Alien-FreeImage
view release on metacpan or search on metacpan
src/Source/LibJPEG/jidctint.c view on Meta::CPAN
& RANGE_MASK];
outptr[4] = range_limit[(int) RIGHT_SHIFT(tmp13 - tmp0,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
wsptr += DCTSIZE; /* advance pointer to next row */
}
}
#ifdef IDCT_SCALING_SUPPORTED
/*
* Perform dequantization and inverse DCT on one block of coefficients,
* producing a 7x7 output block.
*
* Optimized algorithm with 12 multiplications in the 1-D kernel.
* cK represents sqrt(2) * cos(K*pi/14).
*/
GLOBAL(void)
jpeg_idct_7x7 (j_decompress_ptr cinfo, jpeg_component_info * compptr,
JCOEFPTR coef_block,
JSAMPARRAY output_buf, JDIMENSION output_col)
{
INT32 tmp0, tmp1, tmp2, tmp10, tmp11, tmp12, tmp13;
INT32 z1, z2, z3;
JCOEFPTR inptr;
ISLOW_MULT_TYPE * quantptr;
int * wsptr;
JSAMPROW outptr;
JSAMPLE *range_limit = IDCT_range_limit(cinfo);
int ctr;
int workspace[7*7]; /* buffers data between passes */
SHIFT_TEMPS
/* Pass 1: process columns from input, store into work array. */
inptr = coef_block;
quantptr = (ISLOW_MULT_TYPE *) compptr->dct_table;
wsptr = workspace;
for (ctr = 0; ctr < 7; ctr++, inptr++, quantptr++, wsptr++) {
/* Even part */
tmp13 = DEQUANTIZE(inptr[DCTSIZE*0], quantptr[DCTSIZE*0]);
tmp13 <<= CONST_BITS;
/* Add fudge factor here for final descale. */
tmp13 += ONE << (CONST_BITS-PASS1_BITS-1);
z1 = DEQUANTIZE(inptr[DCTSIZE*2], quantptr[DCTSIZE*2]);
z2 = DEQUANTIZE(inptr[DCTSIZE*4], quantptr[DCTSIZE*4]);
z3 = DEQUANTIZE(inptr[DCTSIZE*6], quantptr[DCTSIZE*6]);
tmp10 = MULTIPLY(z2 - z3, FIX(0.881747734)); /* c4 */
tmp12 = MULTIPLY(z1 - z2, FIX(0.314692123)); /* c6 */
tmp11 = tmp10 + tmp12 + tmp13 - MULTIPLY(z2, FIX(1.841218003)); /* c2+c4-c6 */
tmp0 = z1 + z3;
z2 -= tmp0;
tmp0 = MULTIPLY(tmp0, FIX(1.274162392)) + tmp13; /* c2 */
tmp10 += tmp0 - MULTIPLY(z3, FIX(0.077722536)); /* c2-c4-c6 */
tmp12 += tmp0 - MULTIPLY(z1, FIX(2.470602249)); /* c2+c4+c6 */
tmp13 += MULTIPLY(z2, FIX(1.414213562)); /* c0 */
/* Odd part */
z1 = DEQUANTIZE(inptr[DCTSIZE*1], quantptr[DCTSIZE*1]);
z2 = DEQUANTIZE(inptr[DCTSIZE*3], quantptr[DCTSIZE*3]);
z3 = DEQUANTIZE(inptr[DCTSIZE*5], quantptr[DCTSIZE*5]);
tmp1 = MULTIPLY(z1 + z2, FIX(0.935414347)); /* (c3+c1-c5)/2 */
tmp2 = MULTIPLY(z1 - z2, FIX(0.170262339)); /* (c3+c5-c1)/2 */
tmp0 = tmp1 - tmp2;
tmp1 += tmp2;
tmp2 = MULTIPLY(z2 + z3, - FIX(1.378756276)); /* -c1 */
tmp1 += tmp2;
z2 = MULTIPLY(z1 + z3, FIX(0.613604268)); /* c5 */
tmp0 += z2;
tmp2 += z2 + MULTIPLY(z3, FIX(1.870828693)); /* c3+c1-c5 */
/* Final output stage */
wsptr[7*0] = (int) RIGHT_SHIFT(tmp10 + tmp0, CONST_BITS-PASS1_BITS);
wsptr[7*6] = (int) RIGHT_SHIFT(tmp10 - tmp0, CONST_BITS-PASS1_BITS);
wsptr[7*1] = (int) RIGHT_SHIFT(tmp11 + tmp1, CONST_BITS-PASS1_BITS);
wsptr[7*5] = (int) RIGHT_SHIFT(tmp11 - tmp1, CONST_BITS-PASS1_BITS);
wsptr[7*2] = (int) RIGHT_SHIFT(tmp12 + tmp2, CONST_BITS-PASS1_BITS);
wsptr[7*4] = (int) RIGHT_SHIFT(tmp12 - tmp2, CONST_BITS-PASS1_BITS);
wsptr[7*3] = (int) RIGHT_SHIFT(tmp13, CONST_BITS-PASS1_BITS);
}
/* Pass 2: process 7 rows from work array, store into output array. */
wsptr = workspace;
for (ctr = 0; ctr < 7; ctr++) {
outptr = output_buf[ctr] + output_col;
/* Even part */
/* Add fudge factor here for final descale. */
tmp13 = (INT32) wsptr[0] + (ONE << (PASS1_BITS+2));
tmp13 <<= CONST_BITS;
z1 = (INT32) wsptr[2];
z2 = (INT32) wsptr[4];
z3 = (INT32) wsptr[6];
tmp10 = MULTIPLY(z2 - z3, FIX(0.881747734)); /* c4 */
tmp12 = MULTIPLY(z1 - z2, FIX(0.314692123)); /* c6 */
tmp11 = tmp10 + tmp12 + tmp13 - MULTIPLY(z2, FIX(1.841218003)); /* c2+c4-c6 */
tmp0 = z1 + z3;
z2 -= tmp0;
tmp0 = MULTIPLY(tmp0, FIX(1.274162392)) + tmp13; /* c2 */
tmp10 += tmp0 - MULTIPLY(z3, FIX(0.077722536)); /* c2-c4-c6 */
tmp12 += tmp0 - MULTIPLY(z1, FIX(2.470602249)); /* c2+c4+c6 */
tmp13 += MULTIPLY(z2, FIX(1.414213562)); /* c0 */
/* Odd part */
z1 = (INT32) wsptr[1];
z2 = (INT32) wsptr[3];
z3 = (INT32) wsptr[5];
tmp1 = MULTIPLY(z1 + z2, FIX(0.935414347)); /* (c3+c1-c5)/2 */
tmp2 = MULTIPLY(z1 - z2, FIX(0.170262339)); /* (c3+c5-c1)/2 */
tmp0 = tmp1 - tmp2;
tmp1 += tmp2;
tmp2 = MULTIPLY(z2 + z3, - FIX(1.378756276)); /* -c1 */
tmp1 += tmp2;
z2 = MULTIPLY(z1 + z3, FIX(0.613604268)); /* c5 */
tmp0 += z2;
tmp2 += z2 + MULTIPLY(z3, FIX(1.870828693)); /* c3+c1-c5 */
/* Final output stage */
outptr[0] = range_limit[(int) RIGHT_SHIFT(tmp10 + tmp0,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[6] = range_limit[(int) RIGHT_SHIFT(tmp10 - tmp0,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[1] = range_limit[(int) RIGHT_SHIFT(tmp11 + tmp1,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[5] = range_limit[(int) RIGHT_SHIFT(tmp11 - tmp1,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[2] = range_limit[(int) RIGHT_SHIFT(tmp12 + tmp2,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[4] = range_limit[(int) RIGHT_SHIFT(tmp12 - tmp2,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[3] = range_limit[(int) RIGHT_SHIFT(tmp13,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
wsptr += 7; /* advance pointer to next row */
}
}
/*
* Perform dequantization and inverse DCT on one block of coefficients,
* producing a reduced-size 6x6 output block.
*
* Optimized algorithm with 3 multiplications in the 1-D kernel.
* cK represents sqrt(2) * cos(K*pi/12).
*/
GLOBAL(void)
jpeg_idct_6x6 (j_decompress_ptr cinfo, jpeg_component_info * compptr,
JCOEFPTR coef_block,
JSAMPARRAY output_buf, JDIMENSION output_col)
{
src/Source/LibJPEG/jidctint.c view on Meta::CPAN
/* 1x1 is trivial: just take the DC coefficient divided by 8. */
quantptr = (ISLOW_MULT_TYPE *) compptr->dct_table;
dcval = DEQUANTIZE(coef_block[0], quantptr[0]);
dcval = (int) DESCALE((INT32) dcval, 3);
output_buf[0][output_col] = range_limit[dcval & RANGE_MASK];
}
/*
* Perform dequantization and inverse DCT on one block of coefficients,
* producing a 9x9 output block.
*
* Optimized algorithm with 10 multiplications in the 1-D kernel.
* cK represents sqrt(2) * cos(K*pi/18).
