3 |
* XVID MPEG-4 VIDEO CODEC |
* XVID MPEG-4 VIDEO CODEC |
4 |
* - Inverse DCT - |
* - Inverse DCT - |
5 |
* |
* |
6 |
* These routines are from Independent JPEG Group's free JPEG software |
* Copyright (C) 2006-2011 Xvid Solutions GmbH |
|
* Copyright (C) 1991-1998, Thomas G. Lane (see the file README.IJG) |
|
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 |
23 |
* |
* |
24 |
****************************************************************************/ |
****************************************************************************/ |
25 |
|
|
|
/* Copyright (C) 1996, MPEG Software Simulation Group. All Rights Reserved. */ |
|
|
|
|
26 |
/* |
/* |
27 |
* Disclaimer of Warranty |
* Authors: Skal |
|
* |
|
|
* These software programs are available to the user without any license fee or |
|
|
* royalty on an "as is" basis. The MPEG Software Simulation Group disclaims |
|
|
* any and all warranties, whether express, implied, or statuary, including any |
|
|
* implied warranties or merchantability or of fitness for a particular |
|
|
* purpose. In no event shall the copyright-holder be liable for any |
|
|
* incidental, punitive, or consequential damages of any kind whatsoever |
|
|
* arising from the use of these programs. |
|
|
* |
|
|
* This disclaimer of warranty extends to the user of these programs and user's |
|
|
* customers, employees, agents, transferees, successors, and assigns. |
|
28 |
* |
* |
29 |
* The MPEG Software Simulation Group does not represent or warrant that the |
* Walken IDCT |
30 |
* programs furnished hereunder are free of infringement of any third-party |
* Alternative idct implementations for decoding compatibility |
|
* patents. |
|
31 |
* |
* |
32 |
* Commercial implementations of MPEG-1 and MPEG-2 video, including shareware, |
* NOTE: this "C" version is not the original one, |
33 |
* are subject to royalty fees to patent holders. Many of these patents are |
* but is modified to yield the same error profile |
34 |
* general enough such that they are unavoidable regardless of implementation |
* than the MMX version. |
|
* design. |
|
35 |
* |
* |
36 |
* MPEG2AVI |
************************************************************************/ |
|
* -------- |
|
|
* v0.16B33 renamed the initialization function to init_idct_int32() |
|
|
* v0.16B32 removed the unused idct_row() and idct_col() functions |
|
|
* v0.16B3 changed var declarations to static, to enforce data align |
|
|
* v0.16B22 idct_FAST() renamed to idct_int32() |
|
|
