quantize_fs_dither (j_decompress_ptr cinfo, JSAMPARRAY input_buf,
JSAMPARRAY output_buf, int num_rows)
/* General case, with Floyd-Steinberg dithering */
{
my_cquantize_ptr cquantize = (my_cquantize_ptr) cinfo->cquantize;
register LOCFSERROR cur; /* current error or pixel value */
LOCFSERROR belowerr; /* error for pixel below cur */
LOCFSERROR bpreverr; /* error for below/prev col */
LOCFSERROR bnexterr; /* error for below/next col */
LOCFSERROR delta;
register FSERRPTR errorptr; /* => fserrors[] at column before current */
register JSAMPROW input_ptr;
register JSAMPROW output_ptr;
JSAMPROW colorindex_ci;
JSAMPROW colormap_ci;
int pixcode;
int nc = cinfo->out_color_components;
int dir; /* 1 for left-to-right, -1 for right-to-left */
int dirnc; /* dir * nc */
int ci;
int row;
JDIMENSION col;
JDIMENSION width = cinfo->output_width;
JSAMPLE *range_limit = cinfo->sample_range_limit;
SHIFT_TEMPS
for (row = 0; row < num_rows; row++) {
/* Initialize output values to 0 so can process components separately */
jzero_far((void FAR *) output_buf[row],
(size_t) (width * SIZEOF(JSAMPLE)));
for (ci = 0; ci < nc; ci++) {
input_ptr = input_buf[row] + ci;
output_ptr = output_buf[row];
if (cquantize->on_odd_row) {
/* work right to left in this row */
input_ptr += (width-1) * nc; /* so point to rightmost pixel */
output_ptr += width-1;
dir = -1;
dirnc = -nc;
errorptr = cquantize->fserrors[ci] + (width+1); /* => entry after last column */
} else {
/* work left to right in this row */
dir = 1;
dirnc = nc;
errorptr = cquantize->fserrors[ci]; /* => entry before first column */
}
colorindex_ci = cquantize->colorindex[ci];
colormap_ci = cquantize->sv_colormap[ci];
/* Preset error values: no error propagated to first pixel from left */
cur = 0;
/* and no error propagated to row below yet */
belowerr = bpreverr = 0;
for (col = width; col > 0; col--) {
/* cur holds the error propagated from the previous pixel on the
* current line. Add the error propagated from the previous line
* to form the complete error correction term for this pixel, and
* round the error term (which is expressed * 16) to an integer.
* RIGHT_SHIFT rounds towards minus infinity, so adding 8 is correct
* for either sign of the error value.
* Note: errorptr points to *previous* column's array entry.
*/
cur = RIGHT_SHIFT(cur + errorptr[dir] + 8, 4);
/* Form pixel value + error, and range-limit to 0..MAXJSAMPLE.
* The maximum error is +- MAXJSAMPLE; this sets the required size
* of the range_limit array.
*/
cur += GETJSAMPLE(*input_ptr);
cur = GETJSAMPLE(range_limit[cur]);
/* Select output value, accumulate into output code for this pixel */
pixcode = GETJSAMPLE(colorindex_ci[cur]);
*output_ptr += (JSAMPLE) pixcode;
/* Compute actual representation error at this pixel */
/* Note: we can do this even though we don't have the final */
/* pixel code, because the colormap is orthogonal. */
cur -= GETJSAMPLE(colormap_ci[pixcode]);
/* Compute error fractions to be propagated to adjacent pixels.
* Add these into the running sums, and simultaneously shift the
* next-line error sums left by 1 column.
*/
bnexterr = cur;
delta = cur * 2;
cur += delta; /* form error * 3 */
errorptr[0] = (FSERROR) (bpreverr + cur);
cur += delta; /* form error * 5 */
bpreverr = belowerr + cur;
belowerr = bnexterr;
cur += delta; /* form error * 7 */
/* At this point cur contains the 7/16 error value to be propagated
* to the next pixel on the current line, and all the errors for the
* next line have been shifted over. We are therefore ready to move on.
*/
input_ptr += dirnc; /* advance input ptr to next column */
output_ptr += dir; /* advance output ptr to next column */
errorptr += dir; /* advance errorptr to current column */
}
/* Post-loop cleanup: we must unload the final error value into the
* final fserrors[] entry. Note we need not unload belowerr because
* it is for the dummy column before or after the actual array.
*/
errorptr[0] = (FSERROR) bpreverr; /* unload prev err into array */
}
cquantize->on_odd_row = (cquantize->on_odd_row ? FALSE : TRUE);
}
}
get_8bit_row(j_compress_ptr cinfo, cjpeg_source_ptr sinfo)
/* This version is for reading 8-bit colormap indexes */
{
bmp_source_ptr source = (bmp_source_ptr)sinfo;
register JSAMPARRAY colormap = source->colormap;
int cmaplen = source->cmap_length;
JSAMPARRAY image_ptr;
register int t;
register JSAMPROW inptr, outptr;
register JDIMENSION col;
if (source->use_inversion_array) {
/* Fetch next row from virtual array */
source->source_row--;
image_ptr = (*cinfo->mem->access_virt_sarray)
((j_common_ptr)cinfo, source->whole_image,
source->source_row, (JDIMENSION)1, FALSE);
inptr = image_ptr[0];
} else {
if (!ReadOK(source->pub.input_file, source->iobuffer, source->row_width))
ERREXIT(cinfo, JERR_INPUT_EOF);
inptr = source->iobuffer;
}
/* Expand the colormap indexes to real data */
outptr = source->pub.buffer[0];
if (cinfo->in_color_space == JCS_GRAYSCALE) {
for (col = cinfo->image_width; col > 0; col--) {
t = GETJSAMPLE(*inptr++);
if (t >= cmaplen)
ERREXIT(cinfo, JERR_BMP_OUTOFRANGE);
*outptr++ = colormap[0][t];
}
} else if (cinfo->in_color_space == JCS_CMYK) {
for (col = cinfo->image_width; col > 0; col--) {
t = GETJSAMPLE(*inptr++);
if (t >= cmaplen)
ERREXIT(cinfo, JERR_BMP_OUTOFRANGE);
rgb_to_cmyk(colormap[0][t], colormap[1][t], colormap[2][t], outptr,
outptr + 1, outptr + 2, outptr + 3);
outptr += 4;
}
} else {
register int rindex = rgb_red[cinfo->in_color_space];
register int gindex = rgb_green[cinfo->in_color_space];
register int bindex = rgb_blue[cinfo->in_color_space];
register int aindex = alpha_index[cinfo->in_color_space];
register int ps = rgb_pixelsize[cinfo->in_color_space];
if (aindex >= 0) {
for (col = cinfo->image_width; col > 0; col--) {
t = GETJSAMPLE(*inptr++);
if (t >= cmaplen)
ERREXIT(cinfo, JERR_BMP_OUTOFRANGE);
outptr[rindex] = colormap[0][t];
outptr[gindex] = colormap[1][t];
outptr[bindex] = colormap[2][t];
outptr[aindex] = 0xFF;
outptr += ps;
}
} else {
for (col = cinfo->image_width; col > 0; col--) {
t = GETJSAMPLE(*inptr++);
if (t >= cmaplen)
ERREXIT(cinfo, JERR_BMP_OUTOFRANGE);
outptr[rindex] = colormap[0][t];
outptr[gindex] = colormap[1][t];
outptr[bindex] = colormap[2][t];
outptr += ps;
}
}
}
return 1;
}
LOCAL void
extract_block (JSAMPARRAY input_data, int start_row, long start_col,
JBLOCK output_data, QUANT_TBL_PTR quanttbl)
/* Extract one 8x8 block from the specified location in the sample array; */
/* perform forward DCT, quantization scaling, and zigzag reordering on it. */
{
/* This routine is heavily used, so it's worth coding it tightly. */
DCTBLOCK block;
#ifdef DCT_ERR_STATS
DCTBLOCK svblock; /* saves input data for comparison */
#endif
{ register JSAMPROW elemptr;
register DCTELEM *localblkptr = block;
register int elemr;
for (elemr = DCTSIZE; elemr > 0; elemr--) {
elemptr = input_data[start_row++] + start_col;
#if DCTSIZE == 8 /* unroll the inner loop */
*localblkptr++ = (DCTELEM) (GETJSAMPLE(*elemptr++) - CENTERJSAMPLE);
*localblkptr++ = (DCTELEM) (GETJSAMPLE(*elemptr++) - CENTERJSAMPLE);
*localblkptr++ = (DCTELEM) (GETJSAMPLE(*elemptr++) - CENTERJSAMPLE);
*localblkptr++ = (DCTELEM) (GETJSAMPLE(*elemptr++) - CENTERJSAMPLE);
*localblkptr++ = (DCTELEM) (GETJSAMPLE(*elemptr++) - CENTERJSAMPLE);
*localblkptr++ = (DCTELEM) (GETJSAMPLE(*elemptr++) - CENTERJSAMPLE);
*localblkptr++ = (DCTELEM) (GETJSAMPLE(*elemptr++) - CENTERJSAMPLE);
*localblkptr++ = (DCTELEM) (GETJSAMPLE(*elemptr++) - CENTERJSAMPLE);
#else
{ register int elemc;
for (elemc = DCTSIZE; elemc > 0; elemc--) {
*localblkptr++ = (DCTELEM) (GETJSAMPLE(*elemptr++) - CENTERJSAMPLE);
}
}
#endif
}
}
#ifdef DCT_ERR_STATS
MEMCOPY(svblock, block, SIZEOF(DCTBLOCK));
#endif
j_fwd_dct(block);
{ register JCOEF temp;
register QUANT_VAL qval;
register int i;
for (i = 0; i < DCTSIZE2; i++) {
qval = *quanttbl++;
temp = (JCOEF) block[ZAG[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.