*/
GLOBAL(void)
jpeg_idct_9x9 (j_decompress_ptr cinfo, jpeg_component_info * compptr,
JCOEFPTR coef_block,
JSAMPARRAY output_buf, JDIMENSION output_col)
{
INT32 tmp0, tmp1, tmp2, tmp3, tmp10, tmp11, tmp12, tmp13, tmp14;
INT32 z1, z2, z3, z4;
JCOEFPTR inptr;
ISLOW_MULT_TYPE * quantptr;
int * wsptr;
JSAMPROW outptr;
JSAMPLE *range_limit = IDCT_range_limit(cinfo);
int ctr;
int workspace[8*9]; /* buffers data between passes */
SHIFT_TEMPS
/* Pass 1: process columns from input, store into work array. */
inptr = coef_block;
quantptr = (ISLOW_MULT_TYPE *) compptr->dct_table;
wsptr = workspace;
for (ctr = 0; ctr < 8; ctr++, inptr++, quantptr++, wsptr++) {
/* Even part */
tmp0 = DEQUANTIZE(inptr[DCTSIZE*0], quantptr[DCTSIZE*0]);
tmp0 <<= CONST_BITS;
/* Add fudge factor here for final descale. */
tmp0 += ONE << (CONST_BITS-PASS1_BITS-1);
z1 = DEQUANTIZE(inptr[DCTSIZE*2], quantptr[DCTSIZE*2]);
z2 = DEQUANTIZE(inptr[DCTSIZE*4], quantptr[DCTSIZE*4]);
z3 = DEQUANTIZE(inptr[DCTSIZE*6], quantptr[DCTSIZE*6]);
tmp3 = MULTIPLY(z3, FIX(0.707106781)); /* c6 */
tmp1 = tmp0 + tmp3;
tmp2 = tmp0 - tmp3 - tmp3;
tmp0 = MULTIPLY(z1 - z2, FIX(0.707106781)); /* c6 */
tmp11 = tmp2 + tmp0;
tmp14 = tmp2 - tmp0 - tmp0;
tmp0 = MULTIPLY(z1 + z2, FIX(1.328926049)); /* c2 */
tmp2 = MULTIPLY(z1, FIX(1.083350441)); /* c4 */
tmp3 = MULTIPLY(z2, FIX(0.245575608)); /* c8 */
tmp10 = tmp1 + tmp0 - tmp3;
tmp12 = tmp1 - tmp0 + tmp2;
tmp13 = tmp1 - tmp2 + tmp3;
/* Odd part */
z1 = DEQUANTIZE(inptr[DCTSIZE*1], quantptr[DCTSIZE*1]);
z2 = DEQUANTIZE(inptr[DCTSIZE*3], quantptr[DCTSIZE*3]);
z3 = DEQUANTIZE(inptr[DCTSIZE*5], quantptr[DCTSIZE*5]);
z4 = DEQUANTIZE(inptr[DCTSIZE*7], quantptr[DCTSIZE*7]);
z2 = MULTIPLY(z2, - FIX(1.224744871)); /* -c3 */
tmp2 = MULTIPLY(z1 + z3, FIX(0.909038955)); /* c5 */
tmp3 = MULTIPLY(z1 + z4, FIX(0.483689525)); /* c7 */
tmp0 = tmp2 + tmp3 - z2;
tmp1 = MULTIPLY(z3 - z4, FIX(1.392728481)); /* c1 */
tmp2 += z2 - tmp1;
tmp3 += z2 + tmp1;
tmp1 = MULTIPLY(z1 - z3 - z4, FIX(1.224744871)); /* c3 */
/* Final output stage */
wsptr[8*0] = (int) RIGHT_SHIFT(tmp10 + tmp0, CONST_BITS-PASS1_BITS);
wsptr[8*8] = (int) RIGHT_SHIFT(tmp10 - tmp0, CONST_BITS-PASS1_BITS);
wsptr[8*1] = (int) RIGHT_SHIFT(tmp11 + tmp1, CONST_BITS-PASS1_BITS);
wsptr[8*7] = (int) RIGHT_SHIFT(tmp11 - tmp1, CONST_BITS-PASS1_BITS);
wsptr[8*2] = (int) RIGHT_SHIFT(tmp12 + tmp2, CONST_BITS-PASS1_BITS);
wsptr[8*6] = (int) RIGHT_SHIFT(tmp12 - tmp2, CONST_BITS-PASS1_BITS);
wsptr[8*3] = (int) RIGHT_SHIFT(tmp13 + tmp3, CONST_BITS-PASS1_BITS);
wsptr[8*5] = (int) RIGHT_SHIFT(tmp13 - tmp3, CONST_BITS-PASS1_BITS);
wsptr[8*4] = (int) RIGHT_SHIFT(tmp14, CONST_BITS-PASS1_BITS);
}
/* Pass 2: process 9 rows from work array, store into output array. */
wsptr = workspace;
for (ctr = 0; ctr < 9; ctr++) {
outptr = output_buf[ctr] + output_col;
/* Even part */
/* Add fudge factor here for final descale. */
tmp0 = (INT32) wsptr[0] + (ONE << (PASS1_BITS+2));
tmp0 <<= CONST_BITS;
z1 = (INT32) wsptr[2];
z2 = (INT32) wsptr[4];
z3 = (INT32) wsptr[6];
tmp3 = MULTIPLY(z3, FIX(0.707106781)); /* c6 */
tmp1 = tmp0 + tmp3;
tmp2 = tmp0 - tmp3 - tmp3;
tmp0 = MULTIPLY(z1 - z2, FIX(0.707106781)); /* c6 */
tmp11 = tmp2 + tmp0;
tmp14 = tmp2 - tmp0 - tmp0;
tmp0 = MULTIPLY(z1 + z2, FIX(1.328926049)); /* c2 */
tmp2 = MULTIPLY(z1, FIX(1.083350441)); /* c4 */
tmp3 = MULTIPLY(z2, FIX(0.245575608)); /* c8 */
tmp10 = tmp1 + tmp0 - tmp3;
tmp12 = tmp1 - tmp0 + tmp2;
tmp13 = tmp1 - tmp2 + tmp3;
/* Odd part */
z1 = (INT32) wsptr[1];
z2 = (INT32) wsptr[3];
z3 = (INT32) wsptr[5];
z4 = (INT32) wsptr[7];
z2 = MULTIPLY(z2, - FIX(1.224744871)); /* -c3 */
tmp2 = MULTIPLY(z1 + z3, FIX(0.909038955)); /* c5 */
tmp3 = MULTIPLY(z1 + z4, FIX(0.483689525)); /* c7 */
tmp0 = tmp2 + tmp3 - z2;
tmp1 = MULTIPLY(z3 - z4, FIX(1.392728481)); /* c1 */
tmp2 += z2 - tmp1;
tmp3 += z2 + tmp1;
tmp1 = MULTIPLY(z1 - z3 - z4, FIX(1.224744871)); /* c3 */
/* Final output stage */
outptr[0] = range_limit[(int) RIGHT_SHIFT(tmp10 + tmp0,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[8] = range_limit[(int) RIGHT_SHIFT(tmp10 - tmp0,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[1] = range_limit[(int) RIGHT_SHIFT(tmp11 + tmp1,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[7] = range_limit[(int) RIGHT_SHIFT(tmp11 - tmp1,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[2] = range_limit[(int) RIGHT_SHIFT(tmp12 + tmp2,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[6] = range_limit[(int) RIGHT_SHIFT(tmp12 - tmp2,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[3] = range_limit[(int) RIGHT_SHIFT(tmp13 + tmp3,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[5] = range_limit[(int) RIGHT_SHIFT(tmp13 - tmp3,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[4] = range_limit[(int) RIGHT_SHIFT(tmp14,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
wsptr += 8; /* advance pointer to next row */
}
}
/*
* Perform dequantization and inverse DCT on one block of coefficients,
* producing a 10x10 output block.
