1842 | | /* Processing : it calls the reference layer construction followed by */ |
1843 | | /* interplaotion function for luma and cb and cr */ |
1844 | | /* Outputs : inter resampled data of current MB */ |
1845 | | /* Returns : none */ |
1846 | | /* */ |
1847 | | /* Issues : none */ |
1848 | | /* */ |
1849 | | /* Revision History: */ |
1850 | | /* */ |
1851 | | /* DD MM YYYY Author(s) Changes (Describe the changes made) */ |
1852 | | /* 26 06 2021 vijayakumar creation */ |
1853 | | /* */ |
1854 | | /*****************************************************************************/ |
1855 | | WORD32 isvcd_residual_samp_mb(void *pv_residual_samp_ctxt, mem_element_t *ps_ref_luma, |
1856 | | mem_element_t *ps_ref_chroma, mem_element_t *ps_ref_mb_mode, |
1857 | | mem_element_t *ps_out_luma, mem_element_t *ps_out_chroma, |
1858 | | UWORD16 u2_mb_x, UWORD16 u2_mb_y) |
1859 | 109k | { |
1860 | | /* --------------------------------------------------------------------- */ |
1861 | | /* I/O buffer params */ |
1862 | | /* --------------------------------------------------------------------- */ |
1863 | 109k | residual_sampling_ctxt_t *ps_ctxt = (residual_sampling_ctxt_t *) pv_residual_samp_ctxt; |
1864 | 109k | WORD16 *pi2_inp; |
1865 | 109k | WORD16 *pi2_out; |
1866 | 109k | WORD32 i4_inp_stride; |
1867 | 109k | WORD32 i4_out_stride; |
1868 | 109k | WORD32 i4_refarray_wd; |
1869 | 109k | mb_coord_t s_mb_coord = {0}; |
1870 | 109k | WORD32 ret; |
1871 | 109k | s_mb_coord.u2_mb_x = u2_mb_x; |
1872 | 109k | s_mb_coord.u2_mb_y = u2_mb_y; |
1873 | | |
1874 | | /* --------------------------------------------------------------------- */ |
1875 | | /* LUMA PROCESSING */ |
1876 | | /* --------------------------------------------------------------------- */ |
1877 | 109k | pi2_inp = (WORD16 *) ps_ref_luma->pv_buffer; |
1878 | 109k | pi2_out = (WORD16 *) ps_out_luma->pv_buffer; |
1879 | 109k | i4_inp_stride = ps_ref_luma->i4_num_element_stride; |
1880 | 109k | i4_out_stride = ps_out_luma->i4_num_element_stride; |
1881 | | |
1882 | | /* ------- Constructing refSampleArray ----------------------- */ |
1883 | 109k | ret = isvcd_residual_reflayer_const(pv_residual_samp_ctxt, pi2_inp, i4_inp_stride, |
1884 | 109k | ps_ref_mb_mode, &i4_refarray_wd, &s_mb_coord, 0); |
1885 | | |
1886 | 109k | if(ret != OK) return ret; |
1887 | | /* ---- Interpolation process for Residual prediction ------ */ |
1888 | 109k | ps_ctxt->pf_interpolate_residual(pv_residual_samp_ctxt, pi2_out, i4_out_stride, i4_refarray_wd, |
1889 | 109k | s_mb_coord.u2_mb_x, s_mb_coord.u2_mb_y, 0); |
1890 | | |
1891 | | /* --------------------------------------------------------------------- */ |
1892 | | /* CHROMA PROCESSING */ |
1893 | | /* --------------------------------------------------------------------- */ |
1894 | | /* CB */ |
1895 | 109k | pi2_inp = (WORD16 *) ps_ref_chroma->pv_buffer; |
1896 | 109k | pi2_out = (WORD16 *) ps_out_chroma->pv_buffer; |
1897 | 109k | i4_inp_stride = ps_ref_chroma->i4_num_element_stride; |
1898 | 109k | i4_out_stride = ps_out_chroma->i4_num_element_stride; |
1899 | | |
1900 | | /* ------- Constructing refSampleArray ----------------------- */ |
1901 | 109k | ret = isvcd_residual_reflayer_const(pv_residual_samp_ctxt, pi2_inp, i4_inp_stride, |
1902 | 109k | ps_ref_mb_mode, &i4_refarray_wd, &s_mb_coord, 1); |
1903 | | |
1904 | 109k | if(ret != OK) return ret; |
1905 | | /* ---- Interpolation process for Residual prediction ------ */ |
1906 | 109k | ps_ctxt->pf_interpolate_residual(pv_residual_samp_ctxt, pi2_out, i4_out_stride, i4_refarray_wd, |
1907 | 109k | s_mb_coord.u2_mb_x, s_mb_coord.u2_mb_y, 1); |
1908 | | |
1909 | | /* CR */ |
1910 | 109k | pi2_inp += 1; |
1911 | 109k | pi2_out += 1; |
1912 | | |
1913 | | /* ------- Constructing refSampleArray ----------------------- */ |
1914 | 109k | ret = isvcd_residual_reflayer_const(pv_residual_samp_ctxt, pi2_inp, i4_inp_stride, |
1915 | 109k | ps_ref_mb_mode, &i4_refarray_wd, &s_mb_coord, 1); |
1916 | | |
1917 | 109k | if(ret != OK) return ret; |
1918 | | /* ---- Interpolation process for Residual prediction --------- */ |
1919 | 109k | ps_ctxt->pf_interpolate_residual(pv_residual_samp_ctxt, pi2_out, i4_out_stride, i4_refarray_wd, |
1920 | 109k | s_mb_coord.u2_mb_x, s_mb_coord.u2_mb_y, 1); |
1921 | 109k | return OK; |
1922 | 109k | } |
1923 | | /*****************************************************************************/ |
1924 | | /* */ |
1925 | | /* Function Name : isvcd_residual_samp_mb_dyadic */ |
1926 | | /* */ |
1927 | | /* Description : MB level function whcih perform the residual resampling */ |
1928 | | /* of data of an MB (luma and chroma insclusive) */ |
1929 | | /* for Dyadic cases */ |
1930 | | /* Inputs : pv_residual_samp_ctxt : residual sampling context */ |
1931 | | /* ps_ref_luma : reference layer luma data buffer desc */ |