*/
#ifdef FAST_DIVIDE
#define DIVIDE_BY(a,b) a /= b
#else
#define DIVIDE_BY(a,b) (a >= b) ? (a /= b) : (a = 0)
#endif
if (temp < 0) {
temp = -temp;
temp += qval>>1; /* for rounding */
DIVIDE_BY(temp, qval);
temp = -temp;
} else {
temp += qval>>1; /* for rounding */
DIVIDE_BY(temp, qval);
}
*output_data++ = temp;
}
finish_output_rle (j_decompress_ptr cinfo, djpeg_dest_ptr dinfo)
{
rle_dest_ptr dest = (rle_dest_ptr) dinfo;
rle_hdr header; /* Output file information */
rle_pixel **rle_row, *red, *green, *blue;
JSAMPROW output_row;
char cmapcomment[80];
int row, col;
int ci;
#ifdef LIBJPEG_PROGRESS_REPORT
cd_progress_ptr progress = (cd_progress_ptr) cinfo->progress;
#endif
/* Initialize the header info */
header = *rle_hdr_init(NULL);
header.rle_file = dest->pub.output_file;
header.xmin = 0;
header.xmax = cinfo->output_width - 1;
header.ymin = 0;
header.ymax = cinfo->output_height - 1;
header.alpha = 0;
header.ncolors = cinfo->output_components;
for (ci = 0; ci < cinfo->output_components; ci++) {
RLE_SET_BIT(header, ci);
}
if (cinfo->quantize_colors) {
header.ncmap = cinfo->out_color_components;
header.cmaplen = CMAPBITS;
header.cmap = dest->colormap;
/* Add a comment to the output image with the true colormap length. */
sprintf(cmapcomment, "color_map_length=%d", cinfo->actual_number_of_colors);
rle_putcom(cmapcomment, &header);
}
/* Emit the RLE header and color map (if any) */
rle_put_setup(&header);
/* Now output the RLE data from our virtual array.
* We assume here that (a) rle_pixel is represented the same as JSAMPLE,
* and (b) we are not on a machine where FAR pointers differ from regular.
*/
#ifdef LIBJPEG_PROGRESS_REPORT
if (progress != NULL) {
progress->pub.pass_limit = cinfo->output_height;
progress->pub.pass_counter = 0;
(*progress->pub.progress_monitor) ((j_common_ptr) cinfo);
}
#endif
if (cinfo->output_components == 1) {
for (row = cinfo->output_height-1; row >= 0; row--) {
rle_row = (rle_pixel **) (*cinfo->mem->access_virt_sarray)
((j_common_ptr) cinfo, dest->image,
(JDIMENSION) row, (JDIMENSION) 1, FALSE);
rle_putrow(rle_row, (int) cinfo->output_width, &header);
#ifdef LIBJPEG_PROGRESS_REPORT
if (progress != NULL) {
progress->pub.pass_counter++;
(*progress->pub.progress_monitor) ((j_common_ptr) cinfo);
}
#endif
}
} else {
for (row = cinfo->output_height-1; row >= 0; row--) {
rle_row = (rle_pixel **) dest->rle_row;
output_row = * (*cinfo->mem->access_virt_sarray)
((j_common_ptr) cinfo, dest->image,
(JDIMENSION) row, (JDIMENSION) 1, FALSE);
red = rle_row[0];
green = rle_row[1];
blue = rle_row[2];
for (col = cinfo->output_width; col > 0; col--) {
*red++ = GETJSAMPLE(*output_row++);
*green++ = GETJSAMPLE(*output_row++);
*blue++ = GETJSAMPLE(*output_row++);
}
rle_putrow(rle_row, (int) cinfo->output_width, &header);
#ifdef LIBJPEG_PROGRESS_REPORT
if (progress != NULL) {
progress->pub.pass_counter++;
(*progress->pub.progress_monitor) ((j_common_ptr) cinfo);
}
#endif
}
}
#ifdef LIBJPEG_PROGRESS_REPORT
if (progress != NULL)
progress->completed_extra_passes++;
#endif
/* Emit file trailer */
rle_puteof(&header);
fflush(dest->pub.output_file);
if (ferror(dest->pub.output_file))
ERREXIT(cinfo, JERR_FILE_WRITE);
}
start_output_rle (j_decompress_ptr cinfo, djpeg_dest_ptr dinfo)
{
rle_dest_ptr dest = (rle_dest_ptr) dinfo;
size_t cmapsize;
int i, ci;
#ifdef LIBJPEG_PROGRESS_REPORT
cd_progress_ptr progress = (cd_progress_ptr) cinfo->progress;
#endif
/*
* Make sure the image can be stored in RLE format.
*
* - RLE stores image dimensions as *signed* 16 bit integers. JPEG
* uses unsigned, so we have to check the width.
*
* - Colorspace is expected to be grayscale or RGB.
*
* - The number of channels (components) is expected to be 1 (grayscale/
* pseudocolor) or 3 (truecolor/directcolor).