src/Source/LibJPEG/jidctint.c view on Meta::CPAN
INT32 tmp20, tmp21, tmp22, tmp23, tmp24;
INT32 z1, z2, z3, z4, z5;
JCOEFPTR inptr;
ISLOW_MULT_TYPE * quantptr;
int * wsptr;
JSAMPROW outptr;
JSAMPLE *range_limit = IDCT_range_limit(cinfo);
int ctr;
int workspace[8*10]; /* buffers data between passes */
SHIFT_TEMPS
/* Pass 1: process columns from input, store into work array. */
inptr = coef_block;
quantptr = (ISLOW_MULT_TYPE *) compptr->dct_table;
wsptr = workspace;
for (ctr = 0; ctr < 8; ctr++, inptr++, quantptr++, wsptr++) {
/* Even part */
z3 = DEQUANTIZE(inptr[DCTSIZE*0], quantptr[DCTSIZE*0]);
z3 <<= CONST_BITS;
/* Add fudge factor here for final descale. */
z3 += ONE << (CONST_BITS-PASS1_BITS-1);
z4 = DEQUANTIZE(inptr[DCTSIZE*4], quantptr[DCTSIZE*4]);
z1 = MULTIPLY(z4, FIX(1.144122806)); /* c4 */
z2 = MULTIPLY(z4, FIX(0.437016024)); /* c8 */
tmp10 = z3 + z1;
tmp11 = z3 - z2;
tmp22 = RIGHT_SHIFT(z3 - ((z1 - z2) << 1), /* c0 = (c4-c8)*2 */
CONST_BITS-PASS1_BITS);
z2 = DEQUANTIZE(inptr[DCTSIZE*2], quantptr[DCTSIZE*2]);
z3 = DEQUANTIZE(inptr[DCTSIZE*6], quantptr[DCTSIZE*6]);
z1 = MULTIPLY(z2 + z3, FIX(0.831253876)); /* c6 */
tmp12 = z1 + MULTIPLY(z2, FIX(0.513743148)); /* c2-c6 */
tmp13 = z1 - MULTIPLY(z3, FIX(2.176250899)); /* c2+c6 */
tmp20 = tmp10 + tmp12;
tmp24 = tmp10 - tmp12;
tmp21 = tmp11 + tmp13;
tmp23 = tmp11 - tmp13;
/* Odd part */
z1 = DEQUANTIZE(inptr[DCTSIZE*1], quantptr[DCTSIZE*1]);
z2 = DEQUANTIZE(inptr[DCTSIZE*3], quantptr[DCTSIZE*3]);
z3 = DEQUANTIZE(inptr[DCTSIZE*5], quantptr[DCTSIZE*5]);
z4 = DEQUANTIZE(inptr[DCTSIZE*7], quantptr[DCTSIZE*7]);
tmp11 = z2 + z4;
tmp13 = z2 - z4;
tmp12 = MULTIPLY(tmp13, FIX(0.309016994)); /* (c3-c7)/2 */
z5 = z3 << CONST_BITS;
z2 = MULTIPLY(tmp11, FIX(0.951056516)); /* (c3+c7)/2 */
z4 = z5 + tmp12;
tmp10 = MULTIPLY(z1, FIX(1.396802247)) + z2 + z4; /* c1 */
tmp14 = MULTIPLY(z1, FIX(0.221231742)) - z2 + z4; /* c9 */
z2 = MULTIPLY(tmp11, FIX(0.587785252)); /* (c1-c9)/2 */
z4 = z5 - tmp12 - (tmp13 << (CONST_BITS - 1));
tmp12 = (z1 - tmp13 - z3) << PASS1_BITS;
tmp11 = MULTIPLY(z1, FIX(1.260073511)) - z2 - z4; /* c3 */
tmp13 = MULTIPLY(z1, FIX(0.642039522)) - z2 + z4; /* c7 */
/* Final output stage */
wsptr[8*0] = (int) RIGHT_SHIFT(tmp20 + tmp10, CONST_BITS-PASS1_BITS);
wsptr[8*9] = (int) RIGHT_SHIFT(tmp20 - tmp10, CONST_BITS-PASS1_BITS);
wsptr[8*1] = (int) RIGHT_SHIFT(tmp21 + tmp11, CONST_BITS-PASS1_BITS);
wsptr[8*8] = (int) RIGHT_SHIFT(tmp21 - tmp11, CONST_BITS-PASS1_BITS);
wsptr[8*2] = (int) (tmp22 + tmp12);
wsptr[8*7] = (int) (tmp22 - tmp12);
wsptr[8*3] = (int) RIGHT_SHIFT(tmp23 + tmp13, CONST_BITS-PASS1_BITS);
wsptr[8*6] = (int) RIGHT_SHIFT(tmp23 - tmp13, CONST_BITS-PASS1_BITS);
wsptr[8*4] = (int) RIGHT_SHIFT(tmp24 + tmp14, CONST_BITS-PASS1_BITS);
wsptr[8*5] = (int) RIGHT_SHIFT(tmp24 - tmp14, CONST_BITS-PASS1_BITS);
}
/* Pass 2: process 10 rows from work array, store into output array. */
wsptr = workspace;
for (ctr = 0; ctr < 10; ctr++) {
outptr = output_buf[ctr] + output_col;
/* Even part */
/* Add fudge factor here for final descale. */
z3 = (INT32) wsptr[0] + (ONE << (PASS1_BITS+2));
z3 <<= CONST_BITS;
z4 = (INT32) wsptr[4];
z1 = MULTIPLY(z4, FIX(1.144122806)); /* c4 */
z2 = MULTIPLY(z4, FIX(0.437016024)); /* c8 */
tmp10 = z3 + z1;
tmp11 = z3 - z2;
tmp22 = z3 - ((z1 - z2) << 1); /* c0 = (c4-c8)*2 */
z2 = (INT32) wsptr[2];
z3 = (INT32) wsptr[6];
z1 = MULTIPLY(z2 + z3, FIX(0.831253876)); /* c6 */
tmp12 = z1 + MULTIPLY(z2, FIX(0.513743148)); /* c2-c6 */
tmp13 = z1 - MULTIPLY(z3, FIX(2.176250899)); /* c2+c6 */
tmp20 = tmp10 + tmp12;
tmp24 = tmp10 - tmp12;
tmp21 = tmp11 + tmp13;
tmp23 = tmp11 - tmp13;
/* Odd part */
z1 = (INT32) wsptr[1];
z2 = (INT32) wsptr[3];
z3 = (INT32) wsptr[5];
z3 <<= CONST_BITS;
z4 = (INT32) wsptr[7];
tmp11 = z2 + z4;
tmp13 = z2 - z4;
tmp12 = MULTIPLY(tmp13, FIX(0.309016994)); /* (c3-c7)/2 */
z2 = MULTIPLY(tmp11, FIX(0.951056516)); /* (c3+c7)/2 */
z4 = z3 + tmp12;
tmp10 = MULTIPLY(z1, FIX(1.396802247)) + z2 + z4; /* c1 */
tmp14 = MULTIPLY(z1, FIX(0.221231742)) - z2 + z4; /* c9 */
z2 = MULTIPLY(tmp11, FIX(0.587785252)); /* (c1-c9)/2 */
z4 = z3 - tmp12 - (tmp13 << (CONST_BITS - 1));
tmp12 = ((z1 - tmp13) << CONST_BITS) - z3;
tmp11 = MULTIPLY(z1, FIX(1.260073511)) - z2 - z4; /* c3 */
tmp13 = MULTIPLY(z1, FIX(0.642039522)) - z2 + z4; /* c7 */
/* Final output stage */
outptr[0] = range_limit[(int) RIGHT_SHIFT(tmp20 + tmp10,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[9] = range_limit[(int) RIGHT_SHIFT(tmp20 - tmp10,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[1] = range_limit[(int) RIGHT_SHIFT(tmp21 + tmp11,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[8] = range_limit[(int) RIGHT_SHIFT(tmp21 - tmp11,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[2] = range_limit[(int) RIGHT_SHIFT(tmp22 + tmp12,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[7] = range_limit[(int) RIGHT_SHIFT(tmp22 - tmp12,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[3] = range_limit[(int) RIGHT_SHIFT(tmp23 + tmp13,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[6] = range_limit[(int) RIGHT_SHIFT(tmp23 - tmp13,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[4] = range_limit[(int) RIGHT_SHIFT(tmp24 + tmp14,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[5] = range_limit[(int) RIGHT_SHIFT(tmp24 - tmp14,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
wsptr += 8; /* advance pointer to next row */
}
}
/*
* Perform dequantization and inverse DCT on one block of coefficients,
* producing a 11x11 output block.
*
* Optimized algorithm with 24 multiplications in the 1-D kernel.
* cK represents sqrt(2) * cos(K*pi/22).
*/
GLOBAL(void)
jpeg_idct_11x11 (j_decompress_ptr cinfo, jpeg_component_info * compptr,
JCOEFPTR coef_block,
JSAMPARRAY output_buf, JDIMENSION output_col)
src/Source/LibJPEG/jidctint.c view on Meta::CPAN
int ctr;
int workspace[8*16]; /* buffers data between passes */
SHIFT_TEMPS
/* Pass 1: process columns from input, store into work array. */
inptr = coef_block;
quantptr = (ISLOW_MULT_TYPE *) compptr->dct_table;
wsptr = workspace;
for (ctr = 0; ctr < 8; ctr++, inptr++, quantptr++, wsptr++) {
/* Even part */
tmp0 = DEQUANTIZE(inptr[DCTSIZE*0], quantptr[DCTSIZE*0]);
tmp0 <<= CONST_BITS;
/* Add fudge factor here for final descale. */
tmp0 += 1 << (CONST_BITS-PASS1_BITS-1);
z1 = DEQUANTIZE(inptr[DCTSIZE*4], quantptr[DCTSIZE*4]);
tmp1 = MULTIPLY(z1, FIX(1.306562965)); /* c4[16] = c2[8] */
tmp2 = MULTIPLY(z1, FIX_0_541196100); /* c12[16] = c6[8] */
tmp10 = tmp0 + tmp1;
tmp11 = tmp0 - tmp1;
tmp12 = tmp0 + tmp2;
tmp13 = tmp0 - tmp2;
z1 = DEQUANTIZE(inptr[DCTSIZE*2], quantptr[DCTSIZE*2]);
z2 = DEQUANTIZE(inptr[DCTSIZE*6], quantptr[DCTSIZE*6]);
z3 = z1 - z2;
z4 = MULTIPLY(z3, FIX(0.275899379)); /* c14[16] = c7[8] */
z3 = MULTIPLY(z3, FIX(1.387039845)); /* c2[16] = c1[8] */
tmp0 = z3 + MULTIPLY(z2, FIX_2_562915447); /* (c6+c2)[16] = (c3+c1)[8] */
tmp1 = z4 + MULTIPLY(z1, FIX_0_899976223); /* (c6-c14)[16] = (c3-c7)[8] */
tmp2 = z3 - MULTIPLY(z1, FIX(0.601344887)); /* (c2-c10)[16] = (c1-c5)[8] */
tmp3 = z4 - MULTIPLY(z2, FIX(0.509795579)); /* (c10-c14)[16] = (c5-c7)[8] */
tmp20 = tmp10 + tmp0;
tmp27 = tmp10 - tmp0;
tmp21 = tmp12 + tmp1;
tmp26 = tmp12 - tmp1;