* also merged idct_FAST() into a single function, to help VC++ |
|
|
* optimize it. |
|
|
* |
|
|
* v0.14 changed int to long, to avoid confusion when compiling on x86 |
|
|
* platform ( in VC++ "int" -> 32bits ) |
|
|
*/ |
|
|
|
|
|
/**********************************************************/ |
|
|
/* inverse two dimensional DCT, Chen-Wang algorithm */ |
|
|
/* (cf. IEEE ASSP-32, pp. 803-816, Aug. 1984) */ |
|
|
/* 32-bit integer arithmetic (8 bit coefficients) */ |
|
|
/* 11 mults, 29 adds per DCT */ |
|
|
/* sE, 18.8.91 */ |
|
|
/**********************************************************/ |
|
|
/* coefficients extended to 12 bit for IEEE1180-1990 */ |
|
|
/* compliance sE, 2.1.94 */ |
|
|
/**********************************************************/ |
|
|
|
|
|
/* this code assumes >> to be a two's-complement arithmetic */ |
|
|
/* right shift: (-2)>>1 == -1 , (-3)>>1 == -2 */ |
|
37 |
|
|
38 |
#include "idct.h" |
#include "idct.h" |
39 |
|
|
40 |
#define W1 2841 /* 2048*sqrt(2)*cos(1*pi/16) */ |
/* function pointer */ |
41 |
#define W2 2676 /* 2048*sqrt(2)*cos(2*pi/16) */ |
idctFuncPtr idct; |
|
#define W3 2408 /* 2048*sqrt(2)*cos(3*pi/16) */ |
|
|
#define W5 1609 /* 2048*sqrt(2)*cos(5*pi/16) */ |
|
|
#define W6 1108 /* 2048*sqrt(2)*cos(6*pi/16) */ |
|
|
#define W7 565 /* 2048*sqrt(2)*cos(7*pi/16) */ |
|
|
|
|
|
/* private data |
|
|
* Initialized by idct_int32_init so it's mostly RO data, |
|
|
* doesn't hurt thread safety */ |
|
|
static short iclip[1024]; /* clipping table */ |
|
|
static short *iclp; |
|
42 |
|
|
43 |
/* private prototypes */ |
#define XVID_DSP_CLIP_255(x) ( ((x)&~255) ? ((-(x)) >> (8*sizeof((x))-1))&0xff : (x) ) |
44 |
|
|
45 |
/* row (horizontal) IDCT |
#define ROW_SHIFT 11 |
46 |
* |
#define COL_SHIFT 6 |
47 |
* 7 pi 1 |
|
48 |
* dst[k] = sum c[l] * src[l] * cos( -- * ( k + - ) * l ) |
// #define FIX(x) (int)((x) * (1<<ROW_SHIFT)) |
49 |
* l=0 8 2 |
#define Rnd0 65536 // 1<<(COL_SHIFT+ROW_SHIFT-1); |
50 |
* |
#define Rnd1 3597 // FIX (1.75683487303); |
51 |
* where: c[0] = 128 |
#define Rnd2 2260 // FIX (1.10355339059); |
52 |
* c[1..7] = 128*sqrt(2) |
#define Rnd3 1203 // FIX (0.587788325588); |
53 |
*/ |
#define Rnd4 0 |
54 |
|
#define Rnd5 120 // FIX (0.058658283817); |
55 |
#if 0 |
#define Rnd6 512 // FIX (0.25); |
56 |
static void idctrow(blk) |
#define Rnd7 512 // FIX (0.25); |
57 |
short *blk; |
#undef FIX |
58 |
{ |
|
59 |
int X0, X1, X2, X3, X4, X5, X6, X7, X8; |
static const int Tab04[] = { 22725, 21407, 19266, 16384, 12873, 8867, 4520 }; |