* (could be 2 or 4 if using an alpha channel, but we aren't)
*/
if (cinfo->output_width > 32767 || cinfo->output_height > 32767)
ERREXIT2(cinfo, JERR_RLE_DIMENSIONS, cinfo->output_width,
cinfo->output_height);
if (cinfo->out_color_space != JCS_GRAYSCALE &&
cinfo->out_color_space != JCS_RGB)
ERREXIT(cinfo, JERR_RLE_COLORSPACE);
if (cinfo->output_components != 1 && cinfo->output_components != 3)
ERREXIT1(cinfo, JERR_RLE_TOOMANYCHANNELS, cinfo->num_components);
/* Convert colormap, if any, to RLE format. */
dest->colormap = NULL;
if (cinfo->quantize_colors) {
/* Allocate storage for RLE-style cmap, zero any extra entries */
cmapsize = cinfo->out_color_components * CMAPLENGTH * SIZEOF(rle_map);
dest->colormap = (rle_map *) (*cinfo->mem->alloc_small)
((j_common_ptr) cinfo, JPOOL_IMAGE, cmapsize);
MEMZERO(dest->colormap, cmapsize);
/* Save away data in RLE format --- note 8-bit left shift! */
/* Shifting would need adjustment for JSAMPLEs wider than 8 bits. */
for (ci = 0; ci < cinfo->out_color_components; ci++) {
for (i = 0; i < cinfo->actual_number_of_colors; i++) {
dest->colormap[ci * CMAPLENGTH + i] =
GETJSAMPLE(cinfo->colormap[ci][i]) << 8;
}
}
}
/* Set the output buffer to the first row */
dest->pub.buffer = (*cinfo->mem->access_virt_sarray)
((j_common_ptr) cinfo, dest->image, (JDIMENSION) 0, (JDIMENSION) 1, TRUE);
dest->pub.buffer_height = 1;
dest->pub.put_pixel_rows = rle_put_pixel_rows;
#ifdef LIBJPEG_PROGRESS_REPORT
if (progress != NULL) {
progress->total_extra_passes++; /* count file writing as separate pass */
}
#endif
}
void
il_quantize_fs_dither(il_container *ic, const uint8 *mask,
const uint8 *input_buf, int x_offset,
uint8 XP_HUGE *output_buf, int width)
{
my_cquantize_ptr cquantize;
register LOCFSERROR r_cur, g_cur, b_cur; /* current error or pixel
value */
LOCFSERROR r_belowerr, g_belowerr, b_belowerr; /* error for pixel below
cur */
LOCFSERROR r_bpreverr, g_bpreverr, b_bpreverr; /* error for below/prev
col */
LOCFSERROR r_bnexterr, g_bnexterr, b_bnexterr; /* error for below/next
col */
LOCFSERROR delta;
FSERRPTR r_errorptr, g_errorptr, b_errorptr; /* fserrors[] at column
before current */
const JSAMPLE* input_ptr;
JSAMPLE XP_HUGE * output_ptr;
IL_ColorMap *cmap = &ic->image->header.color_space->cmap;
IL_RGB *map = cmap->map; /* The colormap array. */
IL_RGB *map_entry; /* Current entry in the colormap. */
uint8 *lookup_table = cmap->table; /* Lookup table for the colormap. */
const uint8 *maskp;
uint8 map_index;
int dir; /* 1 for left-to-right, -1 for right-to-left */
JDIMENSION col;
JSAMPLE *range_limit = the_sample_range_limit;
SHIFT_TEMPS
cquantize = (my_cquantize_ptr) ic->quantize;
output_buf += x_offset;
/* Initialize output values to 0 so can process components separately */
if (mask) {
output_ptr = output_buf;
maskp = mask;
for (col = width; col > 0; col--)
*output_ptr++ &= ~*maskp++;
} else {
XP_BZERO((void XP_HUGE *) output_buf,
(size_t) (width * SIZEOF(JSAMPLE)));
}
input_ptr = input_buf;
output_ptr = output_buf;
maskp = mask;
if (cquantize->on_odd_row) {
int total_offset;
/* work right to left in this row */
input_ptr += 3 * width - 1; /* so point to the blue sample of the
rightmost pixel */
output_ptr += width-1;
dir = -1;
/* => entry after last column */
total_offset = x_offset + (width + 1);
r_errorptr = cquantize->fserrors[0] + total_offset;
g_errorptr = cquantize->fserrors[1] + total_offset;
b_errorptr = cquantize->fserrors[2] + total_offset;
maskp += (width - 1);
}
else {
/* work left to right in this row */
dir = 1;
/* => entry before first column */
r_errorptr = cquantize->fserrors[0] + x_offset;
g_errorptr = cquantize->fserrors[1] + x_offset;
b_errorptr = cquantize->fserrors[2] + x_offset;
}
/* Preset error values: no error propagated to first pixel from left */
r_cur = g_cur = b_cur = 0;
/* and no error propagated to row below yet */
r_belowerr = g_belowerr = b_belowerr = 0;
r_bpreverr = g_bpreverr = b_bpreverr = 0;
for (col = width; col > 0; col--) {
/* cur holds the error propagated from the previous pixel on the
* current line. Add the error propagated from the previous line
* to form the complete error correction term for this pixel, and
* round the error term (which is expressed * 16) to an integer.
* RIGHT_SHIFT rounds towards minus infinity, so adding 8 is correct
* for either sign of the error value.
* Note: errorptr points to *previous* column's array entry.
*/
r_cur = RIGHT_SHIFT(r_cur + r_errorptr[dir] + 8, 4);
g_cur = RIGHT_SHIFT(g_cur + g_errorptr[dir] + 8, 4);
b_cur = RIGHT_SHIFT(b_cur + b_errorptr[dir] + 8, 4);
/* Form pixel value + error, and range-limit to 0..MAXJSAMPLE.
* The maximum error is +- MAXJSAMPLE; this sets the required size
* of the range_limit array.
*/
if (dir > 0) {
r_cur += GETJSAMPLE(*input_ptr);
r_cur = GETJSAMPLE(range_limit[r_cur]);
input_ptr++;
g_cur += GETJSAMPLE(*input_ptr);
g_cur = GETJSAMPLE(range_limit[g_cur]);
input_ptr++;
b_cur += GETJSAMPLE(*input_ptr);
b_cur = GETJSAMPLE(range_limit[b_cur]);
input_ptr++;
}
else {
b_cur += GETJSAMPLE(*input_ptr);
b_cur = GETJSAMPLE(range_limit[b_cur]);
input_ptr--;
g_cur += GETJSAMPLE(*input_ptr);
g_cur = GETJSAMPLE(range_limit[g_cur]);
input_ptr--;
r_cur += GETJSAMPLE(*input_ptr);
r_cur = GETJSAMPLE(range_limit[r_cur]);
input_ptr--;
}
/* Select output value, accumulate into output code for this pixel */
map_index = COLORMAP_INDEX(lookup_table, r_cur, g_cur, b_cur);
if (mask) {
if (*maskp)
*output_ptr = map_index;
maskp += dir;
} else {
*output_ptr = map_index;
}
/* Compute the actual representation error at this pixel */
map_entry = map + map_index;
r_cur -= GETJSAMPLE(map_entry->red);
g_cur -= GETJSAMPLE(map_entry->green);
b_cur -= GETJSAMPLE(map_entry->blue);
/* Compute error fractions to be propagated to adjacent pixels.
* Add these into the running sums, and simultaneously shift the
* next-line error sums left by 1 column.
*/
r_bnexterr = r_cur;
delta = r_cur * 2;
r_cur += delta; /* form error * 3 */
r_errorptr[0] = (FSERROR) (r_bpreverr + r_cur);
r_cur += delta; /* form error * 5 */
r_bpreverr = r_belowerr + r_cur;
r_belowerr = r_bnexterr;
r_cur += delta; /* form error * 7 */
g_bnexterr = g_cur;
delta = g_cur * 2;
g_cur += delta; /* form error * 3 */
g_errorptr[0] = (FSERROR) (g_bpreverr + g_cur);
g_cur += delta; /* form error * 5 */
g_bpreverr = g_belowerr + g_cur;
g_belowerr = g_bnexterr;
g_cur += delta; /* form error * 7 */
b_bnexterr = b_cur;
delta = b_cur * 2;
b_cur += delta; /* form error * 3 */
b_errorptr[0] = (FSERROR) (b_bpreverr + b_cur);
b_cur += delta; /* form error * 5 */
b_bpreverr = b_belowerr + b_cur;
b_belowerr = b_bnexterr;
b_cur += delta; /* form error * 7 */
/* At this point cur contains the 7/16 error value to be propagated
* to the next pixel on the current line, and all the errors for the
* next line have been shifted over. We are therefore ready to move on.
* Note: the input_ptr has already been advanced.
*/
output_ptr += dir; /* advance output ptr to next column */
r_errorptr += dir; /* advance errorptr to current column */
g_errorptr += dir; /* advance errorptr to current column */
b_errorptr += dir; /* advance errorptr to current column */
}
/* Post-loop cleanup: we must unload the final error value into the
* final fserrors[] entry. Note we need not unload belowerr because
* it is for the dummy column before or after the actual array.
*/
r_errorptr[0] = (FSERROR) r_bpreverr; /* unload prev err into array */
g_errorptr[0] = (FSERROR) g_bpreverr; /* unload prev err into array */
b_errorptr[0] = (FSERROR) b_bpreverr; /* unload prev err into array */
cquantize->on_odd_row = (cquantize->on_odd_row ? FALSE : TRUE);
}
METHODDEF void
fullsize_smooth_downsample(j_compress_ptr cinfo, jpeg_component_info *compptr,
JSAMPARRAY input_data, JSAMPARRAY output_data)
{
int outrow;
JDIMENSION colctr;
JDIMENSION output_cols = compptr->width_in_blocks * DCTSIZE;
register JSAMPROW inptr, above_ptr, below_ptr, outptr;
INT32 membersum, neighsum, memberscale, neighscale;
int colsum, lastcolsum, nextcolsum;
/* Expand input data enough to let all the output samples be generated
* by the standard loop. Special-casing padded output would be more
* efficient.
*/
expand_right_edge(input_data - 1, cinfo->max_v_samp_factor + 2,
cinfo->image_width, output_cols);
/* Each of the eight neighbor pixels contributes a fraction SF to the
* smoothed pixel, while the main pixel contributes (1-8*SF). In order
* to use integer arithmetic, these factors are multiplied by 2^16 = 65536.
* Also recall that SF = smoothing_factor / 1024.