1932 | | /* ps_ref_chroma : reference layer chroma data buffer desc */ |
1933 | | /* ps_ref_luma_bitmap : ref layer luma bit map buffer desc */ |
1934 | | /* ps_ref_chroma_bitmap : ref layer chroma bit map buff des */ |
1935 | | /* ps_ref_mb_mode : ref layer mb mode map buff desc */ |
1936 | | /* ps_out_luma : current layer out luma buffer desc */ |
1937 | | /* ps_out_chroma : current layer out chroma buffer desc */ |
1938 | | /* ps_mb_coord : current mb coorinate */ |
1939 | | /* Globals : none */ |
1940 | | /* Processing : it calls the reference layer construction followed by */ |
1941 | | /* interplaotion function for luma and cb and cr */ |
1942 | | /* Outputs : inter resampled data of current MB */ |
1943 | | /* Returns : none */ |
1944 | | /* */ |
1945 | | /* Issues : none */ |
1946 | | /* */ |
1947 | | /* Revision History: */ |
1948 | | /* */ |
1949 | | /* DD MM YYYY Author(s) Changes (Describe the changes made) */ |
1950 | | /* 26 06 2021 vijayakumar creation */ |
1951 | | /* */ |
1952 | | /*****************************************************************************/ |
1953 | | WORD32 isvcd_residual_samp_mb_dyadic(void *pv_residual_samp_ctxt, mem_element_t *ps_ref_luma, |
1954 | | mem_element_t *ps_ref_chroma, mem_element_t *ps_ref_mb_mode, |
1955 | | mem_element_t *ps_out_luma, mem_element_t *ps_out_chroma, |
1956 | | UWORD16 u2_mb_x, UWORD16 u2_mb_y) |
1957 | 57.2k | { |
1958 | 57.2k | residual_sampling_ctxt_t *ps_ctxt; |
1959 | 57.2k | res_lyr_ctxt *ps_lyr_ctxt; |
1960 | | /* --------------------------------------------------------------------- */ |
1961 | | /* I/O buffer params */ |
1962 | | /* --------------------------------------------------------------------- */ |
1963 | 57.2k | WORD16 *pi2_inp; |
1964 | 57.2k | WORD16 *pi2_out; |
1965 | 57.2k | WORD32 i4_inp_stride; |
1966 | 57.2k | WORD32 i4_out_stride; |
1967 | 57.2k | WORD32 i4_luma_nnz; |
1968 | 57.2k | WORD32 i4_chroma_nnz; |
1969 | 57.2k | WORD32 i4_tx_size; |
1970 | | |
1971 | 57.2k | ps_ctxt = (residual_sampling_ctxt_t *) pv_residual_samp_ctxt; |
1972 | 57.2k | ps_lyr_ctxt = &ps_ctxt->as_res_lyrs[ps_ctxt->i4_res_lyr_id]; |
1973 | | |
1974 | | /* --------------------------------------------------------------------- */ |
1975 | | /* LUMA PROCESSING */ |
1976 | | /* --------------------------------------------------------------------- */ |
1977 | 57.2k | pi2_inp = (WORD16 *) ps_ref_luma->pv_buffer; |
1978 | 57.2k | pi2_out = (WORD16 *) ps_out_luma->pv_buffer; |
1979 | 57.2k | i4_inp_stride = ps_ref_luma->i4_num_element_stride; |
1980 | 57.2k | i4_out_stride = ps_out_luma->i4_num_element_stride; |
1981 | | |
1982 | 57.2k | { |
1983 | 57.2k | WORD32 i4_offset_x, i4_offset_y; |
1984 | 57.2k | residual_samp_map_ctxt_t *ps_luma_map; |
1985 | 57.2k | ref_mb_map_t *ps_x_off_len_luma; |
1986 | 57.2k | ref_mb_map_t *ps_y_off_len_luma; |
1987 | | |
1988 | 57.2k | ps_luma_map = &ps_lyr_ctxt->s_luma_map_ctxt; |
1989 | 57.2k | ps_x_off_len_luma = ps_luma_map->ps_x_offset_length; |
1990 | 57.2k | ps_y_off_len_luma = ps_luma_map->ps_y_offset_length; |
1991 | | |
1992 | | /* get the actual offset for the buffers */ |
1993 | 57.2k | i4_offset_x = ps_x_off_len_luma[u2_mb_x].i2_offset; |
1994 | 57.2k | i4_offset_y = ps_y_off_len_luma[u2_mb_y].i2_offset; |
1995 | | |
1996 | 57.2k | { |
1997 | 57.2k | inter_lyr_mb_prms_t *ps_inter_lyr_mb_prms; |
1998 | 57.2k | WORD32 i4_mb_x, i4_mb_y; |
1999 | 57.2k | UWORD16 u2_luma_mask = 0x0033; |
2000 | 57.2k | UWORD8 u1_chrm_mask = 0x11; |
2001 | 57.2k | WORD32 i4_luma_rt_sft_amt = 0; |
2002 | 57.2k | WORD32 i4_chrm_rt_sft_amt = 0; |
2003 | | |
2004 | 57.2k | i4_mb_x = ((i4_offset_x + 1) >> MB_WIDTH_SHIFT); |
2005 | 57.2k | i4_mb_y = ((i4_offset_y + 1) >> MB_HEIGHT_SHIFT); |
2006 | | |
2007 | | /* get the location of the byte which has the current mb mode */ |
2008 | 57.2k | ps_inter_lyr_mb_prms = (inter_lyr_mb_prms_t *) ps_ref_mb_mode->pv_buffer; |
2009 | 57.2k | ps_inter_lyr_mb_prms += i4_mb_x; |
2010 | 57.2k | ps_inter_lyr_mb_prms += i4_mb_y * ps_ref_mb_mode->i4_num_element_stride; |
2011 | | |
2012 | | /* get the approp block in base layer in horz direction */ |
2013 | 57.2k | if(0 != ((i4_offset_x + 1) & 15)) |
2014 | 29.6k | { |
2015 | 29.6k | u2_luma_mask <<= 2; |
2016 | 29.6k | i4_luma_rt_sft_amt += 2; |
2017 | 29.6k | u1_chrm_mask <<= 1; |
2018 | 29.6k | i4_chrm_rt_sft_amt += 1; |
2019 | 29.6k | } |
2020 | | /* get the approp block in base layer in vert direction */ |
2021 | 57.2k | if(0 != ((i4_offset_y + 1) & 15)) |
2022 | 28.0k | { |
2023 | 28.0k | u2_luma_mask <<= 8; |
2024 | 28.0k | i4_luma_rt_sft_amt += 8; |
2025 | | |
2026 | 28.0k | u1_chrm_mask <<= 2; |
2027 | 28.0k | i4_chrm_rt_sft_amt += 2; |
2028 | 28.0k | } |
2029 | | |
2030 | | /* extract the nnz and store it */ |
2031 | 57.2k | i4_luma_nnz = (ps_inter_lyr_mb_prms->u2_luma_nnz & u2_luma_mask) >> i4_luma_rt_sft_amt; |
2032 | 57.2k | i4_chroma_nnz = |