60 |
|
static const int Tab17[] = { 31521, 29692, 26722, 22725, 17855, 12299, 6270 }; |
61 |
|
static const int Tab26[] = { 29692, 27969, 25172, 21407, 16819, 11585, 5906 }; |
62 |
|
static const int Tab35[] = { 26722, 25172, 22654, 19266, 15137, 10426, 5315 }; |
63 |
|
|
64 |
/* shortcut */ |
static int Idct_Row(short * In, const int * const Tab, int Rnd) |
|
if (!((X1 = blk[4]<<11) | (X2 = blk[6]) | (X3 = blk[2]) | |
|
|
(X4 = blk[1]) | (X5 = blk[7]) | (X6 = blk[5]) | (X7 = blk[3]))) |
|
65 |
{ |
{ |
66 |
blk[0]=blk[1]=blk[2]=blk[3]=blk[4]=blk[5]=blk[6]=blk[7]=blk[0]<<3; |
const int C1 = Tab[0]; |
67 |
return; |
const int C2 = Tab[1]; |
68 |
|
const int C3 = Tab[2]; |
69 |
|
const int C4 = Tab[3]; |
70 |
|
const int C5 = Tab[4]; |
71 |
|
const int C6 = Tab[5]; |
72 |
|
const int C7 = Tab[6]; |
73 |
|
|
74 |
|
const int Right = In[5]|In[6]|In[7]; |
75 |
|
const int Left = In[1]|In[2]|In[3]; |
76 |
|
if (!(Right | In[4])) |
77 |
|
{ |
78 |
|
const int K = C4*In[0] + Rnd; |
79 |
|
if (Left) |
80 |
|
{ |
81 |
|
const int a0 = K + C2*In[2]; |
82 |
|
const int a1 = K + C6*In[2]; |
83 |
|
const int a2 = K - C6*In[2]; |
84 |
|
const int a3 = K - C2*In[2]; |
85 |
|
|
86 |
|
const int b0 = C1*In[1] + C3*In[3]; |
87 |
|
const int b1 = C3*In[1] - C7*In[3]; |
88 |
|
const int b2 = C5*In[1] - C1*In[3]; |
89 |
|
const int b3 = C7*In[1] - C5*In[3]; |
90 |
|
|
91 |
|
In[0] = (a0 + b0) >> ROW_SHIFT; |
92 |
|
In[1] = (a1 + b1) >> ROW_SHIFT; |
93 |
|
In[2] = (a2 + b2) >> ROW_SHIFT; |
94 |
|
In[3] = (a3 + b3) >> ROW_SHIFT; |
95 |
|
In[4] = (a3 - b3) >> ROW_SHIFT; |
96 |
|
In[5] = (a2 - b2) >> ROW_SHIFT; |
97 |
|
In[6] = (a1 - b1) >> ROW_SHIFT; |
98 |
|
In[7] = (a0 - b0) >> ROW_SHIFT; |
99 |
} |
} |
100 |
|
else |
101 |
X0 = (blk[0]<<11) + 128; /* for proper rounding in the fourth stage */ |
{ |
102 |
|
const int a0 = K >> ROW_SHIFT; |
103 |
/* first stage */ |
if (a0) { |
104 |
X8 = W7*(X4+X5); |
In[0] = In[1] = In[2] = In[3] = |
105 |
X4 = X8 + (W1-W7)*X4; |
In[4] = In[5] = In[6] = In[7] = a0; |
|
X5 = X8 - (W1+W7)*X5; |
|
|
X8 = W3*(X6+X7); |
|
|
X6 = X8 - (W3-W5)*X6; |
|
|
X7 = X8 - (W3+W5)*X7; |
|
|
|
|
|
/* second stage */ |
|
|
X8 = X0 + X1; |
|
|
X0 -= X1; |
|
|
X1 = W6*(X3+X2); |
|
|
X2 = X1 - (W2+W6)*X2; |
|
|
X3 = X1 + (W2-W6)*X3; |
|
|
X1 = X4 + X6; |
|
|
X4 -= X6; |
|
|
X6 = X5 + X7; |
|
|
X5 -= X7; |
|
|
|
|
|
/* third stage */ |
|
|
X7 = X8 + X3; |
|
|
X8 -= X3; |
|
|
X3 = X0 + X2; |
|
|
X0 -= X2; |
|
|
X2 = (181*(X4+X5)+128)>>8; |
|
|