*/
memberscale = 65536L - cinfo->smoothing_factor * 512L; /* scaled 1-8*SF */
neighscale = cinfo->smoothing_factor * 64; /* scaled SF */
for(outrow = 0; outrow < compptr->v_samp_factor; outrow++)
{
outptr = output_data[outrow];
inptr = input_data[outrow];
above_ptr = input_data[outrow - 1];
below_ptr = input_data[outrow + 1];
/* Special case for first column */
colsum = GETJSAMPLE(*above_ptr++) + GETJSAMPLE(*below_ptr++) +
GETJSAMPLE(*inptr);
membersum = GETJSAMPLE(*inptr++);
nextcolsum = GETJSAMPLE(*above_ptr) + GETJSAMPLE(*below_ptr) +
GETJSAMPLE(*inptr);
neighsum = colsum + (colsum - membersum) + nextcolsum;
membersum = membersum * memberscale + neighsum * neighscale;
*outptr++ = (JSAMPLE)((membersum + 32768) >> 16);
lastcolsum = colsum;
colsum = nextcolsum;
for(colctr = output_cols - 2; colctr > 0; colctr--)
{
membersum = GETJSAMPLE(*inptr++);
above_ptr++;
below_ptr++;
nextcolsum = GETJSAMPLE(*above_ptr) + GETJSAMPLE(*below_ptr) +
GETJSAMPLE(*inptr);
neighsum = lastcolsum + (colsum - membersum) + nextcolsum;
membersum = membersum * memberscale + neighsum * neighscale;
*outptr++ = (JSAMPLE)((membersum + 32768) >> 16);
lastcolsum = colsum;
colsum = nextcolsum;
}
/* Special case for last column */
membersum = GETJSAMPLE(*inptr);
neighsum = lastcolsum + (colsum - membersum) + colsum;
membersum = membersum * memberscale + neighsum * neighscale;
*outptr = (JSAMPLE)((membersum + 32768) >> 16);
}
}
jpeg_fdct_ifast (DCTELEM * data, JSAMPARRAY sample_data, JDIMENSION start_col)
{
DCTELEM tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
DCTELEM tmp10, tmp11, tmp12, tmp13;
DCTELEM z1, z2, z3, z4, z5, z11, z13;
DCTELEM *dataptr;
JSAMPROW elemptr;
int ctr;
SHIFT_TEMPS
/* Pass 1: process rows. */
dataptr = data;
for (ctr = 0; ctr < DCTSIZE; ctr++) {
elemptr = sample_data[ctr] + start_col;
/* Load data into workspace */
tmp0 = GETJSAMPLE(elemptr[0]) + GETJSAMPLE(elemptr[7]);
tmp7 = GETJSAMPLE(elemptr[0]) - GETJSAMPLE(elemptr[7]);
tmp1 = GETJSAMPLE(elemptr[1]) + GETJSAMPLE(elemptr[6]);
tmp6 = GETJSAMPLE(elemptr[1]) - GETJSAMPLE(elemptr[6]);
tmp2 = GETJSAMPLE(elemptr[2]) + GETJSAMPLE(elemptr[5]);
tmp5 = GETJSAMPLE(elemptr[2]) - GETJSAMPLE(elemptr[5]);
tmp3 = GETJSAMPLE(elemptr[3]) + GETJSAMPLE(elemptr[4]);
tmp4 = GETJSAMPLE(elemptr[3]) - GETJSAMPLE(elemptr[4]);
/* Even part */
tmp10 = tmp0 + tmp3; /* phase 2 */
tmp13 = tmp0 - tmp3;
tmp11 = tmp1 + tmp2;
tmp12 = tmp1 - tmp2;
/* Apply unsigned->signed conversion */
dataptr[0] = tmp10 + tmp11 - 8 * CENTERJSAMPLE; /* phase 3 */
dataptr[4] = tmp10 - tmp11;
z1 = MULTIPLY(tmp12 + tmp13, FIX_0_707106781); /* c4 */
dataptr[2] = tmp13 + z1; /* phase 5 */
dataptr[6] = tmp13 - z1;
/* Odd part */
tmp10 = tmp4 + tmp5; /* phase 2 */
tmp11 = tmp5 + tmp6;
tmp12 = tmp6 + tmp7;
/* The rotator is modified from fig 4-8 to avoid extra negations. */
z5 = MULTIPLY(tmp10 - tmp12, FIX_0_382683433); /* c6 */
z2 = MULTIPLY(tmp10, FIX_0_541196100) + z5; /* c2-c6 */
z4 = MULTIPLY(tmp12, FIX_1_306562965) + z5; /* c2+c6 */
z3 = MULTIPLY(tmp11, FIX_0_707106781); /* c4 */
z11 = tmp7 + z3; /* phase 5 */
z13 = tmp7 - z3;
dataptr[5] = z13 + z2; /* phase 6 */
dataptr[3] = z13 - z2;
dataptr[1] = z11 + z4;
dataptr[7] = z11 - z4;
dataptr += DCTSIZE; /* advance pointer to next row */
}
/* Pass 2: process columns. */
dataptr = data;
for (ctr = DCTSIZE-1; ctr >= 0; ctr--) {
tmp0 = dataptr[DCTSIZE*0] + dataptr[DCTSIZE*7];
tmp7 = dataptr[DCTSIZE*0] - dataptr[DCTSIZE*7];
tmp1 = dataptr[DCTSIZE*1] + dataptr[DCTSIZE*6];
tmp6 = dataptr[DCTSIZE*1] - dataptr[DCTSIZE*6];
tmp2 = dataptr[DCTSIZE*2] + dataptr[DCTSIZE*5];
tmp5 = dataptr[DCTSIZE*2] - dataptr[DCTSIZE*5];
tmp3 = dataptr[DCTSIZE*3] + dataptr[DCTSIZE*4];
tmp4 = dataptr[DCTSIZE*3] - dataptr[DCTSIZE*4];
/* Even part */
tmp10 = tmp0 + tmp3; /* phase 2 */
tmp13 = tmp0 - tmp3;
tmp11 = tmp1 + tmp2;
tmp12 = tmp1 - tmp2;
dataptr[DCTSIZE*0] = tmp10 + tmp11; /* phase 3 */
dataptr[DCTSIZE*4] = tmp10 - tmp11;
z1 = MULTIPLY(tmp12 + tmp13, FIX_0_707106781); /* c4 */
dataptr[DCTSIZE*2] = tmp13 + z1; /* phase 5 */
dataptr[DCTSIZE*6] = tmp13 - z1;
/* Odd part */
tmp10 = tmp4 + tmp5; /* phase 2 */
tmp11 = tmp5 + tmp6;
tmp12 = tmp6 + tmp7;
/* The rotator is modified from fig 4-8 to avoid extra negations. */
z5 = MULTIPLY(tmp10 - tmp12, FIX_0_382683433); /* c6 */
z2 = MULTIPLY(tmp10, FIX_0_541196100) + z5; /* c2-c6 */
z4 = MULTIPLY(tmp12, FIX_1_306562965) + z5; /* c2+c6 */
z3 = MULTIPLY(tmp11, FIX_0_707106781); /* c4 */
z11 = tmp7 + z3; /* phase 5 */
z13 = tmp7 - z3;
dataptr[DCTSIZE*5] = z13 + z2; /* phase 6 */
dataptr[DCTSIZE*3] = z13 - z2;
dataptr[DCTSIZE*1] = z11 + z4;
dataptr[DCTSIZE*7] = z11 - z4;
dataptr++; /* advance pointer to next column */
}
}
h2v2_merged_upsample_internal (j_decompress_ptr cinfo,
JSAMPIMAGE input_buf,
JDIMENSION in_row_group_ctr,
JSAMPARRAY output_buf)
{
my_upsample_ptr upsample = (my_upsample_ptr) cinfo->upsample;
register int y, cred, cgreen, cblue;
int cb, cr;
register JSAMPROW outptr0, outptr1;
JSAMPROW inptr00, inptr01, inptr1, inptr2;
JDIMENSION col;
/* copy these pointers into registers if possible */
register JSAMPLE * range_limit = cinfo->sample_range_limit;
int * Crrtab = upsample->Cr_r_tab;
int * Cbbtab = upsample->Cb_b_tab;
INT32 * Crgtab = upsample->Cr_g_tab;
INT32 * Cbgtab = upsample->Cb_g_tab;
SHIFT_TEMPS
inptr00 = input_buf[0][in_row_group_ctr*2];
inptr01 = input_buf[0][in_row_group_ctr*2 + 1];
inptr1 = input_buf[1][in_row_group_ctr];
inptr2 = input_buf[2][in_row_group_ctr];
outptr0 = output_buf[0];
outptr1 = output_buf[1];
/* Loop for each group of output pixels */
for (col = cinfo->output_width >> 1; col > 0; col--) {
/* Do the chroma part of the calculation */
cb = GETJSAMPLE(*inptr1++);
cr = GETJSAMPLE(*inptr2++);
cred = Crrtab[cr];
