1 |
/************************************************************************** |
/***************************************************************************** |
2 |
* |
* |
3 |
* XVID MPEG-4 VIDEO CODEC |
* XVID MPEG-4 VIDEO CODEC |
4 |
* 8bit<->16bit transfer |
* - 8bit<->16bit transfer - |
5 |
* |
* |
6 |
* This program is an implementation of a part of one or more MPEG-4 |
* Copyright(C) 2001-2003 Peter Ross <pross@xvid.org> |
|
* Video tools as specified in ISO/IEC 14496-2 standard. Those intending |
|
|
* to use this software module in hardware or software products are |
|
|
* advised that its use may infringe existing patents or copyrights, and |
|
|
* any such use would be at such party's own risk. The original |
|
|
* developer of this software module and his/her company, and subsequent |
|
|
* editors and their companies, will have no liability for use of this |
|
|
* software or modifications or derivatives thereof. |
|
7 |
* |
* |
8 |
* This program is free software; you can redistribute it and/or modify |
* This program is free software; you can redistribute it and/or modify |
9 |
* it under the terms of the GNU General Public License as published by |
* it under the terms of the GNU General Public License as published by |
17 |
* |
* |
18 |
* You should have received a copy of the GNU General Public License |
* You should have received a copy of the GNU General Public License |
19 |
* along with this program; if not, write to the Free Software |
* along with this program; if not, write to the Free Software |
20 |
* Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. |
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA |
|
* |
|
|
*************************************************************************/ |
|
|
|
|
|
/************************************************************************** |
|
21 |
* |
* |
22 |
* History: |
* $Id$ |
23 |
* |
* |
24 |
* 14.04.2002 added transfer_8to16sub2 |
****************************************************************************/ |
|
* 07.01.2002 merge functions from compensate; rename functions |
|
|
* 22.12.2001 transfer_8to8add16 limit fix |
|
|
* 07.11.2001 initial version; (c)2001 peter ross <pross@cs.rmit.edu.au> |
|
25 |
|
|
26 |
*************************************************************************/ |
#include "../global.h" |
27 |
|
#include "mem_transfer.h" |
28 |
|
|
29 |
|
/* Function pointers - Initialized in the xvid.c module */ |
30 |
|
|
31 |
|
TRANSFER_8TO16COPY_PTR transfer_8to16copy; |
32 |
|
TRANSFER_16TO8COPY_PTR transfer_16to8copy; |
33 |
|
|
34 |
TRANSFER_8TO16SUB_PTR transfer_8to16sub; |
TRANSFER_8TO16SUB_PTR transfer_8to16sub; |
35 |
|
TRANSFER_8TO16SUBRO_PTR transfer_8to16subro; |
36 |
TRANSFER_8TO16SUB2_PTR transfer_8to16sub2; |
TRANSFER_8TO16SUB2_PTR transfer_8to16sub2; |
37 |
TRANSFER_16TO8ADD_PTR transfer_16to8add; |
TRANSFER_16TO8ADD_PTR transfer_16to8add; |
38 |
|
|
39 |
// function pointers |
TRANSFER8X8_COPY_PTR transfer8x8_copy; |
40 |
|
|
41 |
|
|
42 |
/***************************************************************************** |
/***************************************************************************** |
43 |
* |
* |
44 |
TRANSFER_8TO16SUB_PTR transfer_8to16sub; |
* All these functions are used to transfer data from a 8 bit data array |
45 |
* to a 16 bit data array. |
* to a 16 bit data array. |
46 |
TRANSFER_16TO8ADD_PTR transfer_16to8add; |
* |
47 |
* This is typically used during motion compensation, that's why some |
* This is typically used during motion compensation, that's why some |
48 |
* functions also do the addition/substraction of another buffer during the |
* functions also do the addition/substraction of another buffer during the |
49 |
* so called transfer. |
* so called transfer. |
50 |
* |
* |
51 |
|
****************************************************************************/ |
52 |
|
|
53 |
|
/* |
54 |
|
* SRC - the source buffer |
55 |
|
* DST - the destination buffer |
56 |
|
* |
57 |
|
* Then the function does the 8->16 bit transfer and this serie of operations : |
58 |
|
* |
59 |
|
* SRC (8bit) = SRC |
60 |
|
* DST (16bit) = SRC |
61 |
*/ |
*/ |
62 |
|
void |
63 |
|
transfer_8to16copy_c(int16_t * const dst, |
64 |
|
const uint8_t * const src, |
65 |
|
uint32_t stride) |
66 |
|
{ |
67 |
|