X4 = (181*(X4-X5)+128)>>8; |
|
|
|
|
|
/* fourth stage */ |
|
|
blk[0] = (X7+X1)>>8; |
|
|
blk[1] = (X3+X2)>>8; |
|
|
blk[2] = (X0+X4)>>8; |
|
|
blk[3] = (X8+X6)>>8; |
|
|
blk[4] = (X8-X6)>>8; |
|
|
blk[5] = (X0-X4)>>8; |
|
|
blk[6] = (X3-X2)>>8; |
|
|
blk[7] = (X7-X1)>>8; |
|
106 |
} |
} |
107 |
#endif |
else return 0; |
108 |
|
} |
109 |
/* column (vertical) IDCT |
} |
110 |
* |
else if (!(Left|Right)) |
|
* 7 pi 1 |
|
|
* dst[8*k] = sum c[l] * src[8*l] * cos( -- * ( k + - ) * l ) |
|
|
* l=0 8 2 |
|
|
* |
|
|
* where: c[0] = 1/1024 |
|
|
* c[1..7] = (1/1024)*sqrt(2) |
|
|
*/ |
|
|
|
|
|
#if 0 |
|
|
static void idctcol(blk) |
|
|
short *blk; |
|
111 |
{ |
{ |
112 |
int X0, X1, X2, X3, X4, X5, X6, X7, X8; |
const int a0 = (Rnd + C4*(In[0]+In[4])) >> ROW_SHIFT; |
113 |
|
const int a1 = (Rnd + C4*(In[0]-In[4])) >> ROW_SHIFT; |
114 |
|
|
115 |
/* shortcut */ |
In[0] = a0; |
116 |
if (!((X1 = (blk[8*4]<<8)) | (X2 = blk[8*6]) | (X3 = blk[8*2]) | |
In[3] = a0; |
117 |
(X4 = blk[8*1]) | (X5 = blk[8*7]) | (X6 = blk[8*5]) | (X7 = blk[8*3]))) |
In[4] = a0; |
118 |
|
In[7] = a0; |
119 |
|
In[1] = a1; |
120 |
|
In[2] = a1; |
121 |
|
In[5] = a1; |
122 |
|
In[6] = a1; |
123 |
|
} |
124 |
|
else |
125 |
{ |
{ |
126 |
blk[8*0]=blk[8*1]=blk[8*2]=blk[8*3]=blk[8*4]=blk[8*5]=blk[8*6]=blk[8*7]= |
const int K = C4*In[0] + Rnd; |
127 |
iclp[(blk[8*0]+32)>>6]; |
const int a0 = K + C2*In[2] + C4*In[4] + C6*In[6]; |
128 |
return; |
const int a1 = K + C6*In[2] - C4*In[4] - C2*In[6]; |
129 |
} |
const int a2 = K - C6*In[2] - C4*In[4] + C2*In[6]; |
130 |
|
const int a3 = K - C2*In[2] + C4*In[4] - C6*In[6]; |
131 |
X0 = (blk[8*0]<<8) + 8192; |
|
132 |
|
const int b0 = C1*In[1] + C3*In[3] + C5*In[5] + C7*In[7]; |
133 |
/* first stage */ |
const int b1 = C3*In[1] - C7*In[3] - C1*In[5] - C5*In[7]; |
134 |
X8 = W7*(X4+X5) + 4; |
const int b2 = C5*In[1] - C1*In[3] + C7*In[5] + C3*In[7]; |
135 |
X4 = (X8+(W1-W7)*X4)>>3; |
const int b3 = C7*In[1] - C5*In[3] + C3*In[5] - C1*In[7]; |
136 |
X5 = (X8-(W1+W7)*X5)>>3; |
|
137 |
X8 = W3*(X6+X7) + 4; |
In[0] = (a0 + b0) >> ROW_SHIFT; |
138 |
X6 = (X8-(W3-W5)*X6)>>3; |
In[1] = (a1 + b1) >> ROW_SHIFT; |
139 |
X7 = (X8-(W3+W5)*X7)>>3; |
In[2] = (a2 + b2) >> ROW_SHIFT; |
140 |
|
In[3] = (a3 + b3) >> ROW_SHIFT; |
141 |
/* second stage */ |
In[4] = (a3 - b3) >> ROW_SHIFT; |
142 |
X8 = X0 + X1; |
In[5] = (a2 - b2) >> ROW_SHIFT; |
143 |
X0 -= X1; |
In[6] = (a1 - b1) >> ROW_SHIFT; |
144 |
X1 = W6*(X3+X2) + 4; |
In[7] = (a0 - b0) >> ROW_SHIFT; |
145 |
X2 = (X1-(W2+W6)*X2)>>3; |