cgreen = (int) RIGHT_SHIFT(Cbgtab[cb] + Crgtab[cr], SCALEBITS);
cblue = Cbbtab[cb];
/* Fetch 4 Y values and emit 4 pixels */
y = GETJSAMPLE(*inptr00++);
outptr0[RGB_RED] = range_limit[y + cred];
outptr0[RGB_GREEN] = range_limit[y + cgreen];
outptr0[RGB_BLUE] = range_limit[y + cblue];
#ifdef RGB_ALPHA
outptr0[RGB_ALPHA] = 0xFF;
#endif
outptr0 += RGB_PIXELSIZE;
y = GETJSAMPLE(*inptr00++);
outptr0[RGB_RED] = range_limit[y + cred];
outptr0[RGB_GREEN] = range_limit[y + cgreen];
outptr0[RGB_BLUE] = range_limit[y + cblue];
#ifdef RGB_ALPHA
outptr0[RGB_ALPHA] = 0xFF;
#endif
outptr0 += RGB_PIXELSIZE;
y = GETJSAMPLE(*inptr01++);
outptr1[RGB_RED] = range_limit[y + cred];
outptr1[RGB_GREEN] = range_limit[y + cgreen];
outptr1[RGB_BLUE] = range_limit[y + cblue];
#ifdef RGB_ALPHA
outptr1[RGB_ALPHA] = 0xFF;
#endif
outptr1 += RGB_PIXELSIZE;
y = GETJSAMPLE(*inptr01++);
outptr1[RGB_RED] = range_limit[y + cred];
outptr1[RGB_GREEN] = range_limit[y + cgreen];
outptr1[RGB_BLUE] = range_limit[y + cblue];
#ifdef RGB_ALPHA
outptr1[RGB_ALPHA] = 0xFF;
#endif
outptr1 += RGB_PIXELSIZE;
}
/* If image width is odd, do the last output column separately */
if (cinfo->output_width & 1) {
cb = GETJSAMPLE(*inptr1);
cr = GETJSAMPLE(*inptr2);
cred = Crrtab[cr];
cgreen = (int) RIGHT_SHIFT(Cbgtab[cb] + Crgtab[cr], SCALEBITS);
cblue = Cbbtab[cb];
y = GETJSAMPLE(*inptr00);
outptr0[RGB_RED] = range_limit[y + cred];
outptr0[RGB_GREEN] = range_limit[y + cgreen];
outptr0[RGB_BLUE] = range_limit[y + cblue];
#ifdef RGB_ALPHA
outptr0[RGB_ALPHA] = 0xFF;
#endif
y = GETJSAMPLE(*inptr01);
outptr1[RGB_RED] = range_limit[y + cred];
outptr1[RGB_GREEN] = range_limit[y + cgreen];
outptr1[RGB_BLUE] = range_limit[y + cblue];
#ifdef RGB_ALPHA
outptr1[RGB_ALPHA] = 0xFF;
#endif
}
}
h2v2_merged_upsample_565D (j_decompress_ptr cinfo,
JSAMPIMAGE input_buf, JDIMENSION in_row_group_ctr,
JSAMPARRAY output_buf)
{
my_upsample_ptr upsample = (my_upsample_ptr) cinfo->upsample;
register int y, cred, cgreen, cblue;
int cb, cr;
register JSAMPROW outptr0, outptr1;
JSAMPROW inptr00, inptr01, inptr1, inptr2;
JDIMENSION col;
/* copy these pointers into registers if possible */
register JSAMPLE * range_limit = cinfo->sample_range_limit;
int * Crrtab = upsample->Cr_r_tab;
int * Cbbtab = upsample->Cb_b_tab;
INT32 * Crgtab = upsample->Cr_g_tab;
INT32 * Cbgtab = upsample->Cb_g_tab;
INT32 d0 = dither_matrix[cinfo->output_scanline & DITHER_MASK];
INT32 d1 = dither_matrix[(cinfo->output_scanline+1) & DITHER_MASK];
unsigned int r, g, b;
INT32 rgb;
SHIFT_TEMPS
inptr00 = input_buf[0][in_row_group_ctr*2];
inptr01 = input_buf[0][in_row_group_ctr*2 + 1];
inptr1 = input_buf[1][in_row_group_ctr];
inptr2 = input_buf[2][in_row_group_ctr];
outptr0 = output_buf[0];
outptr1 = output_buf[1];
/* Loop for each group of output pixels */
for (col = cinfo->output_width >> 1; col > 0; col--) {
/* Do the chroma part of the calculation */
cb = GETJSAMPLE(*inptr1++);
cr = GETJSAMPLE(*inptr2++);
cred = Crrtab[cr];
cgreen = (int) RIGHT_SHIFT(Cbgtab[cb] + Crgtab[cr], SCALEBITS);
cblue = Cbbtab[cb];
/* Fetch 4 Y values and emit 4 pixels */
y = GETJSAMPLE(*inptr00++);
r = range_limit[DITHER_565_R(y + cred, d0)];
g = range_limit[DITHER_565_G(y + cgreen, d0)];
b = range_limit[DITHER_565_B(y + cblue, d0)];
d0 = DITHER_ROTATE(d0);
rgb = PACK_SHORT_565(r, g, b);
y = GETJSAMPLE(*inptr00++);
r = range_limit[DITHER_565_R(y + cred, d1)];
g = range_limit[DITHER_565_G(y + cgreen, d1)];
b = range_limit[DITHER_565_B(y + cblue, d1)];
d1 = DITHER_ROTATE(d1);
rgb = PACK_TWO_PIXELS(rgb, PACK_SHORT_565(r, g, b));
WRITE_TWO_PIXELS(outptr0, rgb);
outptr0 += 4;
y = GETJSAMPLE(*inptr01++);
r = range_limit[DITHER_565_R(y + cred, d0)];
g = range_limit[DITHER_565_G(y + cgreen, d0)];
b = range_limit[DITHER_565_B(y + cblue, d0)];
d0 = DITHER_ROTATE(d0);
rgb = PACK_SHORT_565(r, g, b);
y = GETJSAMPLE(*inptr01++);
r = range_limit[DITHER_565_R(y + cred, d1)];
g = range_limit[DITHER_565_G(y + cgreen, d1)];
b = range_limit[DITHER_565_B(y + cblue, d1)];
d1 = DITHER_ROTATE(d1);
rgb = PACK_TWO_PIXELS(rgb, PACK_SHORT_565(r, g, b));
WRITE_TWO_PIXELS(outptr1, rgb);
outptr1 += 4;
}
/* If image width is odd, do the last output column separately */
if (cinfo->output_width & 1) {
cb = GETJSAMPLE(*inptr1);
cr = GETJSAMPLE(*inptr2);
cred = Crrtab[cr];
cgreen = (int) RIGHT_SHIFT(Cbgtab[cb] + Crgtab[cr], SCALEBITS);
cblue = Cbbtab[cb];
y = GETJSAMPLE(*inptr00);
r = range_limit[DITHER_565_R(y + cred, d0)];
g = range_limit[DITHER_565_G(y + cgreen, d0)];
b = range_limit[DITHER_565_B(y + cblue, d0)];
rgb = PACK_SHORT_565(r, g, b);
*(INT16*)outptr0 = rgb;
y = GETJSAMPLE(*inptr01);
r = range_limit[DITHER_565_R(y + cred, d1)];
g = range_limit[DITHER_565_G(y + cgreen, d1)];
b = range_limit[DITHER_565_B(y + cblue, d1)];
rgb = PACK_SHORT_565(r, g, b);
*(INT16*)outptr1 = rgb;
}
}
METHODDEF void
h2v2_fancy_upsample(j_decompress_ptr cinfo, jpeg_component_info *compptr,
JSAMPARRAY input_data, JSAMPARRAY *output_data_ptr)
{
JSAMPARRAY output_data = *output_data_ptr;
register JSAMPROW inptr0, inptr1, outptr;
#if BITS_IN_JSAMPLE == 8
register int thiscolsum, lastcolsum, nextcolsum;
#else
register INT32 thiscolsum, lastcolsum, nextcolsum;
#endif
register JDIMENSION colctr;
int inrow, outrow, v;
inrow = outrow = 0;
while(outrow < cinfo->max_v_samp_factor)
{
for(v = 0; v < 2; v++)
{
/* inptr0 points to nearest input row, inptr1 points to next nearest */
inptr0 = input_data[inrow];
if(v == 0) /* next nearest is row above */
{
inptr1 = input_data[inrow - 1];
}
else /* next nearest is row below */
{
inptr1 = input_data[inrow + 1];
}
outptr = output_data[outrow++];
/* Special case for first column */
thiscolsum = GETJSAMPLE(*inptr0++) * 3 + GETJSAMPLE(*inptr1++);
nextcolsum = GETJSAMPLE(*inptr0++) * 3 + GETJSAMPLE(*inptr1++);