2033 | 57.2k | (ps_inter_lyr_mb_prms->u1_chroma_nnz & u1_chrm_mask) >> i4_chrm_rt_sft_amt; |
2034 | 57.2k | i4_tx_size = |
2035 | 57.2k | (ps_inter_lyr_mb_prms->i1_tx_size < 0) ? 1 : ps_inter_lyr_mb_prms->i1_tx_size; |
2036 | 57.2k | } |
2037 | | |
2038 | | /* since in dyadic case the window width and height will be 10x10 */ |
2039 | | /* and the window start offsets will be always 1 column left and */ |
2040 | | /* 1 row above the block boundary. so the pointer and the required */ |
2041 | | /* positions are appropriately modified */ |
2042 | 57.2k | if(i4_offset_x >= 0) |
2043 | 44.2k | { |
2044 | 44.2k | pi2_inp++; |
2045 | 44.2k | } |
2046 | | |
2047 | 57.2k | if(i4_offset_y >= 0) |
2048 | 51.3k | { |
2049 | 51.3k | pi2_inp += i4_inp_stride; |
2050 | 51.3k | } |
2051 | | |
2052 | 57.2k | ps_ctxt->pf_residual_luma_dyadic(pv_residual_samp_ctxt, pi2_inp, i4_inp_stride, pi2_out, |
2053 | 57.2k | i4_out_stride, ps_ref_mb_mode, u2_mb_x, u2_mb_y, |
2054 | 57.2k | i4_luma_nnz, i4_tx_size); |
2055 | 57.2k | } |
2056 | | |
2057 | | /* --------------------------------------------------------------------- */ |
2058 | | /* CHROMA PROCESSING */ |
2059 | | /* --------------------------------------------------------------------- */ |
2060 | | /* CB */ |
2061 | 57.2k | pi2_inp = (WORD16 *) ps_ref_chroma->pv_buffer; |
2062 | 57.2k | pi2_out = (WORD16 *) ps_out_chroma->pv_buffer; |
2063 | 57.2k | i4_inp_stride = ps_ref_chroma->i4_num_element_stride; |
2064 | 57.2k | i4_out_stride = ps_out_chroma->i4_num_element_stride; |
2065 | | |
2066 | | /* choose the appropriate chroma processing routine */ |
2067 | 57.2k | if(SVCD_FALSE == ps_lyr_ctxt->i4_chrm_alt_proc) |
2068 | 57.2k | { |
2069 | 57.2k | WORD32 i4_offset_x, i4_offset_y; |
2070 | 57.2k | residual_samp_map_ctxt_t *ps_chroma_map; |
2071 | 57.2k | ref_mb_map_t *ps_x_off_len_chroma; |
2072 | 57.2k | ref_mb_map_t *ps_y_off_len_chroma; |
2073 | | |
2074 | 57.2k | ps_chroma_map = &ps_lyr_ctxt->s_chroma_map_ctxt; |
2075 | 57.2k | ps_x_off_len_chroma = ps_chroma_map->ps_x_offset_length; |
2076 | 57.2k | ps_y_off_len_chroma = ps_chroma_map->ps_y_offset_length; |
2077 | | |
2078 | | /* get the actual offset for the buffers */ |
2079 | 57.2k | i4_offset_x = ps_x_off_len_chroma[u2_mb_x].i2_offset; |
2080 | 57.2k | i4_offset_y = ps_y_off_len_chroma[u2_mb_y].i2_offset; |
2081 | | |
2082 | | /* since in dyadic case the window width and height will be 6x6 */ |
2083 | | /* and the window start offsets will be always 1 column left and */ |
2084 | | /* 1 row above the block boundary. so the pointer and the required */ |
2085 | | /* positions are appropriately modified */ |
2086 | 57.2k | if(i4_offset_x >= 0) |
2087 | 44.2k | { |
2088 | 44.2k | pi2_inp += 2; |
2089 | 44.2k | } |
2090 | | |
2091 | 57.2k | if(i4_offset_y >= 0) |
2092 | 51.3k | { |
2093 | 51.3k | pi2_inp += i4_inp_stride; |
2094 | 51.3k | } |
2095 | | |
2096 | 57.2k | if(0 != (i4_chroma_nnz & 0x01)) |
2097 | 871 | { |
2098 | 871 | ps_ctxt->pf_residual_chroma_dyadic(pv_residual_samp_ctxt, pi2_inp, i4_inp_stride, |
2099 | 871 | pi2_out, i4_out_stride); |
2100 | 871 | } |
2101 | 57.2k | } |
2102 | 0 | else |
2103 | 0 | { |
2104 | 0 | ps_ctxt->pf_residual_chroma_dyadic_alt(pv_residual_samp_ctxt, u2_mb_x, u2_mb_y, |
2105 | 0 | ps_ref_mb_mode, pi2_inp, i4_inp_stride, pi2_out, |
2106 | 0 | i4_out_stride, SVCD_FALSE); |
2107 | 0 | } |
2108 | | |
2109 | | /* CR */ |
2110 | 57.2k | pi2_inp += 1; |
2111 | 57.2k | pi2_out += 1; |
2112 | | |
2113 | 57.2k | if(SVCD_FALSE == ps_lyr_ctxt->i4_chrm_alt_proc) |
2114 | 57.2k | { |
2115 | 57.2k | if(0 != (i4_chroma_nnz & 0x10)) |
2116 | 1.14k | { |
2117 | 1.14k | ps_ctxt->pf_residual_chroma_dyadic(pv_residual_samp_ctxt, pi2_inp, i4_inp_stride, |
2118 | 1.14k | pi2_out, i4_out_stride); |
2119 | 1.14k | } |
2120 | 57.2k | } |
2121 | 0 | else |
2122 | 0 | { |
2123 | 0 | ps_ctxt->pf_residual_chroma_dyadic_alt(pv_residual_samp_ctxt, u2_mb_x, u2_mb_y, |
2124 | 0 | ps_ref_mb_mode, pi2_inp, i4_inp_stride, pi2_out, |
2125 | 0 | i4_out_stride, SVCD_TRUE); |
2126 | 0 | } |
2127 | 57.2k | return OK; |
2128 | 57.2k | } |
2129 | | |
2130 | | /*****************************************************************************/ |
2131 | | /* */ |
2132 | | /* Function Name : isvcd_residual_samp_populate_list */ |
2133 | | /* */ |
2134 | | /* Description : This is a seq or frame level init function which fills */ |
2135 | | /* all offsets, projected locations arrays based on */ |
2136 | | /* the two resolutions and cropping parameters */ |
2137 | | /* Inputs : refer to doxygen comments below */ |
2138 | | /* Globals : none */ |
2139 | | /* Processing : it projects the locations and computes the values */ |
2140 | | /* */ |
2141 | | /* Outputs : none */ |
2142 | | /* Returns : none */ |
2143 | | /* */ |
2144 | | /* Issues : none */ |
2145 | | /* */ |
2146 | | /* Revision History: */ |