tmp22 = tmp13 + tmp2;
tmp25 = tmp13 - tmp2;
tmp23 = tmp11 + tmp3;
tmp24 = tmp11 - tmp3;
/* Odd part */
z1 = DEQUANTIZE(inptr[DCTSIZE*1], quantptr[DCTSIZE*1]);
z2 = DEQUANTIZE(inptr[DCTSIZE*3], quantptr[DCTSIZE*3]);
z3 = DEQUANTIZE(inptr[DCTSIZE*5], quantptr[DCTSIZE*5]);
z4 = DEQUANTIZE(inptr[DCTSIZE*7], quantptr[DCTSIZE*7]);
tmp11 = z1 + z3;
tmp1 = MULTIPLY(z1 + z2, FIX(1.353318001)); /* c3 */
tmp2 = MULTIPLY(tmp11, FIX(1.247225013)); /* c5 */
tmp3 = MULTIPLY(z1 + z4, FIX(1.093201867)); /* c7 */
tmp10 = MULTIPLY(z1 - z4, FIX(0.897167586)); /* c9 */
tmp11 = MULTIPLY(tmp11, FIX(0.666655658)); /* c11 */
tmp12 = MULTIPLY(z1 - z2, FIX(0.410524528)); /* c13 */
tmp0 = tmp1 + tmp2 + tmp3 -
MULTIPLY(z1, FIX(2.286341144)); /* c7+c5+c3-c1 */
tmp13 = tmp10 + tmp11 + tmp12 -
MULTIPLY(z1, FIX(1.835730603)); /* c9+c11+c13-c15 */
z1 = MULTIPLY(z2 + z3, FIX(0.138617169)); /* c15 */
tmp1 += z1 + MULTIPLY(z2, FIX(0.071888074)); /* c9+c11-c3-c15 */
tmp2 += z1 - MULTIPLY(z3, FIX(1.125726048)); /* c5+c7+c15-c3 */
z1 = MULTIPLY(z3 - z2, FIX(1.407403738)); /* c1 */
tmp11 += z1 - MULTIPLY(z3, FIX(0.766367282)); /* c1+c11-c9-c13 */
tmp12 += z1 + MULTIPLY(z2, FIX(1.971951411)); /* c1+c5+c13-c7 */
z2 += z4;
z1 = MULTIPLY(z2, - FIX(0.666655658)); /* -c11 */
tmp1 += z1;
tmp3 += z1 + MULTIPLY(z4, FIX(1.065388962)); /* c3+c11+c15-c7 */
z2 = MULTIPLY(z2, - FIX(1.247225013)); /* -c5 */
tmp10 += z2 + MULTIPLY(z4, FIX(3.141271809)); /* c1+c5+c9-c13 */
tmp12 += z2;
z2 = MULTIPLY(z3 + z4, - FIX(1.353318001)); /* -c3 */
tmp2 += z2;
tmp3 += z2;
z2 = MULTIPLY(z4 - z3, FIX(0.410524528)); /* c13 */
tmp10 += z2;
tmp11 += z2;
/* Final output stage */
wsptr[8*0] = (int) RIGHT_SHIFT(tmp20 + tmp0, CONST_BITS-PASS1_BITS);
wsptr[8*15] = (int) RIGHT_SHIFT(tmp20 - tmp0, CONST_BITS-PASS1_BITS);
wsptr[8*1] = (int) RIGHT_SHIFT(tmp21 + tmp1, CONST_BITS-PASS1_BITS);
wsptr[8*14] = (int) RIGHT_SHIFT(tmp21 - tmp1, CONST_BITS-PASS1_BITS);
wsptr[8*2] = (int) RIGHT_SHIFT(tmp22 + tmp2, CONST_BITS-PASS1_BITS);
wsptr[8*13] = (int) RIGHT_SHIFT(tmp22 - tmp2, CONST_BITS-PASS1_BITS);
wsptr[8*3] = (int) RIGHT_SHIFT(tmp23 + tmp3, CONST_BITS-PASS1_BITS);
wsptr[8*12] = (int) RIGHT_SHIFT(tmp23 - tmp3, CONST_BITS-PASS1_BITS);
wsptr[8*4] = (int) RIGHT_SHIFT(tmp24 + tmp10, CONST_BITS-PASS1_BITS);
wsptr[8*11] = (int) RIGHT_SHIFT(tmp24 - tmp10, CONST_BITS-PASS1_BITS);
wsptr[8*5] = (int) RIGHT_SHIFT(tmp25 + tmp11, CONST_BITS-PASS1_BITS);
wsptr[8*10] = (int) RIGHT_SHIFT(tmp25 - tmp11, CONST_BITS-PASS1_BITS);
wsptr[8*6] = (int) RIGHT_SHIFT(tmp26 + tmp12, CONST_BITS-PASS1_BITS);
wsptr[8*9] = (int) RIGHT_SHIFT(tmp26 - tmp12, CONST_BITS-PASS1_BITS);
wsptr[8*7] = (int) RIGHT_SHIFT(tmp27 + tmp13, CONST_BITS-PASS1_BITS);
wsptr[8*8] = (int) RIGHT_SHIFT(tmp27 - tmp13, CONST_BITS-PASS1_BITS);
}
/* Pass 2: process 16 rows from work array, store into output array. */
wsptr = workspace;
for (ctr = 0; ctr < 16; ctr++) {
outptr = output_buf[ctr] + output_col;
/* Even part */
/* Add fudge factor here for final descale. */
tmp0 = (INT32) wsptr[0] + (ONE << (PASS1_BITS+2));
tmp0 <<= CONST_BITS;
z1 = (INT32) wsptr[4];
tmp1 = MULTIPLY(z1, FIX(1.306562965)); /* c4[16] = c2[8] */
tmp2 = MULTIPLY(z1, FIX_0_541196100); /* c12[16] = c6[8] */
tmp10 = tmp0 + tmp1;
tmp11 = tmp0 - tmp1;
tmp12 = tmp0 + tmp2;
tmp13 = tmp0 - tmp2;
z1 = (INT32) wsptr[2];
z2 = (INT32) wsptr[6];
z3 = z1 - z2;
z4 = MULTIPLY(z3, FIX(0.275899379)); /* c14[16] = c7[8] */
z3 = MULTIPLY(z3, FIX(1.387039845)); /* c2[16] = c1[8] */
tmp0 = z3 + MULTIPLY(z2, FIX_2_562915447); /* (c6+c2)[16] = (c3+c1)[8] */
tmp1 = z4 + MULTIPLY(z1, FIX_0_899976223); /* (c6-c14)[16] = (c3-c7)[8] */
tmp2 = z3 - MULTIPLY(z1, FIX(0.601344887)); /* (c2-c10)[16] = (c1-c5)[8] */
tmp3 = z4 - MULTIPLY(z2, FIX(0.509795579)); /* (c10-c14)[16] = (c5-c7)[8] */
tmp20 = tmp10 + tmp0;
tmp27 = tmp10 - tmp0;
tmp21 = tmp12 + tmp1;
tmp26 = tmp12 - tmp1;
tmp22 = tmp13 + tmp2;
tmp25 = tmp13 - tmp2;
tmp23 = tmp11 + tmp3;
tmp24 = tmp11 - tmp3;
/* Odd part */
z1 = (INT32) wsptr[1];
z2 = (INT32) wsptr[3];
z3 = (INT32) wsptr[5];
z4 = (INT32) wsptr[7];
tmp11 = z1 + z3;
tmp1 = MULTIPLY(z1 + z2, FIX(1.353318001)); /* c3 */
tmp2 = MULTIPLY(tmp11, FIX(1.247225013)); /* c5 */
tmp3 = MULTIPLY(z1 + z4, FIX(1.093201867)); /* c7 */
tmp10 = MULTIPLY(z1 - z4, FIX(0.897167586)); /* c9 */
tmp11 = MULTIPLY(tmp11, FIX(0.666655658)); /* c11 */
tmp12 = MULTIPLY(z1 - z2, FIX(0.410524528)); /* c13 */
tmp0 = tmp1 + tmp2 + tmp3 -
MULTIPLY(z1, FIX(2.286341144)); /* c7+c5+c3-c1 */
tmp13 = tmp10 + tmp11 + tmp12 -
MULTIPLY(z1, FIX(1.835730603)); /* c9+c11+c13-c15 */
z1 = MULTIPLY(z2 + z3, FIX(0.138617169)); /* c15 */
tmp1 += z1 + MULTIPLY(z2, FIX(0.071888074)); /* c9+c11-c3-c15 */
tmp2 += z1 - MULTIPLY(z3, FIX(1.125726048)); /* c5+c7+c15-c3 */
z1 = MULTIPLY(z3 - z2, FIX(1.407403738)); /* c1 */
tmp11 += z1 - MULTIPLY(z3, FIX(0.766367282)); /* c1+c11-c9-c13 */
tmp12 += z1 + MULTIPLY(z2, FIX(1.971951411)); /* c1+c5+c13-c7 */
z2 += z4;
z1 = MULTIPLY(z2, - FIX(0.666655658)); /* -c11 */
tmp1 += z1;
tmp3 += z1 + MULTIPLY(z4, FIX(1.065388962)); /* c3+c11+c15-c7 */
z2 = MULTIPLY(z2, - FIX(1.247225013)); /* -c5 */
tmp10 += z2 + MULTIPLY(z4, FIX(3.141271809)); /* c1+c5+c9-c13 */
tmp12 += z2;
z2 = MULTIPLY(z3 + z4, - FIX(1.353318001)); /* -c3 */
tmp2 += z2;
tmp3 += z2;
z2 = MULTIPLY(z4 - z3, FIX(0.410524528)); /* c13 */
tmp10 += z2;
tmp11 += z2;
/* Final output stage */
outptr[0] = range_limit[(int) RIGHT_SHIFT(tmp20 + tmp0,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[15] = range_limit[(int) RIGHT_SHIFT(tmp20 - tmp0,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[1] = range_limit[(int) RIGHT_SHIFT(tmp21 + tmp1,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[14] = range_limit[(int) RIGHT_SHIFT(tmp21 - tmp1,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[2] = range_limit[(int) RIGHT_SHIFT(tmp22 + tmp2,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[13] = range_limit[(int) RIGHT_SHIFT(tmp22 - tmp2,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[3] = range_limit[(int) RIGHT_SHIFT(tmp23 + tmp3,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[12] = range_limit[(int) RIGHT_SHIFT(tmp23 - tmp3,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[4] = range_limit[(int) RIGHT_SHIFT(tmp24 + tmp10,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[11] = range_limit[(int) RIGHT_SHIFT(tmp24 - tmp10,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[5] = range_limit[(int) RIGHT_SHIFT(tmp25 + tmp11,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[10] = range_limit[(int) RIGHT_SHIFT(tmp25 - tmp11,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[6] = range_limit[(int) RIGHT_SHIFT(tmp26 + tmp12,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[9] = range_limit[(int) RIGHT_SHIFT(tmp26 - tmp12,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[7] = range_limit[(int) RIGHT_SHIFT(tmp27 + tmp13,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[8] = range_limit[(int) RIGHT_SHIFT(tmp27 - tmp13,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
wsptr += 8; /* advance pointer to next row */
}
}
/*
src/Source/LibJPEG/jidctint.c view on Meta::CPAN
wsptr++;
}
/* Pass 2: process 8 rows from work array, store into output array.