*outptr++ = (JSAMPLE)((thiscolsum * 4 + 8) >> 4);
*outptr++ = (JSAMPLE)((thiscolsum * 3 + nextcolsum + 7) >> 4);
lastcolsum = thiscolsum;
thiscolsum = nextcolsum;
for(colctr = compptr->downsampled_width - 2; colctr > 0; colctr--)
{
/* General case: 3/4 * nearer pixel + 1/4 * further pixel in each */
/* dimension, thus 9/16, 3/16, 3/16, 1/16 overall */
nextcolsum = GETJSAMPLE(*inptr0++) * 3 + GETJSAMPLE(*inptr1++);
*outptr++ = (JSAMPLE)((thiscolsum * 3 + lastcolsum + 8) >> 4);
*outptr++ = (JSAMPLE)((thiscolsum * 3 + nextcolsum + 7) >> 4);
lastcolsum = thiscolsum;
thiscolsum = nextcolsum;
}
/* Special case for last column */
*outptr++ = (JSAMPLE)((thiscolsum * 3 + lastcolsum + 8) >> 4);
*outptr++ = (JSAMPLE)((thiscolsum * 4 + 7) >> 4);
}
inrow++;
}
}
h2v1_merged_upsample_565 (j_decompress_ptr cinfo,
JSAMPIMAGE input_buf, JDIMENSION in_row_group_ctr,
JSAMPARRAY output_buf)
{
my_upsample_ptr upsample = (my_upsample_ptr) cinfo->upsample;
register int y, cred, cgreen, cblue;
int cb, cr;
register JSAMPROW outptr;
JSAMPROW inptr0, inptr1, inptr2;
JDIMENSION col;
/* copy these pointers into registers if possible */
register JSAMPLE * range_limit = cinfo->sample_range_limit;
int * Crrtab = upsample->Cr_r_tab;
int * Cbbtab = upsample->Cb_b_tab;
INT32 * Crgtab = upsample->Cr_g_tab;
INT32 * Cbgtab = upsample->Cb_g_tab;
unsigned int r, g, b;
INT32 rgb;
SHIFT_TEMPS
inptr0 = input_buf[0][in_row_group_ctr];
inptr1 = input_buf[1][in_row_group_ctr];
inptr2 = input_buf[2][in_row_group_ctr];
outptr = output_buf[0];
/* Loop for each pair of output pixels */
for (col = cinfo->output_width >> 1; col > 0; col--) {
/* Do the chroma part of the calculation */
cb = GETJSAMPLE(*inptr1++);
cr = GETJSAMPLE(*inptr2++);
cred = Crrtab[cr];
cgreen = (int) RIGHT_SHIFT(Cbgtab[cb] + Crgtab[cr], SCALEBITS);
cblue = Cbbtab[cb];
/* Fetch 2 Y values and emit 2 pixels */
y = GETJSAMPLE(*inptr0++);
r = range_limit[y + cred];
g = range_limit[y + cgreen];
b = range_limit[y + cblue];
rgb = PACK_SHORT_565(r, g, b);
y = GETJSAMPLE(*inptr0++);
r = range_limit[y + cred];
g = range_limit[y + cgreen];
b = range_limit[y + cblue];
rgb = PACK_TWO_PIXELS(rgb, PACK_SHORT_565(r, g, b));
WRITE_TWO_PIXELS(outptr, rgb);
outptr += 4;
}
/* If image width is odd, do the last output column separately */
if (cinfo->output_width & 1) {
cb = GETJSAMPLE(*inptr1);
cr = GETJSAMPLE(*inptr2);
cred = Crrtab[cr];
cgreen = (int) RIGHT_SHIFT(Cbgtab[cb] + Crgtab[cr], SCALEBITS);
cblue = Cbbtab[cb];
y = GETJSAMPLE(*inptr0);
r = range_limit[y + cred];
g = range_limit[y + cgreen];
b = range_limit[y + cblue];
rgb = PACK_SHORT_565(r, g, b);
*(INT16*)outptr = rgb;
}
}
write_colormap_buf (j_decompress_ptr cinfo, bmp_dest_ptr dest,
int map_colors, int map_entry_size,char *buf)
{
JSAMPARRAY colormap = cinfo->colormap;
int num_colors = cinfo->actual_number_of_colors;
FILE * outfile = dest->pub.output_file;
JSAMPARRAY outbuffer= dest->pub.buffer;
int i,k=54;
if (colormap != NULL) {
if (cinfo->out_color_components == 3) {
/* Normal case with RGB colormap */
for (i = 0; i < num_colors; i++) {
*(buf+k)=GETJSAMPLE(colormap[2][i]);
k++;
*(buf+k)=GETJSAMPLE(colormap[1][i]);
k++;
*(buf+k)=GETJSAMPLE(colormap[0][i]);
k++;
if (map_entry_size == 4){
//putc(0, outfile);
*(buf+k)=0;
k++;
}
}
} else {
/* Grayscale colormap (only happens with grayscale quantization) */
for (i = 0; i < num_colors; i++) {
*(buf+k)=GETJSAMPLE(colormap[0][i]);
k++;
*(buf+k)=GETJSAMPLE(colormap[0][i]);
k++;
*(buf+k)=GETJSAMPLE(colormap[0][i]);
k++;
if (map_entry_size == 4) {
*(buf+k)=0;
k++;
}
}
}
} else {
/* If no colormap, must be grayscale data. Generate a linear "map". */
for (i = 0; i < 256; i++) {
*(buf+k)=i;
k++;
*(buf+k)=i;
k++;
*(buf+k)=i;
k++;
if (map_entry_size == 4) {
//putc(0, outfile);
*(buf+k)=0;
k++;
}
}
}
/* Pad colormap with zeros to ensure specified number of colormap entries */
if (i > map_colors)
ERREXIT1(cinfo, JERR_TOO_MANY_COLORS, i);
for (; i < map_colors; i++) {
*(buf+k)=0;
k++;
*(buf+k)=0;
k++;
*(buf+k)=0;
k++;
if (map_entry_size == 4) {
//putc(0, outfile);
*(buf+k)=0;
k++;
}
}
}
quantize_ord_dither (j_decompress_ptr cinfo, JSAMPARRAY input_buf,
JSAMPARRAY output_buf, int num_rows)
/* General case, with ordered dithering */
{
my_cquantize_ptr cquantize = (my_cquantize_ptr) cinfo->cquantize;
register JSAMPROW input_ptr;
register JSAMPROW output_ptr;
JSAMPROW colorindex_ci;
int * dither; /* points to active row of dither matrix */
int row_index, col_index; /* current indexes into dither matrix */
int nc = cinfo->out_color_components;
int ci;
int row;
JDIMENSION col;
JDIMENSION width = cinfo->output_width;
#if defined(__INTEL_COMPILER) && 1 /* VDM auto patch */
# pragma ivdep
# pragma swp
# pragma unroll
# pragma prefetch
# if 0
# pragma simd noassert
# endif
#endif /* VDM auto patch */
for (row = 0; row < num_rows; row++) {
/* Initialize output values to 0 so can process components separately */
jzero_far((void FAR *) output_buf[row],
(size_t) (width * SIZEOF(JSAMPLE)));
row_index = cquantize->row_index;
#if defined(__INTEL_COMPILER) && 1 /* VDM auto patch */
# pragma ivdep
# pragma swp
# pragma unroll
# pragma prefetch
# if 0
# pragma simd noassert
# endif
#endif /* VDM auto patch */
for (ci = 0; ci < nc; ci++) {
input_ptr = input_buf[row] + ci;
output_ptr = output_buf[row];
colorindex_ci = cquantize->colorindex[ci];
dither = cquantize->odither[ci][row_index];
col_index = 0;
#if defined(__INTEL_COMPILER) && 1 /* VDM auto patch */
# pragma ivdep
# pragma swp
# pragma unroll
# pragma prefetch
# if 0
# pragma simd noassert
# endif
#endif /* VDM auto patch */
for (col = width; col > 0; col--) {
/* Form pixel value + dither, range-limit to 0..MAXJSAMPLE,
* select output value, accumulate into output code for this pixel.