2147 | | /* */ |
2148 | | /* DD MM YYYY Author(s) Changes (Describe the changes made) */ |
2149 | | /* 06 07 2021 vijayakumar creation */ |
2150 | | /* */ |
2151 | | /*****************************************************************************/ |
2152 | | void isvcd_residual_samp_populate_list(residual_samp_map_ctxt_t *ps_map_ctxt, |
2153 | | dec_seq_params_t *ps_sps, |
2154 | | dec_svc_seq_params_t *ps_subset_sps, |
2155 | | res_prms_t *ps_curr_res_prms, res_prms_t *ps_ref_res_prms, |
2156 | | WORD32 i4_chroma_flag) |
2157 | 68.2k | { |
2158 | | /* --------------------------------------------------------------------- */ |
2159 | | /* Local variables required for finding the mapping between the layers */ |
2160 | | /* --------------------------------------------------------------------- */ |
2161 | 68.2k | UWORD32 u4_shift_x; |
2162 | 68.2k | UWORD32 u4_shift_y; |
2163 | 68.2k | UWORD32 u4_scale_x; |
2164 | 68.2k | UWORD32 u4_scale_y; |
2165 | 68.2k | WORD32 i4_offset_x; |
2166 | 68.2k | WORD32 i4_offset_y; |
2167 | 68.2k | WORD32 i4_add_x; |
2168 | 68.2k | WORD32 i4_add_y; |
2169 | 68.2k | WORD32 i4_delta_x; |
2170 | 68.2k | WORD32 i4_delta_y; |
2171 | 68.2k | WORD32 i4_refphase_x; |
2172 | 68.2k | WORD32 i4_refphase_y; |
2173 | 68.2k | WORD32 i4_phase_x; |
2174 | 68.2k | WORD32 i4_phase_y; |
2175 | 68.2k | WORD32 i4_sub_wd; |
2176 | 68.2k | WORD32 i4_sub_ht; |
2177 | 68.2k | WORD32 i4_mb_wd; |
2178 | 68.2k | WORD32 i4_mb_ht; |
2179 | | /* --------------------------------------------------------------------- */ |
2180 | | /* Local Pointer Declaration for arrays in Mapping context */ |
2181 | | /* --------------------------------------------------------------------- */ |
2182 | 68.2k | ref_mb_map_t *ps_x_off_len; |
2183 | 68.2k | ref_mb_map_t *ps_y_off_len; |
2184 | 68.2k | UWORD32 i4_ref_wd; |
2185 | 68.2k | UWORD32 i4_ref_ht; |
2186 | 68.2k | UWORD32 i4_scaled_wd; |
2187 | 68.2k | UWORD32 i4_scaled_ht; |
2188 | 68.2k | WORD32 i4_curr_lyr_width; |
2189 | 68.2k | WORD32 i4_curr_lyr_height; |
2190 | | |
2191 | | /* --------------------------------------------------------------------- */ |
2192 | | /* Local Flag Declaration */ |
2193 | | /* --------------------------------------------------------------------- */ |
2194 | 68.2k | WORD32 i4_ref_layer_field_pic_flag; |
2195 | 68.2k | WORD32 i4_field_pic_flag; |
2196 | 68.2k | WORD32 i4_frame_mbs_only_flag; |
2197 | 68.2k | WORD32 i4_ref_layer_frame_Mbs_only_flag; |
2198 | 68.2k | WORD32 i4_field_Mb_flag; |
2199 | 68.2k | WORD32 i4_bot_field_flag; |
2200 | | |
2201 | | /* --------------------------------------------------------------------- */ |
2202 | | /* Cropping Parameters Declaration */ |
2203 | | /* --------------------------------------------------------------------- */ |
2204 | 68.2k | WORD32 i4_scaled_ref_layer_left_offset; |
2205 | 68.2k | WORD32 i4_scaled_ref_layer_top_offset; |
2206 | | |
2207 | | /* --------------------------------------------------------------------- */ |
2208 | | /* Hardcoding flag information (assuming no field support) */ |
2209 | | /* --------------------------------------------------------------------- */ |
2210 | 68.2k | i4_ref_layer_field_pic_flag = SVCD_FALSE; |
2211 | 68.2k | i4_field_pic_flag = SVCD_FALSE; |
2212 | 68.2k | i4_frame_mbs_only_flag = SVCD_TRUE; |
2213 | 68.2k | i4_field_Mb_flag = SVCD_FALSE; |
2214 | 68.2k | i4_bot_field_flag = SVCD_FALSE; |
2215 | 68.2k | i4_ref_layer_frame_Mbs_only_flag = SVCD_TRUE; |
2216 | | |
2217 | | /* --------------------------------------------------------------------- */ |
2218 | | /* Pointer and Paramater are intialized - Chroma and Luma */ |
2219 | | /* --------------------------------------------------------------------- */ |
2220 | 68.2k | { |
2221 | 68.2k | WORD32 i4_base_width; |
2222 | 68.2k | WORD32 i4_base_height; |
2223 | 68.2k | WORD32 i4_ref_layer_chroma_phase_x_plus1_flag; |
2224 | 68.2k | WORD32 i4_ref_layer_chroma_phase_y_plus1; |
2225 | 68.2k | WORD32 i4_chroma_phase_x_plus1_flag; |
2226 | 68.2k | WORD32 i4_chroma_phase_y_plus1; |
2227 | | |
2228 | | /* ------------------------------------------------------------- */ |
2229 | | /* HARD CODED FOR 420 */ |
2230 | | /* ------------------------------------------------------------- */ |
2231 | 68.2k | WORD32 i4_sub_wd_chroma = 2; |
2232 | 68.2k | WORD32 i4_sub_ht_chroma = 2; |
2233 | | |
2234 | 68.2k | i4_base_width = ps_ref_res_prms->i4_res_width; |
2235 | 68.2k | i4_base_height = ps_ref_res_prms->i4_res_height; |
2236 | | |
2237 | 68.2k | i4_ref_layer_chroma_phase_x_plus1_flag = |
2238 | 68.2k | ps_curr_res_prms->i1_ref_lyr_chroma_phase_x_plus1_flag; |
2239 | 68.2k | i4_ref_layer_chroma_phase_y_plus1 = ps_curr_res_prms->i1_ref_lyr_chroma_phase_y_plus1; |
2240 | 68.2k | i4_chroma_phase_x_plus1_flag = |
2241 | 68.2k | ps_subset_sps->s_sps_svc_ext.u1_seq_ref_layer_chroma_phase_x_plus1_flag; |