* 16-point IDCT kernel, cK represents sqrt(2) * cos(K*pi/32).
*/
wsptr = workspace;
for (ctr = 0; ctr < 8; ctr++) {
outptr = output_buf[ctr] + output_col;
/* Even part */
/* Add fudge factor here for final descale. */
tmp0 = (INT32) wsptr[0] + (ONE << (PASS1_BITS+2));
tmp0 <<= CONST_BITS;
z1 = (INT32) wsptr[4];
tmp1 = MULTIPLY(z1, FIX(1.306562965)); /* c4[16] = c2[8] */
tmp2 = MULTIPLY(z1, FIX_0_541196100); /* c12[16] = c6[8] */
tmp10 = tmp0 + tmp1;
tmp11 = tmp0 - tmp1;
tmp12 = tmp0 + tmp2;
tmp13 = tmp0 - tmp2;
z1 = (INT32) wsptr[2];
z2 = (INT32) wsptr[6];
z3 = z1 - z2;
z4 = MULTIPLY(z3, FIX(0.275899379)); /* c14[16] = c7[8] */
z3 = MULTIPLY(z3, FIX(1.387039845)); /* c2[16] = c1[8] */
tmp0 = z3 + MULTIPLY(z2, FIX_2_562915447); /* (c6+c2)[16] = (c3+c1)[8] */
tmp1 = z4 + MULTIPLY(z1, FIX_0_899976223); /* (c6-c14)[16] = (c3-c7)[8] */
tmp2 = z3 - MULTIPLY(z1, FIX(0.601344887)); /* (c2-c10)[16] = (c1-c5)[8] */
tmp3 = z4 - MULTIPLY(z2, FIX(0.509795579)); /* (c10-c14)[16] = (c5-c7)[8] */
tmp20 = tmp10 + tmp0;
tmp27 = tmp10 - tmp0;
tmp21 = tmp12 + tmp1;
tmp26 = tmp12 - tmp1;
tmp22 = tmp13 + tmp2;
tmp25 = tmp13 - tmp2;
tmp23 = tmp11 + tmp3;
tmp24 = tmp11 - tmp3;
/* Odd part */
z1 = (INT32) wsptr[1];
z2 = (INT32) wsptr[3];
z3 = (INT32) wsptr[5];
z4 = (INT32) wsptr[7];
tmp11 = z1 + z3;
tmp1 = MULTIPLY(z1 + z2, FIX(1.353318001)); /* c3 */
tmp2 = MULTIPLY(tmp11, FIX(1.247225013)); /* c5 */
tmp3 = MULTIPLY(z1 + z4, FIX(1.093201867)); /* c7 */
tmp10 = MULTIPLY(z1 - z4, FIX(0.897167586)); /* c9 */
tmp11 = MULTIPLY(tmp11, FIX(0.666655658)); /* c11 */
tmp12 = MULTIPLY(z1 - z2, FIX(0.410524528)); /* c13 */
tmp0 = tmp1 + tmp2 + tmp3 -
MULTIPLY(z1, FIX(2.286341144)); /* c7+c5+c3-c1 */
tmp13 = tmp10 + tmp11 + tmp12 -
MULTIPLY(z1, FIX(1.835730603)); /* c9+c11+c13-c15 */
z1 = MULTIPLY(z2 + z3, FIX(0.138617169)); /* c15 */
tmp1 += z1 + MULTIPLY(z2, FIX(0.071888074)); /* c9+c11-c3-c15 */
tmp2 += z1 - MULTIPLY(z3, FIX(1.125726048)); /* c5+c7+c15-c3 */
z1 = MULTIPLY(z3 - z2, FIX(1.407403738)); /* c1 */
tmp11 += z1 - MULTIPLY(z3, FIX(0.766367282)); /* c1+c11-c9-c13 */
tmp12 += z1 + MULTIPLY(z2, FIX(1.971951411)); /* c1+c5+c13-c7 */
z2 += z4;
z1 = MULTIPLY(z2, - FIX(0.666655658)); /* -c11 */
tmp1 += z1;
tmp3 += z1 + MULTIPLY(z4, FIX(1.065388962)); /* c3+c11+c15-c7 */
z2 = MULTIPLY(z2, - FIX(1.247225013)); /* -c5 */
tmp10 += z2 + MULTIPLY(z4, FIX(3.141271809)); /* c1+c5+c9-c13 */
tmp12 += z2;
z2 = MULTIPLY(z3 + z4, - FIX(1.353318001)); /* -c3 */
tmp2 += z2;
tmp3 += z2;
z2 = MULTIPLY(z4 - z3, FIX(0.410524528)); /* c13 */
tmp10 += z2;
tmp11 += z2;
/* Final output stage */
outptr[0] = range_limit[(int) RIGHT_SHIFT(tmp20 + tmp0,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[15] = range_limit[(int) RIGHT_SHIFT(tmp20 - tmp0,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[1] = range_limit[(int) RIGHT_SHIFT(tmp21 + tmp1,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[14] = range_limit[(int) RIGHT_SHIFT(tmp21 - tmp1,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[2] = range_limit[(int) RIGHT_SHIFT(tmp22 + tmp2,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[13] = range_limit[(int) RIGHT_SHIFT(tmp22 - tmp2,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[3] = range_limit[(int) RIGHT_SHIFT(tmp23 + tmp3,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[12] = range_limit[(int) RIGHT_SHIFT(tmp23 - tmp3,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[4] = range_limit[(int) RIGHT_SHIFT(tmp24 + tmp10,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[11] = range_limit[(int) RIGHT_SHIFT(tmp24 - tmp10,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[5] = range_limit[(int) RIGHT_SHIFT(tmp25 + tmp11,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[10] = range_limit[(int) RIGHT_SHIFT(tmp25 - tmp11,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[6] = range_limit[(int) RIGHT_SHIFT(tmp26 + tmp12,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[9] = range_limit[(int) RIGHT_SHIFT(tmp26 - tmp12,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[7] = range_limit[(int) RIGHT_SHIFT(tmp27 + tmp13,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[8] = range_limit[(int) RIGHT_SHIFT(tmp27 - tmp13,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
wsptr += 8; /* advance pointer to next row */
}
}
/*
* Perform dequantization and inverse DCT on one block of coefficients,
* producing a 14x7 output block.
*
* 7-point IDCT in pass 1 (columns), 14-point in pass 2 (rows).
*/
GLOBAL(void)
jpeg_idct_14x7 (j_decompress_ptr cinfo, jpeg_component_info * compptr,
JCOEFPTR coef_block,
JSAMPARRAY output_buf, JDIMENSION output_col)
{
INT32 tmp10, tmp11, tmp12, tmp13, tmp14, tmp15, tmp16;
INT32 tmp20, tmp21, tmp22, tmp23, tmp24, tmp25, tmp26;
INT32 z1, z2, z3, z4;
JCOEFPTR inptr;
ISLOW_MULT_TYPE * quantptr;
int * wsptr;
JSAMPROW outptr;
JSAMPLE *range_limit = IDCT_range_limit(cinfo);
int ctr;
int workspace[8*7]; /* buffers data between passes */
SHIFT_TEMPS
/* Pass 1: process columns from input, store into work array.
* 7-point IDCT kernel, cK represents sqrt(2) * cos(K*pi/14).
*/
inptr = coef_block;
quantptr = (ISLOW_MULT_TYPE *) compptr->dct_table;
wsptr = workspace;
for (ctr = 0; ctr < 8; ctr++, inptr++, quantptr++, wsptr++) {
/* Even part */
tmp23 = DEQUANTIZE(inptr[DCTSIZE*0], quantptr[DCTSIZE*0]);
tmp23 <<= CONST_BITS;
/* Add fudge factor here for final descale. */
tmp23 += ONE << (CONST_BITS-PASS1_BITS-1);
z1 = DEQUANTIZE(inptr[DCTSIZE*2], quantptr[DCTSIZE*2]);
z2 = DEQUANTIZE(inptr[DCTSIZE*4], quantptr[DCTSIZE*4]);
z3 = DEQUANTIZE(inptr[DCTSIZE*6], quantptr[DCTSIZE*6]);
tmp20 = MULTIPLY(z2 - z3, FIX(0.881747734)); /* c4 */
tmp22 = MULTIPLY(z1 - z2, FIX(0.314692123)); /* c6 */
tmp21 = tmp20 + tmp22 + tmp23 - MULTIPLY(z2, FIX(1.841218003)); /* c2+c4-c6 */
tmp10 = z1 + z3;
z2 -= tmp10;
tmp10 = MULTIPLY(tmp10, FIX(1.274162392)) + tmp23; /* c2 */
tmp20 += tmp10 - MULTIPLY(z3, FIX(0.077722536)); /* c2-c4-c6 */
tmp22 += tmp10 - MULTIPLY(z1, FIX(2.470602249)); /* c2+c4+c6 */
tmp23 += MULTIPLY(z2, FIX(1.414213562)); /* c0 */
/* Odd part */
z1 = DEQUANTIZE(inptr[DCTSIZE*1], quantptr[DCTSIZE*1]);
z2 = DEQUANTIZE(inptr[DCTSIZE*3], quantptr[DCTSIZE*3]);
z3 = DEQUANTIZE(inptr[DCTSIZE*5], quantptr[DCTSIZE*5]);
tmp11 = MULTIPLY(z1 + z2, FIX(0.935414347)); /* (c3+c1-c5)/2 */
tmp12 = MULTIPLY(z1 - z2, FIX(0.170262339)); /* (c3+c5-c1)/2 */
tmp10 = tmp11 - tmp12;
tmp11 += tmp12;
tmp12 = MULTIPLY(z2 + z3, - FIX(1.378756276)); /* -c1 */
tmp11 += tmp12;
z2 = MULTIPLY(z1 + z3, FIX(0.613604268)); /* c5 */
tmp10 += z2;
tmp12 += z2 + MULTIPLY(z3, FIX(1.870828693)); /* c3+c1-c5 */
/* Final output stage */
wsptr[8*0] = (int) RIGHT_SHIFT(tmp20 + tmp10, CONST_BITS-PASS1_BITS);
wsptr[8*6] = (int) RIGHT_SHIFT(tmp20 - tmp10, CONST_BITS-PASS1_BITS);
wsptr[8*1] = (int) RIGHT_SHIFT(tmp21 + tmp11, CONST_BITS-PASS1_BITS);
wsptr[8*5] = (int) RIGHT_SHIFT(tmp21 - tmp11, CONST_BITS-PASS1_BITS);
wsptr[8*2] = (int) RIGHT_SHIFT(tmp22 + tmp12, CONST_BITS-PASS1_BITS);
wsptr[8*4] = (int) RIGHT_SHIFT(tmp22 - tmp12, CONST_BITS-PASS1_BITS);
wsptr[8*3] = (int) RIGHT_SHIFT(tmp23, CONST_BITS-PASS1_BITS);
}
/* Pass 2: process 7 rows from work array, store into output array.