uint32_t i, j; |
68 |
|
|
69 |
|
for (j = 0; j < 8; j++) { |
70 |
|
for (i = 0; i < 8; i++) { |
71 |
dst[j * 8 + i] = (int16_t) src[j * stride + i]; |
dst[j * 8 + i] = (int16_t) src[j * stride + i]; |
72 |
} |
} |
73 |
} |
} |
82 |
* SRC (16bit) = SRC |
* SRC (16bit) = SRC |
83 |
* DST (8bit) = max(min(SRC, 255), 0) |
* DST (8bit) = max(min(SRC, 255), 0) |
84 |
*/ |
*/ |
85 |
|
void |
86 |
|
transfer_16to8copy_c(uint8_t * const dst, |
87 |
|
const int16_t * const src, |
88 |
|
uint32_t stride) |
89 |
|
{ |
90 |
|
uint32_t i, j; |
91 |
|
|
92 |
|
for (j = 0; j < 8; j++) { |
93 |
|
for (i = 0; i < 8; i++) { |
94 |
int16_t pixel = src[j * 8 + i]; |
int16_t pixel = src[j * 8 + i]; |
95 |
|
|
96 |
if (pixel < 0) { |
if (pixel < 0) { |
116 |
* R (8bit) = R |
* R (8bit) = R |
117 |
* C (8bit) = R |
* C (8bit) = R |
118 |
* DCT (16bit) = C - R |
* DCT (16bit) = C - R |
|
perform motion compensation (and 8bit->16bit dct transfer) |
|
|
... with write back |
|
119 |
*/ |
*/ |
120 |
|
void |
121 |
|
transfer_8to16sub_c(int16_t * const dct, |
122 |
|
uint8_t * const cur, |
123 |
|
const uint8_t * ref, |
124 |
|
const uint32_t stride) |
125 |
|
{ |
126 |
|
uint32_t i, j; |
127 |
|
|
128 |
|
for (j = 0; j < 8; j++) { |
129 |
for (i = 0; i < 8; i++) { |
for (i = 0; i < 8; i++) { |
130 |
uint8_t c = cur[j * stride + i]; |
uint8_t c = cur[j * stride + i]; |
131 |
uint8_t r = ref[j * stride + i]; |
uint8_t r = ref[j * stride + i]; |
137 |
} |
} |
138 |
|
|
139 |
|
|
140 |
|
void |
141 |
|
transfer_8to16subro_c(int16_t * const dct, |
142 |
|
const uint8_t * const cur, |
143 |
|
const uint8_t * ref, |
144 |
|
const uint32_t stride) |
145 |
|
{ |
146 |
|
uint32_t i, j; |
147 |
|
|
148 |
|
for (j = 0; j < 8; j++) { |
149 |
|
for (i = 0; i < 8; i++) { |
150 |
|
uint8_t c = cur[j * stride + i]; |
151 |
|
uint8_t r = ref[j * stride + i]; |
152 |
|
dct[j * 8 + i] = (int16_t) c - (int16_t) r; |
153 |
|
} |
154 |
|
} |
155 |
|
} |
156 |
|
|
157 |
|
|
158 |
|
|
159 |
/* |
/* |
160 |
* C - the current buffer |
* C - the current buffer |
161 |
* R1 - the 1st reference buffer |
* R1 - the 1st reference buffer |
166 |
* |
* |
167 |
* R1 (8bit) = R1 |
* R1 (8bit) = R1 |
168 |
* R2 (8bit) = R2 |
* R2 (8bit) = R2 |
169 |
performs bi motion compensation (and 8bit->16bit dct transfer) |
* R (temp) = min((R1 + R2)/2, 255) |
170 |
.. but does not write back |
* DCT (16bit)= C - R |
171 |
|
* C (8bit) = R |
172 |
*/ |
*/ |
173 |
|
void |
174 |
|
transfer_8to16sub2_c(int16_t * const dct, |
175 |
|
uint8_t * const cur, |
176 |
|
const uint8_t * ref1, |
177 |
|
const uint8_t * ref2, |
178 |
|
const uint32_t stride) |
179 |
|
{ |
180 |
|
uint32_t i, j; |
181 |
|
|
182 |
for (j = 0; j < 8; j++) { |
for (j = 0; j < 8; j++) { |
183 |
for (i = 0; i < 8; i++) { |
for (i = 0; i < 8; i++) { |
184 |
uint8_t c = cur[j * stride + i]; |
uint8_t c = cur[j * stride + i]; |
187 |
if (r > 255) { |
if (r > 255) { |
188 |
r = 255; |
r = 255; |
189 |
} |
} |
190 |
//cur[j * stride + i] = r; |
cur[j * stride + i] = r; |
191 |
dct[j * 8 + i] = (int16_t) c - (int16_t) r; |
dct[j * 8 + i] = (int16_t) c - (int16_t) r; |
192 |
} |
} |
193 |
} |
} |
203 |
* SRC (16bit) = SRC |
* SRC (16bit) = SRC |
204 |
* DST (8bit) = max(min(DST+SRC, 255), 0) |
* DST (8bit) = max(min(DST+SRC, 255), 0) |
205 |
*/ |
*/ |
206 |
|
void |
207 |
|
transfer_16to8add_c(uint8_t * const dst, |
208 |
|
const int16_t * const src, |
209 |
|
uint32_t stride) |
210 |
|
{ |
211 |
|
uint32_t i, j; |
212 |
|
|
213 |
|
for (j = 0; j < 8; j++) { |
214 |
|
for (i = 0; i < 8; i++) { |
215 |
int16_t pixel = (int16_t) dst[j * stride + i] + src[j * 8 + i]; |
int16_t pixel = (int16_t) dst[j * stride + i] + src[j * 8 + i]; |
216 |
|
|
217 |
if (pixel < 0) { |
if (pixel < 0) { |
233 |
* SRC (8bit) = SRC |
* SRC (8bit) = SRC |
234 |
* DST (8bit) = SRC |
* DST (8bit) = SRC |
235 |
*/ |
*/ |
236 |
|
void |
237 |
|
transfer8x8_copy_c(uint8_t * const dst, |
238 |
|
const uint8_t * const src, |
239 |
|
const uint32_t stride) |
240 |
|
{ |
241 |
|
uint32_t j; |
242 |
|
|
243 |
|
for (j = 0; j < 8; j++) { |
244 |
|
uint32_t *d= (uint32_t*)(dst + j*stride); |
245 |
dst[j * stride + i] = src[j * stride + i]; |
const uint32_t *s = (const uint32_t*)(src + j*stride); |
246 |
} |
*(d+0) = *(s+0); |
247 |
|
*(d+1) = *(s+1); |
248 |
} |
} |
249 |
} |
} |