2242 | 68.2k | i4_chroma_phase_y_plus1 = |
2243 | 68.2k | ps_subset_sps->s_sps_svc_ext.u1_seq_ref_layer_chroma_phase_y_plus1; |
2244 | 68.2k | i4_scaled_ref_layer_left_offset = ps_curr_res_prms->s_ref_lyr_scaled_offset.i2_left; |
2245 | 68.2k | i4_scaled_ref_layer_top_offset = ps_curr_res_prms->s_ref_lyr_scaled_offset.i2_top; |
2246 | | |
2247 | | /* ----------------------------------------------------------------- */ |
2248 | | /* Computing Effective Frame Dimensions */ |
2249 | | /* ------------------------------------------------------------------*/ |
2250 | 68.2k | i4_ref_wd = (i4_base_width >> i4_chroma_flag); |
2251 | 68.2k | i4_ref_ht = (i4_base_height >> i4_chroma_flag) * (1 + i4_ref_layer_field_pic_flag); |
2252 | 68.2k | i4_scaled_wd = ps_curr_res_prms->u2_scaled_ref_width; |
2253 | 68.2k | i4_scaled_ht = ps_curr_res_prms->u2_scaled_ref_height; |
2254 | 68.2k | i4_scaled_wd = (i4_scaled_wd >> i4_chroma_flag); |
2255 | 68.2k | i4_scaled_ht = (i4_scaled_ht >> i4_chroma_flag) * (1 + i4_field_pic_flag); |
2256 | | |
2257 | 68.2k | if(1 == i4_chroma_flag) |
2258 | 34.1k | { |
2259 | 34.1k | i4_refphase_x = i4_ref_layer_chroma_phase_x_plus1_flag - 1; |
2260 | 34.1k | i4_refphase_y = i4_ref_layer_chroma_phase_y_plus1 - 1; |
2261 | 34.1k | i4_phase_x = i4_chroma_phase_x_plus1_flag - 1; |
2262 | 34.1k | i4_phase_y = i4_chroma_phase_y_plus1 - 1; |
2263 | 34.1k | i4_sub_wd = i4_sub_wd_chroma; |
2264 | 34.1k | i4_sub_ht = i4_sub_ht_chroma; |
2265 | 34.1k | i4_mb_wd = MB_WIDTH >> 1; |
2266 | 34.1k | i4_mb_ht = MB_HEIGHT >> 1; |
2267 | 34.1k | } |
2268 | 34.1k | else |
2269 | 34.1k | { |
2270 | 34.1k | i4_refphase_x = 0; |
2271 | 34.1k | i4_refphase_y = 0; |
2272 | 34.1k | i4_phase_x = 0; |
2273 | 34.1k | i4_phase_y = 0; |
2274 | 34.1k | i4_sub_wd = 1; |
2275 | 34.1k | i4_sub_ht = 1; |
2276 | 34.1k | i4_mb_wd = MB_WIDTH; |
2277 | 34.1k | i4_mb_ht = MB_HEIGHT; |
2278 | 34.1k | } |
2279 | 68.2k | } |
2280 | | |
2281 | | /* --------------------------------------------------------------------- */ |
2282 | | /* Derive shift x and y based on level idd */ |
2283 | | /* --------------------------------------------------------------------- */ |
2284 | 68.2k | if(ps_sps->u1_level_idc <= 30) |
2285 | 64.3k | { |
2286 | 64.3k | u4_shift_x = 16; |
2287 | 64.3k | u4_shift_y = 16; |
2288 | 64.3k | } |
2289 | 3.92k | else |
2290 | 3.92k | { |
2291 | 3.92k | u4_shift_x = 31 - isvcd_get_ceil_log2(i4_ref_wd); |
2292 | 3.92k | u4_shift_y = 31 - isvcd_get_ceil_log2(i4_ref_ht); |
2293 | 3.92k | } |
2294 | | |
2295 | | /* --------------------------------------------------------------------- */ |
2296 | | /* The following condition is not true in our case for time being */ |
2297 | | /* --------------------------------------------------------------------- */ |
2298 | 68.2k | if((SVCD_FALSE == i4_frame_mbs_only_flag) || (SVCD_FALSE == i4_ref_layer_frame_Mbs_only_flag)) |
2299 | 0 | { |
2300 | 0 | i4_phase_y = i4_phase_y + 4 * i4_bot_field_flag; |
2301 | |
|
2302 | 0 | if(1 == i4_ref_layer_frame_Mbs_only_flag) |
2303 | 0 | i4_refphase_y = (2 * i4_refphase_y) + 2; |
2304 | 0 | else |
2305 | 0 | i4_refphase_y = i4_refphase_y + (4 * i4_bot_field_flag); |
2306 | 0 | } |
2307 | | |
2308 | | /* --------------------------------------------------------------------- */ |
2309 | | /* Dx and Dy Computation - Ratio of the base and enhance layer width */ |
2310 | | /* --------------------------------------------------------------------- */ |
2311 | 68.2k | u4_scale_x = ((i4_ref_wd << u4_shift_x) + (i4_scaled_wd >> 1)) / (i4_scaled_wd); |
2312 | | |
2313 | 68.2k | u4_scale_y = ((i4_ref_ht << u4_shift_y) + (i4_scaled_ht >> 1)) / (i4_scaled_ht); |
2314 | | |
2315 | 68.2k | i4_offset_x = i4_scaled_ref_layer_left_offset / i4_sub_wd; |
2316 | 68.2k | i4_add_x = (((i4_ref_wd * (2 + i4_phase_x)) << (u4_shift_x - 2)) + (i4_scaled_wd >> 1)) / |
2317 | 68.2k | i4_scaled_wd + |
2318 | 68.2k | (1 << (u4_shift_x - 5)); |
2319 | 68.2k | i4_delta_x = 4 * (2 + i4_refphase_x); |
2320 | | |
2321 | 68.2k | if((SVCD_TRUE == i4_frame_mbs_only_flag) && (SVCD_TRUE == i4_ref_layer_frame_Mbs_only_flag)) |
2322 | 68.2k | { |
2323 | 68.2k | i4_offset_y = i4_scaled_ref_layer_top_offset / i4_sub_ht; |
2324 | 68.2k | i4_add_y = (((i4_ref_ht * (2 + i4_phase_y)) << (u4_shift_y - 2)) + (i4_scaled_ht >> 1)) / |
2325 | 68.2k | i4_scaled_ht + |
2326 | 68.2k | (1 << (u4_shift_y - 5)); |
2327 | 68.2k | i4_delta_y = 4 * (2 + i4_refphase_y); |
2328 | 68.2k | } |
2329 | 0 | else |
2330 | 0 | { |
2331 | 0 | i4_offset_y = i4_scaled_ref_layer_top_offset / (2 * i4_sub_ht); |
2332 | 0 | i4_add_y = (((i4_ref_ht * (2 + i4_phase_y)) << (u4_shift_y - 3)) + (i4_scaled_ht >> 1)) / |
2333 | 0 | i4_scaled_ht + |
2334 | 0 | (1 << (u4_shift_y - 5)); |
2335 | 0 | i4_delta_y = 2 * (2 + i4_refphase_y); |
2336 | 0 | } |
2337 | | |
2338 | | /* --------------------------------------------------------------------- */ |