} |
146 |
X3 = (X1+(W2-W6)*X3)>>3; |
return 1; |
|
X1 = X4 + X6; |
|
|
X4 -= X6; |
|
|
X6 = X5 + X7; |
|
|
X5 -= X7; |
|
|
|
|
|
/* third stage */ |
|
|
X7 = X8 + X3; |
|
|
X8 -= X3; |
|
|
X3 = X0 + X2; |
|
|
X0 -= X2; |
|
|
X2 = (181*(X4+X5)+128)>>8; |
|
|
X4 = (181*(X4-X5)+128)>>8; |
|
|
|
|
|
/* fourth stage */ |
|
|
blk[8*0] = iclp[(X7+X1)>>14]; |
|
|
blk[8*1] = iclp[(X3+X2)>>14]; |
|
|
blk[8*2] = iclp[(X0+X4)>>14]; |
|
|
blk[8*3] = iclp[(X8+X6)>>14]; |
|
|
blk[8*4] = iclp[(X8-X6)>>14]; |
|
|
blk[8*5] = iclp[(X0-X4)>>14]; |
|
|
blk[8*6] = iclp[(X3-X2)>>14]; |
|
|
blk[8*7] = iclp[(X7-X1)>>14]; |
|
147 |
} |
} |
|
#endif |
|
148 |
|
|
149 |
/* function pointer */ |
#define Tan1 0x32ec |
150 |
idctFuncPtr idct; |
#define Tan2 0x6a0a |
151 |
|
#define Tan3 0xab0e |
152 |
|
#define Sqrt2 0x5a82 |
153 |
|
|
154 |
|
#define MULT(c,x, n) ( ((c) * (x)) >> (n) ) |
155 |
|
// 12b version => #define MULT(c,x, n) ( (((c)>>3) * (x)) >> ((n)-3) ) |
156 |
|
// 12b zero-testing version: |
157 |
|
|
158 |
|
#define BUTF(a, b, tmp) \ |
159 |
|
(tmp) = (a)+(b); \ |
160 |
|
(b) = (a)-(b); \ |
161 |
|
(a) = (tmp) |
162 |
|
|
163 |
|
#define LOAD_BUTF(m1, m2, a, b, tmp, S) \ |
164 |
|
(m1) = (S)[(a)] + (S)[(b)]; \ |
165 |
|
(m2) = (S)[(a)] - (S)[(b)] |
166 |
|
|
167 |
/* two dimensional inverse discrete cosine transform */ |
static void Idct_Col_8(short * const In) |
|
void |
|
|
idct_int32(short *const block) |
|
168 |
{ |
{ |
169 |
|
int mm0, mm1, mm2, mm3, mm4, mm5, mm6, mm7, Spill; |
170 |
|
|
171 |
/* |
// odd |
|
* idct_int32_init() must be called before the first call to this |
|
|
* function! |
|
|
*/ |
|
|
|
|
172 |
|
|
173 |
#if 0 |
mm4 = (int)In[7*8]; |
174 |
int i; |
mm5 = (int)In[5*8]; |
175 |
long i; |
mm6 = (int)In[3*8]; |
176 |
|
mm7 = (int)In[1*8]; |
177 |
|
|
178 |
|
mm0 = MULT(Tan1, mm4, 16) + mm7; |
179 |
|
mm1 = MULT(Tan1, mm7, 16) - mm4; |
180 |
|
mm2 = MULT(Tan3, mm5, 16) + mm6; |
181 |
|
mm3 = MULT(Tan3, mm6, 16) - mm5; |
182 |
|
|
183 |
|
mm7 = mm0 + mm2; |
184 |
|
mm4 = mm1 - mm3; |
185 |
|
mm0 = mm0 - mm2; |
186 |
|
mm1 = mm1 + mm3; |
187 |
|
mm6 = mm0 + mm1; |
188 |
|
mm5 = mm0 - mm1; |
189 |
|
mm5 = 2*MULT(Sqrt2, mm5, 16); // 2*sqrt2 |
190 |
|
mm6 = 2*MULT(Sqrt2, mm6, 16); // Watch out: precision loss but done to match |
191 |
|
// the pmulhw used in mmx/sse versions |
192 |
|
|
193 |
|
// even |
194 |
|
|
195 |
|
mm1 = (int)In[2*8]; |
196 |
|
mm2 = (int)In[6*8]; |
197 |
|
mm3 = MULT(Tan2,mm2, 16) + mm1; |
198 |
|
mm2 = MULT(Tan2,mm1, 16) - mm2; |