* 14-point IDCT kernel, cK represents sqrt(2) * cos(K*pi/28).
*/
wsptr = workspace;
for (ctr = 0; ctr < 7; ctr++) {
outptr = output_buf[ctr] + output_col;
/* Even part */
/* Add fudge factor here for final descale. */
z1 = (INT32) wsptr[0] + (ONE << (PASS1_BITS+2));
z1 <<= CONST_BITS;
z4 = (INT32) wsptr[4];
z2 = MULTIPLY(z4, FIX(1.274162392)); /* c4 */
z3 = MULTIPLY(z4, FIX(0.314692123)); /* c12 */
z4 = MULTIPLY(z4, FIX(0.881747734)); /* c8 */
tmp10 = z1 + z2;
tmp11 = z1 + z3;
tmp12 = z1 - z4;
tmp23 = z1 - ((z2 + z3 - z4) << 1); /* c0 = (c4+c12-c8)*2 */
z1 = (INT32) wsptr[2];
z2 = (INT32) wsptr[6];
z3 = MULTIPLY(z1 + z2, FIX(1.105676686)); /* c6 */
tmp13 = z3 + MULTIPLY(z1, FIX(0.273079590)); /* c2-c6 */
tmp14 = z3 - MULTIPLY(z2, FIX(1.719280954)); /* c6+c10 */
src/Source/LibJPEG/jidctint.c view on Meta::CPAN
tmp14 = z1 - MULTIPLY(z3, FIX(2.176250899)); /* c1+c3 */
/* Final output stage */
wsptr[8*0] = (int) RIGHT_SHIFT(tmp10 + tmp13, CONST_BITS-PASS1_BITS);
wsptr[8*4] = (int) RIGHT_SHIFT(tmp10 - tmp13, CONST_BITS-PASS1_BITS);
wsptr[8*1] = (int) RIGHT_SHIFT(tmp11 + tmp14, CONST_BITS-PASS1_BITS);
wsptr[8*3] = (int) RIGHT_SHIFT(tmp11 - tmp14, CONST_BITS-PASS1_BITS);
wsptr[8*2] = (int) RIGHT_SHIFT(tmp12, CONST_BITS-PASS1_BITS);
}
/* Pass 2: process 5 rows from work array, store into output array.
* 10-point IDCT kernel, cK represents sqrt(2) * cos(K*pi/20).
*/
wsptr = workspace;
for (ctr = 0; ctr < 5; ctr++) {
outptr = output_buf[ctr] + output_col;
/* Even part */
/* Add fudge factor here for final descale. */
z3 = (INT32) wsptr[0] + (ONE << (PASS1_BITS+2));
z3 <<= CONST_BITS;
z4 = (INT32) wsptr[4];
z1 = MULTIPLY(z4, FIX(1.144122806)); /* c4 */
z2 = MULTIPLY(z4, FIX(0.437016024)); /* c8 */
tmp10 = z3 + z1;
tmp11 = z3 - z2;
tmp22 = z3 - ((z1 - z2) << 1); /* c0 = (c4-c8)*2 */
z2 = (INT32) wsptr[2];
z3 = (INT32) wsptr[6];
z1 = MULTIPLY(z2 + z3, FIX(0.831253876)); /* c6 */
tmp12 = z1 + MULTIPLY(z2, FIX(0.513743148)); /* c2-c6 */
tmp13 = z1 - MULTIPLY(z3, FIX(2.176250899)); /* c2+c6 */
tmp20 = tmp10 + tmp12;
tmp24 = tmp10 - tmp12;
tmp21 = tmp11 + tmp13;
tmp23 = tmp11 - tmp13;
/* Odd part */
z1 = (INT32) wsptr[1];
z2 = (INT32) wsptr[3];
z3 = (INT32) wsptr[5];
z3 <<= CONST_BITS;
z4 = (INT32) wsptr[7];
tmp11 = z2 + z4;
tmp13 = z2 - z4;
tmp12 = MULTIPLY(tmp13, FIX(0.309016994)); /* (c3-c7)/2 */
z2 = MULTIPLY(tmp11, FIX(0.951056516)); /* (c3+c7)/2 */
z4 = z3 + tmp12;
tmp10 = MULTIPLY(z1, FIX(1.396802247)) + z2 + z4; /* c1 */
tmp14 = MULTIPLY(z1, FIX(0.221231742)) - z2 + z4; /* c9 */
z2 = MULTIPLY(tmp11, FIX(0.587785252)); /* (c1-c9)/2 */
z4 = z3 - tmp12 - (tmp13 << (CONST_BITS - 1));
tmp12 = ((z1 - tmp13) << CONST_BITS) - z3;
tmp11 = MULTIPLY(z1, FIX(1.260073511)) - z2 - z4; /* c3 */
tmp13 = MULTIPLY(z1, FIX(0.642039522)) - z2 + z4; /* c7 */
/* Final output stage */
outptr[0] = range_limit[(int) RIGHT_SHIFT(tmp20 + tmp10,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[9] = range_limit[(int) RIGHT_SHIFT(tmp20 - tmp10,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[1] = range_limit[(int) RIGHT_SHIFT(tmp21 + tmp11,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[8] = range_limit[(int) RIGHT_SHIFT(tmp21 - tmp11,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[2] = range_limit[(int) RIGHT_SHIFT(tmp22 + tmp12,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[7] = range_limit[(int) RIGHT_SHIFT(tmp22 - tmp12,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[3] = range_limit[(int) RIGHT_SHIFT(tmp23 + tmp13,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[6] = range_limit[(int) RIGHT_SHIFT(tmp23 - tmp13,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[4] = range_limit[(int) RIGHT_SHIFT(tmp24 + tmp14,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[5] = range_limit[(int) RIGHT_SHIFT(tmp24 - tmp14,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
wsptr += 8; /* advance pointer to next row */
}
}
/*
* Perform dequantization and inverse DCT on one block of coefficients,
* producing a 8x4 output block.
*
* 4-point IDCT in pass 1 (columns), 8-point in pass 2 (rows).
*/
GLOBAL(void)
jpeg_idct_8x4 (j_decompress_ptr cinfo, jpeg_component_info * compptr,
JCOEFPTR coef_block,
JSAMPARRAY output_buf, JDIMENSION output_col)
{
src/Source/LibJPEG/jidctint.c view on Meta::CPAN
SHIFT_TEMPS
/* Pass 1: process columns from input, store into work array.
* 16-point IDCT kernel, cK represents sqrt(2) * cos(K*pi/32).