2339 | | /* Intializing Local Pointers - Chroma and Luma */ |
2340 | | /* --------------------------------------------------------------------- */ |
2341 | 68.2k | ps_x_off_len = ps_map_ctxt->ps_x_offset_length; |
2342 | 68.2k | ps_y_off_len = ps_map_ctxt->ps_y_offset_length; |
2343 | 68.2k | i4_curr_lyr_width = ps_curr_res_prms->i4_res_width >> i4_chroma_flag; |
2344 | 68.2k | i4_curr_lyr_height = ps_curr_res_prms->i4_res_height >> i4_chroma_flag; |
2345 | | |
2346 | 68.2k | { |
2347 | 68.2k | WORD32 i4_i, i4_j; |
2348 | | |
2349 | | /* ----------------------------------------------------------------- */ |
2350 | | /* Computation of offsetX refArrayW Xmin and Xmax Lists */ |
2351 | | /* ----------------------------------------------------------------- */ |
2352 | 310k | for(i4_i = 0; i4_i < i4_curr_lyr_width; i4_i = i4_i + i4_mb_wd) |
2353 | 241k | { |
2354 | 241k | WORD32 i4_x_refmin16; |
2355 | 241k | WORD32 i4_x_refmax16; |
2356 | 241k | WORD32 i4_x_offset; |
2357 | | |
2358 | 241k | i4_x_refmin16 = (WORD64) (((WORD64) ((i4_i - i4_offset_x) * u4_scale_x) + i4_add_x) >> |
2359 | 241k | ((WORD32) (u4_shift_x - 4))) - |
2360 | 241k | i4_delta_x; |
2361 | | |
2362 | 241k | i4_x_refmax16 = |
2363 | 241k | (WORD64) (((WORD64) (i4_i + i4_mb_wd - 1 - i4_offset_x) * u4_scale_x + i4_add_x) >> |
2364 | 241k | ((WORD32) (u4_shift_x - 4))) - |
2365 | 241k | i4_delta_x; |
2366 | | |
2367 | | /* AC205 */ |
2368 | 241k | i4_x_offset = i4_x_refmin16 >> 4; |
2369 | 241k | ps_x_off_len->i2_offset = i4_x_offset; |
2370 | 241k | ps_x_off_len->i2_length = (i4_x_refmax16 >> 4) - i4_x_offset + 2; |
2371 | | |
2372 | | /* increment the pointer */ |
2373 | 241k | ps_x_off_len++; |
2374 | | |
2375 | 241k | } /* end of loop over current layer width */ |
2376 | | |
2377 | | /* ----------------------------------------------------------------- */ |
2378 | | /* Computation of offsetY refArrayH Ymin and Ymax Lists */ |
2379 | | /* ----------------------------------------------------------------- */ |
2380 | 704k | for(i4_j = 0; i4_j < i4_curr_lyr_height; i4_j = i4_j + i4_mb_ht) |
2381 | 636k | { |
2382 | 636k | WORD32 i4_y_refmin16; |
2383 | 636k | WORD32 i4_y_refmax16; |
2384 | 636k | WORD32 i4_y_offset; |
2385 | | |
2386 | 636k | i4_y_refmin16 = (WORD64) (((WORD64) (i4_j - i4_offset_y) * u4_scale_y + i4_add_y) >> |
2387 | 636k | ((WORD32) (u4_shift_y - 4))) - |
2388 | 636k | i4_delta_y; |
2389 | | |
2390 | 636k | i4_y_refmax16 = |
2391 | 636k | (WORD64) (((WORD64) (i4_j + i4_mb_ht - 1 - i4_offset_y) * u4_scale_y + i4_add_y) >> |
2392 | 636k | ((WORD32) (u4_shift_y - 4))) - |
2393 | 636k | i4_delta_y; |
2394 | | |
2395 | | /* AC205 */ |
2396 | 636k | i4_y_offset = i4_y_refmin16 >> 4; |
2397 | 636k | ps_y_off_len->i2_offset = i4_y_offset; |
2398 | 636k | ps_y_off_len->i2_length = (i4_y_refmax16 >> 4) - i4_y_offset + 2; |
2399 | | |
2400 | | /* increment the pointer */ |
2401 | 636k | ps_y_off_len++; |
2402 | | |
2403 | 636k | } /* end of loop over current layer height */ |
2404 | 68.2k | } |
2405 | | |
2406 | | /* --------------------------------------------------------------------- */ |
2407 | | /* Computation of Xref and Xphase List as per standard */ |
2408 | | /* --------------------------------------------------------------------- */ |
2409 | 68.2k | ps_x_off_len = ps_map_ctxt->ps_x_offset_length; |
2410 | 68.2k | ps_y_off_len = ps_map_ctxt->ps_y_offset_length; |
2411 | | |
2412 | 68.2k | { |
2413 | 68.2k | WORD32 i4_xc; |
2414 | 68.2k | WORD32 i4_offset_x_index; |
2415 | 68.2k | ref_pixel_map_t *ps_x_pos_phase; |
2416 | | |
2417 | 68.2k | ps_x_pos_phase = ps_map_ctxt->ps_x_pos_phase; |
2418 | | |
2419 | 2.97M | for(i4_xc = 0; i4_xc < i4_curr_lyr_width; i4_xc++) |
2420 | 2.90M | { |
2421 | 2.90M | WORD32 i4_x_offset; |
2422 | 2.90M | WORD32 i4_x_ref16; |
2423 | | |
2424 | 2.90M | i4_offset_x_index = i4_xc / i4_mb_wd; |
2425 | 2.90M | i4_x_offset = ps_x_off_len[i4_offset_x_index].i2_offset; |
2426 | 2.90M | i4_x_ref16 = (WORD64) (((WORD64) (i4_xc - i4_offset_x) * u4_scale_x + i4_add_x) >> |
2427 | 2.90M | ((WORD32) (u4_shift_x - 4))) - |
2428 | 2.90M | i4_delta_x; |
2429 | | |
2430 | | /* store the values */ |
2431 | 2.90M | ps_x_pos_phase->i2_ref_pos = (i4_x_ref16 >> 4) - i4_x_offset; |
2432 | 2.90M | ps_x_pos_phase->i2_phase = (i4_x_ref16 - (16 * i4_x_offset)) & 15; |
2433 | | |
2434 | | /* increment the pointer */ |
2435 | 2.90M | ps_x_pos_phase++; |
2436 | 2.90M | } /* end of loop over scaled width */ |
2437 | 68.2k | } |
2438 | | |
2439 | | /* --------------------------------------------------------------------- */ |
2440 | | /* Computation of Yref and Yphase List as per standard */ |
2441 | | /* --------------------------------------------------------------------- */ |
2442 | 68.2k | { |
2443 | 68.2k | WORD32 i4_yc; |
2444 | 68.2k | WORD32 i4_offset_y_index; |