199 |
|
|
200 |
|
LOAD_BUTF(mm0, mm1, 0*8, 4*8, Spill, In); |
201 |
|
|
202 |
|
BUTF(mm0, mm3, Spill); |
203 |
|
BUTF(mm0, mm7, Spill); |
204 |
|
In[8*0] = (int16_t) (mm0 >> COL_SHIFT); |
205 |
|
In[8*7] = (int16_t) (mm7 >> COL_SHIFT); |
206 |
|
BUTF(mm3, mm4, mm0); |
207 |
|
In[8*3] = (int16_t) (mm3 >> COL_SHIFT); |
208 |
|
In[8*4] = (int16_t) (mm4 >> COL_SHIFT); |
209 |
|
|
210 |
|
BUTF(mm1, mm2, mm0); |
211 |
|
BUTF(mm1, mm6, mm0); |
212 |
|
In[8*1] = (int16_t) (mm1 >> COL_SHIFT); |
213 |
|
In[8*6] = (int16_t) (mm6 >> COL_SHIFT); |
214 |
|
BUTF(mm2, mm5, mm0); |
215 |
|
In[8*2] = (int16_t) (mm2 >> COL_SHIFT); |
216 |
|
In[8*5] = (int16_t) (mm5 >> COL_SHIFT); |
217 |
|
} |
218 |
|
|
219 |
for (i=0; i<8; i++) |
static void Idct_Col_4(short * const In) |
220 |
idctrow(block+8*i); |
{ |
221 |
|
int mm0, mm1, mm2, mm3, mm4, mm5, mm6, mm7, Spill; |
222 |
|
|
223 |
for (i=0; i<8; i++) |
// odd |
|
idctcol(block+i); |
|
|
#endif |
|
224 |
|
|
225 |
short *blk; |
mm0 = (int)In[1*8]; |
226 |
long i; |
mm2 = (int)In[3*8]; |
|
long X0, X1, X2, X3, X4, X5, X6, X7, X8; |
|
227 |
|
|
228 |
|
mm1 = MULT(Tan1, mm0, 16); |
229 |
|
mm3 = MULT(Tan3, mm2, 16); |
230 |
|
|
231 |
|
mm7 = mm0 + mm2; |
232 |
|
mm4 = mm1 - mm3; |
233 |
|
mm0 = mm0 - mm2; |
234 |
|
mm1 = mm1 + mm3; |
235 |
|
mm6 = mm0 + mm1; |
236 |
|
mm5 = mm0 - mm1; |
237 |
|
mm6 = 2*MULT(Sqrt2, mm6, 16); // 2*sqrt2 |
238 |
|
mm5 = 2*MULT(Sqrt2, mm5, 16); |
239 |
|
|
240 |
|
// even |
241 |
|
|
242 |
|
mm0 = mm1 = (int)In[0*8]; |
243 |
|
mm3 = (int)In[2*8]; |
244 |
|
mm2 = MULT(Tan2,mm3, 16); |
245 |
|
|
246 |
|
BUTF(mm0, mm3, Spill); |
247 |
|
BUTF(mm0, mm7, Spill); |
248 |
|
In[8*0] = (int16_t) (mm0 >> COL_SHIFT); |
249 |
|
In[8*7] = (int16_t) (mm7 >> COL_SHIFT); |
250 |
|
BUTF(mm3, mm4, mm0); |
251 |
|
In[8*3] = (int16_t) (mm3 >> COL_SHIFT); |
252 |
|
In[8*4] = (int16_t) (mm4 >> COL_SHIFT); |
253 |
|
|
254 |
|
BUTF(mm1, mm2, mm0); |
255 |
|
BUTF(mm1, mm6, mm0); |
256 |
|
In[8*1] = (int16_t) (mm1 >> COL_SHIFT); |
257 |
|
In[8*6] = (int16_t) (mm6 >> COL_SHIFT); |
258 |
|
BUTF(mm2, mm5, mm0); |
259 |
|
In[8*2] = (int16_t) (mm2 >> COL_SHIFT); |
260 |
|
In[8*5] = (int16_t) (mm5 >> COL_SHIFT); |
261 |
|
} |
262 |
|
|
263 |
for (i = 0; i < 8; i++) /* idct rows */ |
static void Idct_Col_3(short * const In) |
264 |
{ |
{ |
265 |
blk = block + (i << 3); |
int mm0, mm1, mm2, mm3, mm4, mm5, mm6, mm7, Spill; |
|
if (! |
|
|
((X1 = blk[4] << 11) | (X2 = blk[6]) | (X3 = blk[2]) | (X4 = |
|
|
blk[1]) | |
|
|