* Range-limiting need not be done explicitly, as we have extended
* the colorindex table to produce the right answers for out-of-range
* inputs. The maximum dither is +- MAXJSAMPLE; this sets the
* required amount of padding.
*/
*output_ptr += colorindex_ci[GETJSAMPLE(*input_ptr)+dither[col_index]];
input_ptr += nc;
output_ptr++;
col_index = (col_index + 1) & ODITHER_MASK;
}
}
/* Advance row index for next row */
row_index = (row_index + 1) & ODITHER_MASK;
cquantize->row_index = row_index;
}
}
jpeg_fdct_float (FAST_FLOAT * data, JSAMPARRAY sample_data, JDIMENSION start_col)
{
FAST_FLOAT tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
FAST_FLOAT tmp10, tmp11, tmp12, tmp13;
FAST_FLOAT z1, z2, z3, z4, z5, z11, z13;
FAST_FLOAT *dataptr;
JSAMPROW elemptr;
int ctr;
/* Pass 1: process rows. */
dataptr = data;
for (ctr = 0; ctr < DCTSIZE; ctr++) {
elemptr = sample_data[ctr] + start_col;
/* Load data into workspace */
tmp0 = (FAST_FLOAT) (GETJSAMPLE(elemptr[0]) + GETJSAMPLE(elemptr[7]));
tmp7 = (FAST_FLOAT) (GETJSAMPLE(elemptr[0]) - GETJSAMPLE(elemptr[7]));
tmp1 = (FAST_FLOAT) (GETJSAMPLE(elemptr[1]) + GETJSAMPLE(elemptr[6]));
tmp6 = (FAST_FLOAT) (GETJSAMPLE(elemptr[1]) - GETJSAMPLE(elemptr[6]));
tmp2 = (FAST_FLOAT) (GETJSAMPLE(elemptr[2]) + GETJSAMPLE(elemptr[5]));
tmp5 = (FAST_FLOAT) (GETJSAMPLE(elemptr[2]) - GETJSAMPLE(elemptr[5]));
tmp3 = (FAST_FLOAT) (GETJSAMPLE(elemptr[3]) + GETJSAMPLE(elemptr[4]));
tmp4 = (FAST_FLOAT) (GETJSAMPLE(elemptr[3]) - GETJSAMPLE(elemptr[4]));
/* Even part */
tmp10 = tmp0 + tmp3; /* phase 2 */
tmp13 = tmp0 - tmp3;
tmp11 = tmp1 + tmp2;
tmp12 = tmp1 - tmp2;
/* Apply unsigned->signed conversion. */
dataptr[0] = tmp10 + tmp11 - 8 * CENTERJSAMPLE; /* phase 3 */
dataptr[4] = tmp10 - tmp11;
z1 = (tmp12 + tmp13) * ((FAST_FLOAT) 0.707106781); /* c4 */
dataptr[2] = tmp13 + z1; /* phase 5 */
dataptr[6] = tmp13 - z1;
/* Odd part */
tmp10 = tmp4 + tmp5; /* phase 2 */
tmp11 = tmp5 + tmp6;
tmp12 = tmp6 + tmp7;
/* The rotator is modified from fig 4-8 to avoid extra negations. */
z5 = (tmp10 - tmp12) * ((FAST_FLOAT) 0.382683433); /* c6 */
z2 = ((FAST_FLOAT) 0.541196100) * tmp10 + z5; /* c2-c6 */
z4 = ((FAST_FLOAT) 1.306562965) * tmp12 + z5; /* c2+c6 */
z3 = tmp11 * ((FAST_FLOAT) 0.707106781); /* c4 */
z11 = tmp7 + z3; /* phase 5 */
z13 = tmp7 - z3;
dataptr[5] = z13 + z2; /* phase 6 */
dataptr[3] = z13 - z2;
dataptr[1] = z11 + z4;
dataptr[7] = z11 - z4;
dataptr += DCTSIZE; /* advance pointer to next row */
}
/* Pass 2: process columns. */
dataptr = data;
for (ctr = DCTSIZE-1; ctr >= 0; ctr--) {
tmp0 = dataptr[DCTSIZE*0] + dataptr[DCTSIZE*7];
tmp7 = dataptr[DCTSIZE*0] - dataptr[DCTSIZE*7];
tmp1 = dataptr[DCTSIZE*1] + dataptr[DCTSIZE*6];
tmp6 = dataptr[DCTSIZE*1] - dataptr[DCTSIZE*6];
tmp2 = dataptr[DCTSIZE*2] + dataptr[DCTSIZE*5];
tmp5 = dataptr[DCTSIZE*2] - dataptr[DCTSIZE*5];
tmp3 = dataptr[DCTSIZE*3] + dataptr[DCTSIZE*4];
tmp4 = dataptr[DCTSIZE*3] - dataptr[DCTSIZE*4];
/* Even part */
tmp10 = tmp0 + tmp3; /* phase 2 */
tmp13 = tmp0 - tmp3;
tmp11 = tmp1 + tmp2;
tmp12 = tmp1 - tmp2;
dataptr[DCTSIZE*0] = tmp10 + tmp11; /* phase 3 */
dataptr[DCTSIZE*4] = tmp10 - tmp11;
z1 = (tmp12 + tmp13) * ((FAST_FLOAT) 0.707106781); /* c4 */
dataptr[DCTSIZE*2] = tmp13 + z1; /* phase 5 */
dataptr[DCTSIZE*6] = tmp13 - z1;
/* Odd part */
tmp10 = tmp4 + tmp5; /* phase 2 */
tmp11 = tmp5 + tmp6;
tmp12 = tmp6 + tmp7;
/* The rotator is modified from fig 4-8 to avoid extra negations. */
z5 = (tmp10 - tmp12) * ((FAST_FLOAT) 0.382683433); /* c6 */
z2 = ((FAST_FLOAT) 0.541196100) * tmp10 + z5; /* c2-c6 */
z4 = ((FAST_FLOAT) 1.306562965) * tmp12 + z5; /* c2+c6 */
z3 = tmp11 * ((FAST_FLOAT) 0.707106781); /* c4 */
z11 = tmp7 + z3; /* phase 5 */
z13 = tmp7 - z3;
dataptr[DCTSIZE*5] = z13 + z2; /* phase 6 */
dataptr[DCTSIZE*3] = z13 - z2;
dataptr[DCTSIZE*1] = z11 + z4;
dataptr[DCTSIZE*7] = z11 - z4;
dataptr++; /* advance pointer to next column */
}
}
METHODDEF void
h2v2_smooth_downsample(j_compress_ptr cinfo, jpeg_component_info *compptr,
JSAMPARRAY input_data, JSAMPARRAY output_data)
{
int inrow, outrow;
JDIMENSION colctr;
JDIMENSION output_cols = compptr->width_in_blocks * DCTSIZE;
register JSAMPROW inptr0, inptr1, above_ptr, below_ptr, outptr;
INT32 membersum, neighsum, memberscale, neighscale;
/* Expand input data enough to let all the output samples be generated
* by the standard loop. Special-casing padded output would be more
* efficient.
*/
expand_right_edge(input_data - 1, cinfo->max_v_samp_factor + 2,
cinfo->image_width, output_cols * 2);
/* We don't bother to form the individual "smoothed" input pixel values;
* we can directly compute the output which is the average of the four
* smoothed values. Each of the four member pixels contributes a fraction
* (1-8*SF) to its own smoothed image and a fraction SF to each of the three
* other smoothed pixels, therefore a total fraction (1-5*SF)/4 to the final
* output. The four corner-adjacent neighbor pixels contribute a fraction
* SF to just one smoothed pixel, or SF/4 to the final output; while the
* eight edge-adjacent neighbors contribute SF to each of two smoothed
* pixels, or SF/2 overall. In order to use integer arithmetic, these
* factors are scaled by 2^16 = 65536.
* Also recall that SF = smoothing_factor / 1024.