2445 | 68.2k | ref_pixel_map_t *ps_y_pos_phase; |
2446 | | |
2447 | 68.2k | ps_y_pos_phase = ps_map_ctxt->ps_y_pos_phase; |
2448 | | |
2449 | 7.70M | for(i4_yc = 0; i4_yc < i4_curr_lyr_height; i4_yc++) |
2450 | 7.63M | { |
2451 | 7.63M | WORD32 i4_y_offset; |
2452 | 7.63M | WORD32 i4_y_ref16; |
2453 | | |
2454 | 7.63M | i4_offset_y_index = i4_yc / i4_mb_ht; |
2455 | 7.63M | i4_y_offset = ps_y_off_len[i4_offset_y_index].i2_offset; |
2456 | | |
2457 | 7.63M | if((SVCD_FALSE == i4_frame_mbs_only_flag) || |
2458 | 7.63M | (SVCD_FALSE == i4_ref_layer_frame_Mbs_only_flag)) |
2459 | 0 | { |
2460 | 0 | i4_yc = i4_yc >> (1 - i4_field_Mb_flag); |
2461 | 0 | } |
2462 | | |
2463 | 7.63M | i4_y_ref16 = (WORD64) ((((WORD64) (i4_yc - i4_offset_y) * u4_scale_y + i4_add_y) >> |
2464 | 7.63M | ((WORD32) (u4_shift_y - 4))) - |
2465 | 7.63M | i4_delta_y); |
2466 | 7.63M | ps_y_pos_phase->i2_ref_pos = (i4_y_ref16 >> 4) - i4_y_offset; |
2467 | 7.63M | ps_y_pos_phase->i2_phase = (i4_y_ref16 - (16 * i4_y_offset)) & 15; |
2468 | | |
2469 | | /* increment the pointer */ |
2470 | 7.63M | ps_y_pos_phase++; |
2471 | 7.63M | } /* end of loop over scaled height */ |
2472 | 68.2k | } |
2473 | 68.2k | return; |
2474 | 68.2k | } |
2475 | | |
2476 | | /*****************************************************************************/ |
2477 | | /* */ |
2478 | | /* Function Name : isvcd_residual_samp_res_init */ |
2479 | | /* */ |
2480 | | /* Description : this function calculates the scale factors and initialise*/ |
2481 | | /* the context structure */ |
2482 | | /* */ |
2483 | | /* Inputs : pv_residual_samp_ctxt: handle to private structure */ |
2484 | | /* ps_curr_lyr_res_prms: pointer to current resolution */ |
2485 | | /* params */ |
2486 | | /* Globals : none */ |
2487 | | /* Processing : it stores the layer dimensions */ |
2488 | | /* */ |
2489 | | /* Outputs : none */ |
2490 | | /* Returns : none */ |
2491 | | /* */ |
2492 | | /* Issues : none */ |
2493 | | /* */ |
2494 | | /* Revision History: */ |
2495 | | /* */ |
2496 | | /* DD MM YYYY Author(s) Changes (Describe the changes made) */ |
2497 | | /* 26 06 2021 vijayakumar creation */ |
2498 | | /* */ |
2499 | | /*****************************************************************************/ |
2500 | | WORD32 isvcd_residual_samp_res_init(void *pv_svc_dec) |
2501 | 137k | { |
2502 | 137k | residual_sampling_ctxt_t *ps_ctxt; |
2503 | 137k | res_lyr_ctxt *ps_lyr_ctxt; |
2504 | 137k | dec_seq_params_t *ps_sps; |
2505 | 137k | dec_svc_seq_params_t *ps_subset_sps; |
2506 | 137k | svc_dec_lyr_struct_t *ps_svc_lyr_dec = (svc_dec_lyr_struct_t *) pv_svc_dec; |
2507 | 137k | dec_struct_t *ps_dec = &ps_svc_lyr_dec->s_dec; |
2508 | | |
2509 | 137k | void *pv_residual_samp_ctxt = ps_svc_lyr_dec->pv_residual_sample_ctxt; |
2510 | 137k | res_prms_t *ps_curr_lyr_res_prms = &ps_svc_lyr_dec->s_res_prms; |
2511 | 137k | ref_mb_map_t **pps_luma_map_horz = &ps_svc_lyr_dec->ps_ressam_luma_map_horz; |
2512 | 137k | ref_mb_map_t **pps_chroma_map_horz = &ps_svc_lyr_dec->ps_ressam_chroma_map_horz; |
2513 | 137k | ref_mb_map_t **pps_luma_map_vert = &ps_svc_lyr_dec->ps_ressam_luma_map_vert; |
2514 | 137k | ref_mb_map_t **pps_chroma_map_vert = &ps_svc_lyr_dec->ps_ressam_chroma_map_vert; |
2515 | | |
2516 | 137k | if((NULL == pv_residual_samp_ctxt) || (NULL == ps_curr_lyr_res_prms) || |
2517 | 137k | (NULL == pps_luma_map_horz) || (NULL == pps_chroma_map_horz) || |
2518 | 137k | (NULL == pps_luma_map_vert) || (NULL == pps_chroma_map_vert)) |
2519 | 0 | { |
2520 | 0 | return NOT_OK; |
2521 | 0 | } |
2522 | | |
2523 | 137k | ps_ctxt = (residual_sampling_ctxt_t *) pv_residual_samp_ctxt; |
2524 | | |
2525 | | /* if called for base resolution store deafult values */ |
2526 | 137k | if(SVCD_TRUE == ps_svc_lyr_dec->u1_base_res_flag) |
2527 | 103k | { |
2528 | 103k | *pps_luma_map_horz = NULL; |
2529 | 103k | *pps_chroma_map_horz = NULL; |
2530 | 103k | *pps_luma_map_vert = NULL; |
2531 | 103k | *pps_chroma_map_vert = NULL; |
2532 | 103k | ps_ctxt->i4_res_lyr_id = -1; |
2533 | 103k | ps_ctxt->i4_ref_width = ps_curr_lyr_res_prms->i4_res_width; |
2534 | 103k | ps_ctxt->i4_ref_height = ps_curr_lyr_res_prms->i4_res_height; |
2535 | 103k | return OK; |
2536 | 103k | } |
2537 | | |
2538 | | /* derive the current sps */ |
2539 | 34.1k | ps_sps = ps_dec->ps_cur_sps; |
2540 | 34.1k | ps_subset_sps = ps_svc_lyr_dec->ps_cur_subset_sps; |
2541 | | |
2542 | | /* store the res id appropriately */ |
2543 | 34.1k | ps_ctxt->i4_res_lyr_id = ps_svc_lyr_dec->u1_layer_id - 1; |
2544 | | |
2545 | | /* get the current layer ctxt */ |
2546 | 34.1k | ps_lyr_ctxt = &ps_ctxt->as_res_lyrs[ps_svc_lyr_dec->u1_layer_id - 1]; |
2547 | | |
2548 | | /* get the width and heights */ |
2549 | 34.1k | ps_lyr_ctxt->i4_curr_width = ps_curr_lyr_res_prms->i4_res_width; |
2550 | 34.1k | ps_lyr_ctxt->i4_curr_height = ps_curr_lyr_res_prms->i4_res_height; |