(X5 = blk[7]) | (X6 = blk[5]) | (X7 = blk[3]))) { |
|
|
blk[0] = blk[1] = blk[2] = blk[3] = blk[4] = blk[5] = blk[6] = |
|
|
blk[7] = blk[0] << 3; |
|
|
continue; |
|
|
} |
|
|
|
|
|
X0 = (blk[0] << 11) + 128; /* for proper rounding in the fourth stage */ |
|
|
|
|
|
/* first stage */ |
|
|
X8 = W7 * (X4 + X5); |
|
|
X4 = X8 + (W1 - W7) * X4; |
|
|
X5 = X8 - (W1 + W7) * X5; |
|
|
X8 = W3 * (X6 + X7); |
|
|
X6 = X8 - (W3 - W5) * X6; |
|
|
X7 = X8 - (W3 + W5) * X7; |
|
|
|
|
|
/* second stage */ |
|
|
X8 = X0 + X1; |
|
|
X0 -= X1; |
|
|
X1 = W6 * (X3 + X2); |
|
|
X2 = X1 - (W2 + W6) * X2; |
|
|
X3 = X1 + (W2 - W6) * X3; |
|
|
X1 = X4 + X6; |
|
|
X4 -= X6; |
|
|
X6 = X5 + X7; |
|
|
X5 -= X7; |
|
|
|
|
|
/* third stage */ |
|
|
X7 = X8 + X3; |
|
|
X8 -= X3; |
|
|
X3 = X0 + X2; |
|
|
X0 -= X2; |
|
|
X2 = (181 * (X4 + X5) + 128) >> 8; |
|
|
X4 = (181 * (X4 - X5) + 128) >> 8; |
|
|
|
|
|
/* fourth stage */ |
|
|
|
|
|
blk[0] = (short) ((X7 + X1) >> 8); |
|
|
blk[1] = (short) ((X3 + X2) >> 8); |
|
|
blk[2] = (short) ((X0 + X4) >> 8); |
|
|
blk[3] = (short) ((X8 + X6) >> 8); |
|
|
blk[4] = (short) ((X8 - X6) >> 8); |
|
|
blk[5] = (short) ((X0 - X4) >> 8); |
|
|
blk[6] = (short) ((X3 - X2) >> 8); |
|
|
blk[7] = (short) ((X7 - X1) >> 8); |
|
266 |
|
|
267 |
} /* end for ( i = 0; i < 8; ++i ) IDCT-rows */ |
// odd |
268 |
|
|
269 |
|
mm7 = (int)In[1*8]; |
270 |
|
mm4 = MULT(Tan1, mm7, 16); |
271 |
|
|
272 |
|
mm6 = mm7 + mm4; |
273 |
for (i = 0; i < 8; i++) /* idct columns */ |
mm5 = mm7 - mm4; |
274 |
{ |
mm6 = 2*MULT(Sqrt2, mm6, 16); // 2*sqrt2 |
275 |
blk = block + i; |
mm5 = 2*MULT(Sqrt2, mm5, 16); |
276 |
/* shortcut */ |
|
277 |
if (! |
// even |
278 |
((X1 = (blk[8 * 4] << 8)) | (X2 = blk[8 * 6]) | (X3 = |
|
279 |
blk[8 * |
mm0 = mm1 = (int)In[0*8]; |
280 |
2]) | (X4 = |
mm3 = (int)In[2*8]; |
281 |
blk[8 * |
mm2 = MULT(Tan2,mm3, 16); |
282 |
1]) |
|
283 |
| (X5 = blk[8 * 7]) | (X6 = blk[8 * 5]) | (X7 = blk[8 * 3]))) { |
BUTF(mm0, mm3, Spill); |
284 |
blk[8 * 0] = blk[8 * 1] = blk[8 * 2] = blk[8 * 3] = blk[8 * 4] = |
BUTF(mm0, mm7, Spill); |
285 |
blk[8 * 5] = blk[8 * 6] = blk[8 * 7] = |
In[8*0] = (int16_t) (mm0 >> COL_SHIFT); |
286 |
iclp[(blk[8 * 0] + 32) >> 6]; |
In[8*7] = (int16_t) (mm7 >> COL_SHIFT); |
287 |
continue; |
BUTF(mm3, mm4, mm0); |
288 |
} |
In[8*3] = (int16_t) (mm3 >> COL_SHIFT); |
289 |
|
In[8*4] = (int16_t) (mm4 >> COL_SHIFT); |
290 |
X0 = (blk[8 * 0] << 8) + 8192; |
|
291 |
|
BUTF(mm1, mm2, mm0); |
292 |
/* first stage */ |
BUTF(mm1, mm6, mm0); |
293 |