*/
inptr = coef_block;
quantptr = (ISLOW_MULT_TYPE *) compptr->dct_table;
wsptr = workspace;
for (ctr = 0; ctr < 8; ctr++, inptr++, quantptr++, wsptr++) {
/* Even part */
tmp0 = DEQUANTIZE(inptr[DCTSIZE*0], quantptr[DCTSIZE*0]);
tmp0 <<= CONST_BITS;
/* Add fudge factor here for final descale. */
tmp0 += ONE << (CONST_BITS-PASS1_BITS-1);
z1 = DEQUANTIZE(inptr[DCTSIZE*4], quantptr[DCTSIZE*4]);
tmp1 = MULTIPLY(z1, FIX(1.306562965)); /* c4[16] = c2[8] */
tmp2 = MULTIPLY(z1, FIX_0_541196100); /* c12[16] = c6[8] */
tmp10 = tmp0 + tmp1;
tmp11 = tmp0 - tmp1;
tmp12 = tmp0 + tmp2;
tmp13 = tmp0 - tmp2;
z1 = DEQUANTIZE(inptr[DCTSIZE*2], quantptr[DCTSIZE*2]);
z2 = DEQUANTIZE(inptr[DCTSIZE*6], quantptr[DCTSIZE*6]);
z3 = z1 - z2;
z4 = MULTIPLY(z3, FIX(0.275899379)); /* c14[16] = c7[8] */
z3 = MULTIPLY(z3, FIX(1.387039845)); /* c2[16] = c1[8] */
tmp0 = z3 + MULTIPLY(z2, FIX_2_562915447); /* (c6+c2)[16] = (c3+c1)[8] */
tmp1 = z4 + MULTIPLY(z1, FIX_0_899976223); /* (c6-c14)[16] = (c3-c7)[8] */
tmp2 = z3 - MULTIPLY(z1, FIX(0.601344887)); /* (c2-c10)[16] = (c1-c5)[8] */
tmp3 = z4 - MULTIPLY(z2, FIX(0.509795579)); /* (c10-c14)[16] = (c5-c7)[8] */
tmp20 = tmp10 + tmp0;
tmp27 = tmp10 - tmp0;
tmp21 = tmp12 + tmp1;
tmp26 = tmp12 - tmp1;
tmp22 = tmp13 + tmp2;
tmp25 = tmp13 - tmp2;
tmp23 = tmp11 + tmp3;
tmp24 = tmp11 - tmp3;
/* Odd part */
z1 = DEQUANTIZE(inptr[DCTSIZE*1], quantptr[DCTSIZE*1]);
z2 = DEQUANTIZE(inptr[DCTSIZE*3], quantptr[DCTSIZE*3]);
z3 = DEQUANTIZE(inptr[DCTSIZE*5], quantptr[DCTSIZE*5]);
z4 = DEQUANTIZE(inptr[DCTSIZE*7], quantptr[DCTSIZE*7]);
tmp11 = z1 + z3;
tmp1 = MULTIPLY(z1 + z2, FIX(1.353318001)); /* c3 */
tmp2 = MULTIPLY(tmp11, FIX(1.247225013)); /* c5 */
tmp3 = MULTIPLY(z1 + z4, FIX(1.093201867)); /* c7 */
tmp10 = MULTIPLY(z1 - z4, FIX(0.897167586)); /* c9 */
tmp11 = MULTIPLY(tmp11, FIX(0.666655658)); /* c11 */
tmp12 = MULTIPLY(z1 - z2, FIX(0.410524528)); /* c13 */
tmp0 = tmp1 + tmp2 + tmp3 -
MULTIPLY(z1, FIX(2.286341144)); /* c7+c5+c3-c1 */
tmp13 = tmp10 + tmp11 + tmp12 -
MULTIPLY(z1, FIX(1.835730603)); /* c9+c11+c13-c15 */
z1 = MULTIPLY(z2 + z3, FIX(0.138617169)); /* c15 */
tmp1 += z1 + MULTIPLY(z2, FIX(0.071888074)); /* c9+c11-c3-c15 */
tmp2 += z1 - MULTIPLY(z3, FIX(1.125726048)); /* c5+c7+c15-c3 */
z1 = MULTIPLY(z3 - z2, FIX(1.407403738)); /* c1 */
tmp11 += z1 - MULTIPLY(z3, FIX(0.766367282)); /* c1+c11-c9-c13 */
tmp12 += z1 + MULTIPLY(z2, FIX(1.971951411)); /* c1+c5+c13-c7 */
z2 += z4;
z1 = MULTIPLY(z2, - FIX(0.666655658)); /* -c11 */
tmp1 += z1;
tmp3 += z1 + MULTIPLY(z4, FIX(1.065388962)); /* c3+c11+c15-c7 */
z2 = MULTIPLY(z2, - FIX(1.247225013)); /* -c5 */
tmp10 += z2 + MULTIPLY(z4, FIX(3.141271809)); /* c1+c5+c9-c13 */
tmp12 += z2;
z2 = MULTIPLY(z3 + z4, - FIX(1.353318001)); /* -c3 */
tmp2 += z2;
tmp3 += z2;
z2 = MULTIPLY(z4 - z3, FIX(0.410524528)); /* c13 */
tmp10 += z2;
tmp11 += z2;
/* Final output stage */
wsptr[8*0] = (int) RIGHT_SHIFT(tmp20 + tmp0, CONST_BITS-PASS1_BITS);
wsptr[8*15] = (int) RIGHT_SHIFT(tmp20 - tmp0, CONST_BITS-PASS1_BITS);
wsptr[8*1] = (int) RIGHT_SHIFT(tmp21 + tmp1, CONST_BITS-PASS1_BITS);
wsptr[8*14] = (int) RIGHT_SHIFT(tmp21 - tmp1, CONST_BITS-PASS1_BITS);
wsptr[8*2] = (int) RIGHT_SHIFT(tmp22 + tmp2, CONST_BITS-PASS1_BITS);
wsptr[8*13] = (int) RIGHT_SHIFT(tmp22 - tmp2, CONST_BITS-PASS1_BITS);
wsptr[8*3] = (int) RIGHT_SHIFT(tmp23 + tmp3, CONST_BITS-PASS1_BITS);
wsptr[8*12] = (int) RIGHT_SHIFT(tmp23 - tmp3, CONST_BITS-PASS1_BITS);
wsptr[8*4] = (int) RIGHT_SHIFT(tmp24 + tmp10, CONST_BITS-PASS1_BITS);
wsptr[8*11] = (int) RIGHT_SHIFT(tmp24 - tmp10, CONST_BITS-PASS1_BITS);
wsptr[8*5] = (int) RIGHT_SHIFT(tmp25 + tmp11, CONST_BITS-PASS1_BITS);
wsptr[8*10] = (int) RIGHT_SHIFT(tmp25 - tmp11, CONST_BITS-PASS1_BITS);
wsptr[8*6] = (int) RIGHT_SHIFT(tmp26 + tmp12, CONST_BITS-PASS1_BITS);
wsptr[8*9] = (int) RIGHT_SHIFT(tmp26 - tmp12, CONST_BITS-PASS1_BITS);
wsptr[8*7] = (int) RIGHT_SHIFT(tmp27 + tmp13, CONST_BITS-PASS1_BITS);
wsptr[8*8] = (int) RIGHT_SHIFT(tmp27 - tmp13, CONST_BITS-PASS1_BITS);
}
/* Pass 2: process rows from work array, store into output array.
* Note that we must descale the results by a factor of 8 == 2**3,
* and also undo the PASS1_BITS scaling.
* 8-point IDCT kernel, cK represents sqrt(2) * cos(K*pi/16).
*/
wsptr = workspace;
for (ctr = 0; ctr < 16; ctr++) {
outptr = output_buf[ctr] + output_col;
/* Even part: reverse the even part of the forward DCT.
* The rotator is c(-6).
*/
z2 = (INT32) wsptr[2];
z3 = (INT32) wsptr[6];
z1 = MULTIPLY(z2 + z3, FIX_0_541196100); /* c6 */
tmp2 = z1 + MULTIPLY(z2, FIX_0_765366865); /* c2-c6 */
tmp3 = z1 - MULTIPLY(z3, FIX_1_847759065); /* c2+c6 */
/* Add fudge factor here for final descale. */
z2 = (INT32) wsptr[0] + (ONE << (PASS1_BITS+2));
z3 = (INT32) wsptr[4];
tmp0 = (z2 + z3) << CONST_BITS;
tmp1 = (z2 - z3) << CONST_BITS;
tmp10 = tmp0 + tmp2;
tmp13 = tmp0 - tmp2;
tmp11 = tmp1 + tmp3;
tmp12 = tmp1 - tmp3;
/* Odd part per figure 8; the matrix is unitary and hence its
* transpose is its inverse. i0..i3 are y7,y5,y3,y1 respectively.
*/
src/Source/LibJPEG/jidctint.c view on Meta::CPAN
tmp12 = MULTIPLY(tmp14, FIX(1.197448846)); /* c5 */
tmp10 = tmp11 + tmp12 + tmp13 - MULTIPLY(z1, FIX(1.126980169)); /* c3+c5-c1 */
tmp14 = MULTIPLY(tmp14, FIX(0.752406978)); /* c9 */
tmp16 = tmp14 - MULTIPLY(z1, FIX(1.061150426)); /* c9+c11-c13 */
z1 -= z2;
tmp15 = MULTIPLY(z1, FIX(0.467085129)) - tmp13; /* c11 */
tmp16 += tmp15;
z1 += z4;
z4 = MULTIPLY(z2 + z3, - FIX(0.158341681)) - tmp13; /* -c13 */
tmp11 += z4 - MULTIPLY(z2, FIX(0.424103948)); /* c3-c9-c13 */
tmp12 += z4 - MULTIPLY(z3, FIX(2.373959773)); /* c3+c5-c13 */
z4 = MULTIPLY(z3 - z2, FIX(1.405321284)); /* c1 */
tmp14 += z4 + tmp13 - MULTIPLY(z3, FIX(1.6906431334)); /* c1+c9-c11 */
tmp15 += z4 + MULTIPLY(z2, FIX(0.674957567)); /* c1+c11-c5 */
tmp13 = (z1 - z3) << PASS1_BITS;
/* Final output stage */
wsptr[7*0] = (int) RIGHT_SHIFT(tmp20 + tmp10, CONST_BITS-PASS1_BITS);
wsptr[7*13] = (int) RIGHT_SHIFT(tmp20 - tmp10, CONST_BITS-PASS1_BITS);
wsptr[7*1] = (int) RIGHT_SHIFT(tmp21 + tmp11, CONST_BITS-PASS1_BITS);
wsptr[7*12] = (int) RIGHT_SHIFT(tmp21 - tmp11, CONST_BITS-PASS1_BITS);
wsptr[7*2] = (int) RIGHT_SHIFT(tmp22 + tmp12, CONST_BITS-PASS1_BITS);
wsptr[7*11] = (int) RIGHT_SHIFT(tmp22 - tmp12, CONST_BITS-PASS1_BITS);
wsptr[7*3] = (int) (tmp23 + tmp13);
wsptr[7*10] = (int) (tmp23 - tmp13);
wsptr[7*4] = (int) RIGHT_SHIFT(tmp24 + tmp14, CONST_BITS-PASS1_BITS);
wsptr[7*9] = (int) RIGHT_SHIFT(tmp24 - tmp14, CONST_BITS-PASS1_BITS);
wsptr[7*5] = (int) RIGHT_SHIFT(tmp25 + tmp15, CONST_BITS-PASS1_BITS);
wsptr[7*8] = (int) RIGHT_SHIFT(tmp25 - tmp15, CONST_BITS-PASS1_BITS);
wsptr[7*6] = (int) RIGHT_SHIFT(tmp26 + tmp16, CONST_BITS-PASS1_BITS);
wsptr[7*7] = (int) RIGHT_SHIFT(tmp26 - tmp16, CONST_BITS-PASS1_BITS);
}
/* Pass 2: process 14 rows from work array, store into output array.
* 7-point IDCT kernel, cK represents sqrt(2) * cos(K*pi/14).