*/
memberscale = 16384 - cinfo->smoothing_factor * 80; /* scaled (1-5*SF)/4 */
neighscale = cinfo->smoothing_factor * 16; /* scaled SF/4 */
inrow = 0;
for(outrow = 0; outrow < compptr->v_samp_factor; outrow++)
{
outptr = output_data[outrow];
inptr0 = input_data[inrow];
inptr1 = input_data[inrow + 1];
above_ptr = input_data[inrow - 1];
below_ptr = input_data[inrow + 2];
/* Special case for first column: pretend column -1 is same as column 0 */
membersum = GETJSAMPLE(*inptr0) + GETJSAMPLE(inptr0[1]) +
GETJSAMPLE(*inptr1) + GETJSAMPLE(inptr1[1]);
neighsum = GETJSAMPLE(*above_ptr) + GETJSAMPLE(above_ptr[1]) +
GETJSAMPLE(*below_ptr) + GETJSAMPLE(below_ptr[1]) +
GETJSAMPLE(*inptr0) + GETJSAMPLE(inptr0[2]) +
GETJSAMPLE(*inptr1) + GETJSAMPLE(inptr1[2]);
neighsum += neighsum;
neighsum += GETJSAMPLE(*above_ptr) + GETJSAMPLE(above_ptr[2]) +
GETJSAMPLE(*below_ptr) + GETJSAMPLE(below_ptr[2]);
membersum = membersum * memberscale + neighsum * neighscale;
*outptr++ = (JSAMPLE)((membersum + 32768) >> 16);
inptr0 += 2;
inptr1 += 2;
above_ptr += 2;
below_ptr += 2;
for(colctr = output_cols - 2; colctr > 0; colctr--)
{
/* sum of pixels directly mapped to this output element */
membersum = GETJSAMPLE(*inptr0) + GETJSAMPLE(inptr0[1]) +
GETJSAMPLE(*inptr1) + GETJSAMPLE(inptr1[1]);
/* sum of edge-neighbor pixels */
neighsum = GETJSAMPLE(*above_ptr) + GETJSAMPLE(above_ptr[1]) +
GETJSAMPLE(*below_ptr) + GETJSAMPLE(below_ptr[1]) +
GETJSAMPLE(inptr0[-1]) + GETJSAMPLE(inptr0[2]) +
GETJSAMPLE(inptr1[-1]) + GETJSAMPLE(inptr1[2]);
/* The edge-neighbors count twice as much as corner-neighbors */
neighsum += neighsum;
/* Add in the corner-neighbors */
neighsum += GETJSAMPLE(above_ptr[-1]) + GETJSAMPLE(above_ptr[2]) +
GETJSAMPLE(below_ptr[-1]) + GETJSAMPLE(below_ptr[2]);
/* form final output scaled up by 2^16 */
membersum = membersum * memberscale + neighsum * neighscale;
/* round, descale and output it */
*outptr++ = (JSAMPLE)((membersum + 32768) >> 16);
inptr0 += 2;
inptr1 += 2;
above_ptr += 2;
below_ptr += 2;
}
/* Special case for last column */
membersum = GETJSAMPLE(*inptr0) + GETJSAMPLE(inptr0[1]) +
GETJSAMPLE(*inptr1) + GETJSAMPLE(inptr1[1]);
neighsum = GETJSAMPLE(*above_ptr) + GETJSAMPLE(above_ptr[1]) +
GETJSAMPLE(*below_ptr) + GETJSAMPLE(below_ptr[1]) +
GETJSAMPLE(inptr0[-1]) + GETJSAMPLE(inptr0[1]) +
GETJSAMPLE(inptr1[-1]) + GETJSAMPLE(inptr1[1]);
neighsum += neighsum;
neighsum += GETJSAMPLE(above_ptr[-1]) + GETJSAMPLE(above_ptr[1]) +
GETJSAMPLE(below_ptr[-1]) + GETJSAMPLE(below_ptr[1]);
membersum = membersum * memberscale + neighsum * neighscale;
*outptr = (JSAMPLE)((membersum + 32768) >> 16);
inrow += 2;
}
}
forward_DCT (j_compress_ptr cinfo, jpeg_component_info * compptr,
JSAMPARRAY sample_data, JBLOCKROW coef_blocks,
JDIMENSION start_row, JDIMENSION start_col,
JDIMENSION num_blocks)
/* This version is used for integer DCT implementations. */
{
/* This routine is heavily used, so it's worth coding it tightly. */
j_lossy_c_ptr lossyc = (j_lossy_c_ptr) cinfo->codec;
fdct_ptr fdct = (fdct_ptr) lossyc->fdct_private;
forward_DCT_method_ptr do_dct = fdct->do_dct;
DCTELEM * divisors = fdct->divisors[compptr->quant_tbl_no];
DCTELEM workspace[DCTSIZE2]; /* work area for FDCT subroutine */
JDIMENSION bi;
sample_data += start_row; /* fold in the vertical offset once */
for (bi = 0; bi < num_blocks; bi++, start_col += DCTSIZE) {
/* Load data into workspace, applying unsigned->signed conversion */
{ register DCTELEM *workspaceptr;
register JSAMPROW elemptr;
register int elemr;
workspaceptr = workspace;
for (elemr = 0; elemr < DCTSIZE; elemr++) {
elemptr = sample_data[elemr] + start_col;
#if DCTSIZE == 8 /* unroll the inner loop */
*workspaceptr++ = GETJSAMPLE(*elemptr++) - CENTERJSAMPLE;
*workspaceptr++ = GETJSAMPLE(*elemptr++) - CENTERJSAMPLE;
*workspaceptr++ = GETJSAMPLE(*elemptr++) - CENTERJSAMPLE;
*workspaceptr++ = GETJSAMPLE(*elemptr++) - CENTERJSAMPLE;
*workspaceptr++ = GETJSAMPLE(*elemptr++) - CENTERJSAMPLE;
*workspaceptr++ = GETJSAMPLE(*elemptr++) - CENTERJSAMPLE;
*workspaceptr++ = GETJSAMPLE(*elemptr++) - CENTERJSAMPLE;
*workspaceptr++ = GETJSAMPLE(*elemptr++) - CENTERJSAMPLE;
#else
{ register int elemc;
for (elemc = DCTSIZE; elemc > 0; elemc--) {
*workspaceptr++ = GETJSAMPLE(*elemptr++) - CENTERJSAMPLE;
}
}
#endif
}
}
/* Perform the DCT */
(*do_dct) (workspace);
/* Quantize/descale the coefficients, and store into coef_blocks[] */
{ register DCTELEM temp, qval;
register int i;
register JCOEFPTR output_ptr = coef_blocks[bi];
for (i = 0; i < DCTSIZE2; i++) {
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.
*/
#ifdef FAST_DIVIDE
#define DIVIDE_BY(a,b) a /= b
#else
#define DIVIDE_BY(a,b) if (a >= b) a /= b; else a = 0
#endif
if (temp < 0) {
temp = -temp;
temp += qval>>1; /* for rounding */
DIVIDE_BY(temp, qval);
temp = -temp;
} else {
temp += qval>>1; /* for rounding */
DIVIDE_BY(temp, qval);
}
output_ptr[i] = (JCOEF) temp;
}
}
static int bmJpegRememberScanline( JSAMPARRAY pixel_data,
struct jpeg_decompress_struct * cinfo )
{
BmJpegInputSource * bjis= (BmJpegInputSource *)cinfo->src;
unsigned char * to;
register JSAMPROW ptr0;
long col;
switch( cinfo->out_color_space )
{
case JCS_GRAYSCALE:
switch( bjis->bjisBd.bdBitsPerSample )
{
case 8:
to= bjis->bjisBitmapBuffer+
bjis->bjisBd.bdBytesPerRow*
(bjis->bjisRowsReceived++);
ptr0 = pixel_data[0];
for ( col = 0; col < cinfo->image_width; col++ )
{ *(to++)= GETJSAMPLE(*(ptr0++)); }
break;
default:
LDEB(bjis->bjisBd.bdBitsPerSample);
return -1;
}
break;
case JCS_RGB:
switch( bjis->bjisBd.bdBitsPerSample )
{
case 8:
to= bjis->bjisBitmapBuffer+
bjis->bjisBd.bdBytesPerRow*
(bjis->bjisRowsReceived++);
ptr0 = pixel_data[0];
for ( col = 0; col < cinfo->image_width; col++ )
{
*(to++)= GETJSAMPLE(*(ptr0++)); /* red */
*(to++)= GETJSAMPLE(*(ptr0++)); /* green */
*(to++)= GETJSAMPLE(*(ptr0++)); /* blue */
}
break;
default:
LDEB(bjis->bjisBd.bdBitsPerSample);
return -1;
}
break;
case JCS_UNKNOWN:
case JCS_YCbCr:
case JCS_YCCK:
case JCS_CMYK:
LDEB(cinfo->out_color_space);
return -1;
break;
default:
LDEB(cinfo->out_color_space);
return -1;
break;
}
return 0;
}