X8 = W7 * (X4 + X5) + 4; |
In[8*1] = (int16_t) (mm1 >> COL_SHIFT); |
294 |
X4 = (X8 + (W1 - W7) * X4) >> 3; |
In[8*6] = (int16_t) (mm6 >> COL_SHIFT); |
295 |
X5 = (X8 - (W1 + W7) * X5) >> 3; |
BUTF(mm2, mm5, mm0); |
296 |
X8 = W3 * (X6 + X7) + 4; |
In[8*2] = (int16_t) (mm2 >> COL_SHIFT); |
297 |
X6 = (X8 - (W3 - W5) * X6) >> 3; |
In[8*5] = (int16_t) (mm5 >> COL_SHIFT); |
|
X7 = (X8 - (W3 + W5) * X7) >> 3; |
|
|
|
|
|
/* second stage */ |
|
|
X8 = X0 + X1; |
|
|
X0 -= X1; |
|
|
X1 = W6 * (X3 + X2) + 4; |
|
|
X2 = (X1 - (W2 + W6) * X2) >> 3; |
|
|
X3 = (X1 + (W2 - W6) * X3) >> 3; |
|
|
X1 = X4 + X6; |
|
|
X4 -= X6; |
|
|
X6 = X5 + X7; |
|
|
X5 -= X7; |
|
|
|
|
|
/* third stage */ |
|
|
X7 = X8 + X3; |
|
|
X8 -= X3; |
|
|
X3 = X0 + X2; |
|
|
X0 -= X2; |
|
|
X2 = (181 * (X4 + X5) + 128) >> 8; |
|
|
X4 = (181 * (X4 - X5) + 128) >> 8; |
|
|
|
|
|
/* fourth stage */ |
|
|
blk[8 * 0] = iclp[(X7 + X1) >> 14]; |
|
|
blk[8 * 1] = iclp[(X3 + X2) >> 14]; |
|
|
blk[8 * 2] = iclp[(X0 + X4) >> 14]; |
|
|
blk[8 * 3] = iclp[(X8 + X6) >> 14]; |
|
|
blk[8 * 4] = iclp[(X8 - X6) >> 14]; |
|
|
blk[8 * 5] = iclp[(X0 - X4) >> 14]; |
|
|
blk[8 * 6] = iclp[(X3 - X2) >> 14]; |
|
|
blk[8 * 7] = iclp[(X7 - X1) >> 14]; |
|
298 |
} |
} |
299 |
|
|
300 |
} /* end function idct_int32(block) */ |
#undef Tan1 |
301 |
|
#undef Tan2 |
302 |
|
#undef Tan3 |
303 |
|
#undef Sqrt2 |
304 |
|
|
305 |
|
#undef ROW_SHIFT |
306 |
|
#undef COL_SHIFT |
307 |
|
|
308 |
|
////////////////////////////////////////////////////////// |
309 |
|
|
310 |
void |
void idct_int32(short *const In) |
|
idct_int32_init(void) |
|
311 |
{ |
{ |
312 |
int i; |
int i, Rows = 0x07; |
313 |
|
|
314 |
iclp = iclip + 512; |
Idct_Row(In + 0*8, Tab04, Rnd0); |
315 |
for (i = -512; i < 512; i++) |
Idct_Row(In + 1*8, Tab17, Rnd1); |
316 |
iclp[i] = (i < -256) ? -256 : ((i > 255) ? 255 : i); |
Idct_Row(In + 2*8, Tab26, Rnd2); |
317 |
|
if (Idct_Row(In + 3*8, Tab35, Rnd3)) Rows |= 0x08; |
318 |
|
if (Idct_Row(In + 4*8, Tab04, Rnd4)) Rows |= 0x10; |
319 |
|
if (Idct_Row(In + 5*8, Tab35, Rnd5)) Rows |= 0x20; |
320 |
|
if (Idct_Row(In + 6*8, Tab26, Rnd6)) Rows |= 0x40; |
321 |
|
if (Idct_Row(In + 7*8, Tab17, Rnd7)) Rows |= 0x80; |
322 |
|
|
323 |
|
if (Rows&0xf0) { |
324 |
|
for(i=0; i<8; i++) |
325 |
|
Idct_Col_8(In + i); |
326 |
|
} |
327 |
|
else if (Rows&0x08) { |
328 |
|
for(i=0; i<8; i++) |
329 |
|
Idct_Col_4(In + i); |
330 |
|
} |
331 |
|
else { |
332 |
|
for(i=0; i<8; i++) |
333 |
|
Idct_Col_3(In + i); |
334 |
|
} |
335 |
} |
} |