*/
wsptr = workspace;
for (ctr = 0; ctr < 14; ctr++) {
outptr = output_buf[ctr] + output_col;
/* Even part */
/* Add fudge factor here for final descale. */
tmp23 = (INT32) wsptr[0] + (ONE << (PASS1_BITS+2));
tmp23 <<= CONST_BITS;
z1 = (INT32) wsptr[2];
z2 = (INT32) wsptr[4];
z3 = (INT32) wsptr[6];
tmp20 = MULTIPLY(z2 - z3, FIX(0.881747734)); /* c4 */
tmp22 = MULTIPLY(z1 - z2, FIX(0.314692123)); /* c6 */
tmp21 = tmp20 + tmp22 + tmp23 - MULTIPLY(z2, FIX(1.841218003)); /* c2+c4-c6 */
tmp10 = z1 + z3;
z2 -= tmp10;
tmp10 = MULTIPLY(tmp10, FIX(1.274162392)) + tmp23; /* c2 */
tmp20 += tmp10 - MULTIPLY(z3, FIX(0.077722536)); /* c2-c4-c6 */
tmp22 += tmp10 - MULTIPLY(z1, FIX(2.470602249)); /* c2+c4+c6 */
tmp23 += MULTIPLY(z2, FIX(1.414213562)); /* c0 */
/* Odd part */
z1 = (INT32) wsptr[1];
z2 = (INT32) wsptr[3];
z3 = (INT32) wsptr[5];
tmp11 = MULTIPLY(z1 + z2, FIX(0.935414347)); /* (c3+c1-c5)/2 */
tmp12 = MULTIPLY(z1 - z2, FIX(0.170262339)); /* (c3+c5-c1)/2 */
tmp10 = tmp11 - tmp12;
tmp11 += tmp12;
tmp12 = MULTIPLY(z2 + z3, - FIX(1.378756276)); /* -c1 */
tmp11 += tmp12;
z2 = MULTIPLY(z1 + z3, FIX(0.613604268)); /* c5 */
tmp10 += z2;
tmp12 += z2 + MULTIPLY(z3, FIX(1.870828693)); /* c3+c1-c5 */
/* Final output stage */
outptr[0] = range_limit[(int) RIGHT_SHIFT(tmp20 + tmp10,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[6] = range_limit[(int) RIGHT_SHIFT(tmp20 - tmp10,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[1] = range_limit[(int) RIGHT_SHIFT(tmp21 + tmp11,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[5] = range_limit[(int) RIGHT_SHIFT(tmp21 - tmp11,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[2] = range_limit[(int) RIGHT_SHIFT(tmp22 + tmp12,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[4] = range_limit[(int) RIGHT_SHIFT(tmp22 - tmp12,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[3] = range_limit[(int) RIGHT_SHIFT(tmp23,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
wsptr += 7; /* advance pointer to next row */
}
}
/*
* Perform dequantization and inverse DCT on one block of coefficients,
* producing a 6x12 output block.
*
* 12-point IDCT in pass 1 (columns), 6-point in pass 2 (rows).
*/
GLOBAL(void)
jpeg_idct_6x12 (j_decompress_ptr cinfo, jpeg_component_info * compptr,
JCOEFPTR coef_block,
JSAMPARRAY output_buf, JDIMENSION output_col)
{
INT32 tmp10, tmp11, tmp12, tmp13, tmp14, tmp15;
src/Source/LibJPEG/jidctint.c view on Meta::CPAN
JCOEFPTR inptr;
ISLOW_MULT_TYPE * quantptr;
int * wsptr;
JSAMPROW outptr;
JSAMPLE *range_limit = IDCT_range_limit(cinfo);
int ctr;
int workspace[5*10]; /* buffers data between passes */
SHIFT_TEMPS
/* Pass 1: process columns from input, store into work array.
* 10-point IDCT kernel, cK represents sqrt(2) * cos(K*pi/20).
*/
inptr = coef_block;
quantptr = (ISLOW_MULT_TYPE *) compptr->dct_table;
wsptr = workspace;
for (ctr = 0; ctr < 5; ctr++, inptr++, quantptr++, wsptr++) {
/* Even part */
z3 = DEQUANTIZE(inptr[DCTSIZE*0], quantptr[DCTSIZE*0]);
z3 <<= CONST_BITS;
/* Add fudge factor here for final descale. */
z3 += ONE << (CONST_BITS-PASS1_BITS-1);
z4 = DEQUANTIZE(inptr[DCTSIZE*4], quantptr[DCTSIZE*4]);
z1 = MULTIPLY(z4, FIX(1.144122806)); /* c4 */
z2 = MULTIPLY(z4, FIX(0.437016024)); /* c8 */
tmp10 = z3 + z1;
tmp11 = z3 - z2;
tmp22 = RIGHT_SHIFT(z3 - ((z1 - z2) << 1), /* c0 = (c4-c8)*2 */
CONST_BITS-PASS1_BITS);
z2 = DEQUANTIZE(inptr[DCTSIZE*2], quantptr[DCTSIZE*2]);
z3 = DEQUANTIZE(inptr[DCTSIZE*6], quantptr[DCTSIZE*6]);
z1 = MULTIPLY(z2 + z3, FIX(0.831253876)); /* c6 */
tmp12 = z1 + MULTIPLY(z2, FIX(0.513743148)); /* c2-c6 */
tmp13 = z1 - MULTIPLY(z3, FIX(2.176250899)); /* c2+c6 */
tmp20 = tmp10 + tmp12;
tmp24 = tmp10 - tmp12;
tmp21 = tmp11 + tmp13;
tmp23 = tmp11 - tmp13;
/* Odd part */
z1 = DEQUANTIZE(inptr[DCTSIZE*1], quantptr[DCTSIZE*1]);
z2 = DEQUANTIZE(inptr[DCTSIZE*3], quantptr[DCTSIZE*3]);
z3 = DEQUANTIZE(inptr[DCTSIZE*5], quantptr[DCTSIZE*5]);
z4 = DEQUANTIZE(inptr[DCTSIZE*7], quantptr[DCTSIZE*7]);
tmp11 = z2 + z4;
tmp13 = z2 - z4;
tmp12 = MULTIPLY(tmp13, FIX(0.309016994)); /* (c3-c7)/2 */
z5 = z3 << CONST_BITS;
z2 = MULTIPLY(tmp11, FIX(0.951056516)); /* (c3+c7)/2 */
z4 = z5 + tmp12;
tmp10 = MULTIPLY(z1, FIX(1.396802247)) + z2 + z4; /* c1 */
tmp14 = MULTIPLY(z1, FIX(0.221231742)) - z2 + z4; /* c9 */
z2 = MULTIPLY(tmp11, FIX(0.587785252)); /* (c1-c9)/2 */
z4 = z5 - tmp12 - (tmp13 << (CONST_BITS - 1));
tmp12 = (z1 - tmp13 - z3) << PASS1_BITS;
tmp11 = MULTIPLY(z1, FIX(1.260073511)) - z2 - z4; /* c3 */
tmp13 = MULTIPLY(z1, FIX(0.642039522)) - z2 + z4; /* c7 */
/* Final output stage */
wsptr[5*0] = (int) RIGHT_SHIFT(tmp20 + tmp10, CONST_BITS-PASS1_BITS);
wsptr[5*9] = (int) RIGHT_SHIFT(tmp20 - tmp10, CONST_BITS-PASS1_BITS);
wsptr[5*1] = (int) RIGHT_SHIFT(tmp21 + tmp11, CONST_BITS-PASS1_BITS);
wsptr[5*8] = (int) RIGHT_SHIFT(tmp21 - tmp11, CONST_BITS-PASS1_BITS);
wsptr[5*2] = (int) (tmp22 + tmp12);
wsptr[5*7] = (int) (tmp22 - tmp12);
wsptr[5*3] = (int) RIGHT_SHIFT(tmp23 + tmp13, CONST_BITS-PASS1_BITS);
wsptr[5*6] = (int) RIGHT_SHIFT(tmp23 - tmp13, CONST_BITS-PASS1_BITS);
wsptr[5*4] = (int) RIGHT_SHIFT(tmp24 + tmp14, CONST_BITS-PASS1_BITS);
wsptr[5*5] = (int) RIGHT_SHIFT(tmp24 - tmp14, CONST_BITS-PASS1_BITS);
}
/* Pass 2: process 10 rows from work array, store into output array.
* 5-point IDCT kernel, cK represents sqrt(2) * cos(K*pi/10).
*/
wsptr = workspace;
for (ctr = 0; ctr < 10; ctr++) {
outptr = output_buf[ctr] + output_col;
/* Even part */
/* Add fudge factor here for final descale. */
tmp12 = (INT32) wsptr[0] + (ONE << (PASS1_BITS+2));
tmp12 <<= CONST_BITS;
tmp13 = (INT32) wsptr[2];
tmp14 = (INT32) wsptr[4];
z1 = MULTIPLY(tmp13 + tmp14, FIX(0.790569415)); /* (c2+c4)/2 */
z2 = MULTIPLY(tmp13 - tmp14, FIX(0.353553391)); /* (c2-c4)/2 */
z3 = tmp12 + z2;
tmp10 = z3 + z1;
tmp11 = z3 - z1;
tmp12 -= z2 << 2;
/* Odd part */
z2 = (INT32) wsptr[1];
z3 = (INT32) wsptr[3];
z1 = MULTIPLY(z2 + z3, FIX(0.831253876)); /* c3 */
tmp13 = z1 + MULTIPLY(z2, FIX(0.513743148)); /* c1-c3 */
tmp14 = z1 - MULTIPLY(z3, FIX(2.176250899)); /* c1+c3 */
/* Final output stage */
outptr[0] = range_limit[(int) RIGHT_SHIFT(tmp10 + tmp13,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
( run in 0.472 second using v1.01-cache-2.11-cpan-56fb94df46f )