diff options
Diffstat (limited to 'src/lib/libjpeg/jcdctmgr.c')
-rw-r--r-- | src/lib/libjpeg/jcdctmgr.c | 56 |
1 files changed, 28 insertions, 28 deletions
diff --git a/src/lib/libjpeg/jcdctmgr.c b/src/lib/libjpeg/jcdctmgr.c index 0bbdbb685d1..83bea7f5f2d 100644 --- a/src/lib/libjpeg/jcdctmgr.c +++ b/src/lib/libjpeg/jcdctmgr.c @@ -90,17 +90,17 @@ forward_DCT (j_compress_ptr cinfo, jpeg_component_info * compptr, qval = divisors[i]; temp = workspace[i]; /* Divide the coefficient value by qval, ensuring proper rounding. - * Since C does not specify the direction of rounding for negative - * quotients, we have to force the dividend positive for portability. - * - * In most files, at least half of the output values will be zero - * (at default quantization settings, more like three-quarters...) - * so we should ensure that this case is fast. On many machines, - * a comparison is enough cheaper than a divide to make a special test - * a win. Since both inputs will be nonnegative, we need only test - * for a < b to discover whether a/b is 0. - * If your machine's division is fast enough, define FAST_DIVIDE. - */ + * Since C does not specify the direction of rounding for negative + * quotients, we have to force the dividend positive for portability. + * + * In most files, at least half of the output values will be zero + * (at default quantization settings, more like three-quarters...) + * so we should ensure that this case is fast. On many machines, + * a comparison is enough cheaper than a divide to make a special test + * a win. Since both inputs will be nonnegative, we need only test + * for a < b to discover whether a/b is 0. + * If your machine's division is fast enough, define FAST_DIVIDE. + */ #ifdef FAST_DIVIDE #define DIVIDE_BY(a,b) a /= b #else @@ -153,11 +153,11 @@ forward_DCT_float (j_compress_ptr cinfo, jpeg_component_info * compptr, /* Apply the quantization and scaling factor */ temp = workspace[i] * divisors[i]; /* Round to nearest integer. - * Since C does not specify the direction of rounding for negative - * quotients, we have to force the dividend positive for portability. - * The maximum coefficient size is +-16K (for 12-bit data), so this - * code should work for either 16-bit or 32-bit ints. - */ + * Since C does not specify the direction of rounding for negative + * quotients, we have to force the dividend positive for portability. + * The maximum coefficient size is +-16K (for 12-bit data), so this + * code should work for either 16-bit or 32-bit ints. + */ output_ptr[i] = (JCOEF) ((int) (temp + (FAST_FLOAT) 16384.5) - 16384); } } @@ -376,11 +376,11 @@ start_pass_fdctmgr (j_compress_ptr cinfo) case JDCT_IFAST: { /* For AA&N IDCT method, divisors are equal to quantization - * coefficients scaled by scalefactor[row]*scalefactor[col], where - * scalefactor[0] = 1 - * scalefactor[k] = cos(k*PI/16) * sqrt(2) for k=1..7 - * We apply a further scale factor of 8. - */ + * coefficients scaled by scalefactor[row]*scalefactor[col], where + * scalefactor[0] = 1 + * scalefactor[k] = cos(k*PI/16) * sqrt(2) for k=1..7 + * We apply a further scale factor of 8. + */ #define CONST_BITS 14 static const INT16 aanscales[DCTSIZE2] = { /* precomputed values scaled up by 14 bits */ @@ -415,13 +415,13 @@ start_pass_fdctmgr (j_compress_ptr cinfo) case JDCT_FLOAT: { /* For float AA&N IDCT method, divisors are equal to quantization - * coefficients scaled by scalefactor[row]*scalefactor[col], where - * scalefactor[0] = 1 - * scalefactor[k] = cos(k*PI/16) * sqrt(2) for k=1..7 - * We apply a further scale factor of 8. - * What's actually stored is 1/divisor so that the inner loop can - * use a multiplication rather than a division. - */ + * coefficients scaled by scalefactor[row]*scalefactor[col], where + * scalefactor[0] = 1 + * scalefactor[k] = cos(k*PI/16) * sqrt(2) for k=1..7 + * We apply a further scale factor of 8. + * What's actually stored is 1/divisor so that the inner loop can + * use a multiplication rather than a division. + */ FAST_FLOAT * fdtbl; int row, col; static const double aanscalefactor[DCTSIZE] = { |