2551 | 34.1k | ps_lyr_ctxt->i4_ref_width = ps_ctxt->i4_ref_width; |
2552 | 34.1k | ps_lyr_ctxt->i4_ref_height = ps_ctxt->i4_ref_height; |
2553 | | |
2554 | | /* store the strcuture pointer containing projected locations */ |
2555 | 34.1k | *pps_luma_map_horz = ps_lyr_ctxt->s_luma_map_ctxt.ps_x_offset_length; |
2556 | 34.1k | *pps_chroma_map_horz = ps_lyr_ctxt->s_chroma_map_ctxt.ps_x_offset_length; |
2557 | 34.1k | *pps_luma_map_vert = ps_lyr_ctxt->s_luma_map_ctxt.ps_y_offset_length; |
2558 | 34.1k | *pps_chroma_map_vert = ps_lyr_ctxt->s_chroma_map_ctxt.ps_y_offset_length; |
2559 | | |
2560 | | /* check for recomputation of mapping required */ |
2561 | 34.1k | if(SVCD_TRUE == ps_curr_lyr_res_prms->u1_remap_req_flag) |
2562 | 34.1k | { |
2563 | 34.1k | res_prms_t s_ref_res_prms = {0}; |
2564 | | |
2565 | | /* store the reference layer resolution width and height */ |
2566 | 34.1k | s_ref_res_prms.i4_res_width = ps_ctxt->i4_ref_width; |
2567 | 34.1k | s_ref_res_prms.i4_res_height = ps_ctxt->i4_ref_height; |
2568 | | |
2569 | | /* call the frame level projections calculation function */ |
2570 | 34.1k | isvcd_residual_samp_populate_list(&ps_lyr_ctxt->s_luma_map_ctxt, ps_sps, ps_subset_sps, |
2571 | 34.1k | ps_curr_lyr_res_prms, &s_ref_res_prms, 0); |
2572 | 34.1k | isvcd_residual_samp_populate_list(&ps_lyr_ctxt->s_chroma_map_ctxt, ps_sps, ps_subset_sps, |
2573 | 34.1k | ps_curr_lyr_res_prms, &s_ref_res_prms, 1); |
2574 | | |
2575 | | /* default values for flags */ |
2576 | 34.1k | ps_lyr_ctxt->pf_residual_samp_mb = &isvcd_residual_samp_mb; |
2577 | 34.1k | ps_lyr_ctxt->i4_chrm_horz_int_mode = 0; |
2578 | 34.1k | ps_lyr_ctxt->i4_chrm_vert_int_mode = 0; |
2579 | 34.1k | ps_lyr_ctxt->i4_chrm_alt_proc = SVCD_FALSE; |
2580 | | |
2581 | | /* Store the Dyadic flag */ |
2582 | 34.1k | ps_lyr_ctxt->i4_dyadic_flag = ps_curr_lyr_res_prms->u1_dyadic_flag; |
2583 | | |
2584 | | /* set the appropriate chroma processing routine based on */ |
2585 | | /* phase values */ |
2586 | 34.1k | if(SVCD_TRUE == ps_curr_lyr_res_prms->u1_dyadic_flag) |
2587 | 18.4k | { |
2588 | 18.4k | WORD32 i4_ref_layer_chroma_phase_x_plus1_flag; |
2589 | 18.4k | WORD32 i4_ref_layer_chroma_phase_y_plus1; |
2590 | 18.4k | WORD32 i4_chroma_phase_x_plus1_flag; |
2591 | 18.4k | WORD32 i4_chroma_phase_y_plus1; |
2592 | | |
2593 | 18.4k | ps_lyr_ctxt->pf_residual_samp_mb = &isvcd_residual_samp_mb_dyadic; |
2594 | 18.4k | i4_ref_layer_chroma_phase_x_plus1_flag = |
2595 | 18.4k | ps_curr_lyr_res_prms->i1_ref_lyr_chroma_phase_x_plus1_flag; |
2596 | 18.4k | i4_ref_layer_chroma_phase_y_plus1 = |
2597 | 18.4k | ps_curr_lyr_res_prms->i1_ref_lyr_chroma_phase_y_plus1; |
2598 | 18.4k | i4_chroma_phase_x_plus1_flag = |
2599 | 18.4k | ps_subset_sps->s_sps_svc_ext.u1_seq_ref_layer_chroma_phase_x_plus1_flag; |
2600 | 18.4k | i4_chroma_phase_y_plus1 = |
2601 | 18.4k | ps_subset_sps->s_sps_svc_ext.u1_seq_ref_layer_chroma_phase_y_plus1; |
2602 | 18.4k | if((0 == i4_ref_layer_chroma_phase_x_plus1_flag) && (1 == i4_chroma_phase_x_plus1_flag)) |
2603 | 0 | { |
2604 | 0 | ps_lyr_ctxt->i4_chrm_horz_int_mode = 1; |
2605 | 0 | ps_lyr_ctxt->i4_chrm_alt_proc = SVCD_TRUE; |
2606 | 0 | } |
2607 | | |
2608 | 18.4k | if((0 == i4_ref_layer_chroma_phase_y_plus1) && (1 == i4_chroma_phase_y_plus1)) |
2609 | 0 | { |
2610 | 0 | ps_lyr_ctxt->i4_chrm_vert_int_mode = 1; |
2611 | 0 | ps_lyr_ctxt->i4_chrm_alt_proc = SVCD_TRUE; |
2612 | 0 | } |
2613 | | |
2614 | 18.4k | if((0 == i4_ref_layer_chroma_phase_y_plus1) && (2 == i4_chroma_phase_y_plus1)) |
2615 | 0 | { |
2616 | 0 | ps_lyr_ctxt->i4_chrm_vert_int_mode = 1; |
2617 | 0 | ps_lyr_ctxt->i4_chrm_alt_proc = SVCD_TRUE; |
2618 | 0 | } |
2619 | | |
2620 | 18.4k | if((2 == i4_ref_layer_chroma_phase_y_plus1) && (0 == i4_chroma_phase_y_plus1)) |
2621 | 0 | { |
2622 | 0 | ps_lyr_ctxt->i4_chrm_vert_int_mode = 2; |
2623 | 0 | ps_lyr_ctxt->i4_chrm_alt_proc = SVCD_TRUE; |
2624 | 0 | } |
2625 | 18.4k | } |
2626 | 34.1k | } |
2627 | 0 | else |
2628 | 0 | { |
2629 | | /* should take false value */ |
2630 | 0 | if(SVCD_FALSE != ps_curr_lyr_res_prms->u1_remap_req_flag) |
2631 | 0 | { |
2632 | 0 | return NOT_OK; |
2633 | 0 | } |
2634 | 0 | } |
2635 | | |
2636 | | /* store the current layer width and height to context */ |
2637 | 34.1k | ps_ctxt->i4_ref_width = ps_curr_lyr_res_prms->i4_res_width; |
2638 | 34.1k | ps_ctxt->i4_ref_height = ps_curr_lyr_res_prms->i4_res_height; |
2639 | | |
2640 | | /* assert on max ranges of width and shift values */ |
2641 | 34.1k | if((ps_lyr_ctxt->i4_curr_width > H264_MAX_FRAME_WIDTH) || |
2642 | 34.1k | (ps_lyr_ctxt->i4_ref_width > H264_MAX_FRAME_WIDTH) || |
2643 | 34.1k | (ps_lyr_ctxt->i4_curr_height > H264_MAX_FRAME_HEIGHT) || |
2644 | 34.1k | (ps_lyr_ctxt->i4_ref_height > H264_MAX_FRAME_HEIGHT)) |
2645 | 0 | { |
2646 | 0 | return NOT_OK; |
2647 | 0 | } |
2648 | 34.1k | return OK; |
2649 | 34.1k | } |