// Downsample pixel values of a single component.
// One row group is processed per call.
// This version handles arbitrary integral sampling ratios, without smoothing.
// Note that this version is not actually used for customary sampling ratios.
static void int_downsample(jpeg_compress cinfo, jpeg_component_info compptr, byte[][] input_data, uint in_row_index, byte[][] output_data, uint out_row_index)
{
uint output_cols = compptr.width_in_blocks * cinfo.DCT_size;
int h_expand = cinfo.max_h_samp_factor / compptr.h_samp_factor;
int v_expand = cinfo.max_v_samp_factor / compptr.v_samp_factor;
int numpix = h_expand * v_expand;
int numpix2 = numpix / 2;
// 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, in_row_index, cinfo.max_v_samp_factor, cinfo.image_width, (uint)(output_cols * h_expand));
int inrow = (int)in_row_index;
for (int outrow = 0; outrow < compptr.v_samp_factor; outrow++)
{
byte[] outptr = output_data[out_row_index + outrow];
for (uint outcol = 0, outcol_h = 0; outcol < output_cols; outcol++, outcol_h += (uint)h_expand)
{
byte[] inptr;
int outvalue = 0;
for (int v = 0; v < v_expand; v++)
{
inptr = input_data[inrow + v];
for (int h = 0; h < h_expand; h++)
{
outvalue += (int)inptr[outcol_h + h];
}
}
outptr[outcol] = (byte)((outvalue + numpix2) / numpix);
}
inrow += v_expand;
}
}
// Downsample pixel values of a single component.
// This version handles the standard case of 2:1 horizontal and 2:1 vertical,
// without smoothing.
static void h2v2_downsample(jpeg_compress cinfo, jpeg_component_info compptr, byte[][] input_data, uint in_row_index, byte[][] output_data, uint out_row_index)
{
uint output_cols = compptr.width_in_blocks * cinfo.DCT_size;
// 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, in_row_index, cinfo.max_v_samp_factor, cinfo.image_width, output_cols * 2);
int inrow = (int)in_row_index;
for (int outrow = 0; outrow < compptr.v_samp_factor; outrow++)
{
byte[] outptr = output_data[out_row_index + outrow];
byte[] inptr0 = input_data[inrow];
byte[] inptr1 = input_data[inrow + 1];
int bias = 1; // bias = 1,2,1,2,... for successive samples
for (uint outcol = 0, ind = 0; outcol < output_cols; outcol++, ind += 2)
{
outptr[outcol] = (byte)((inptr0[ind] + inptr0[ind + 1] + inptr1[ind] + inptr1[ind + 1] + bias) >> 2);
bias ^= 3; // 1=>2, 2=>1
}
inrow += 2;
}
}
// These are the routines invoked by sep_upsample to upsample pixel values
// of a single component. One row group is processed per call.
// For full-size components, we just make color_buf[ci] point at the
// input buffer, and thus avoid copying any data. Note that this is
// safe only because sep_upsample doesn't declare the input row group
// "consumed" until we are done color converting and emitting it.
static void fullsize_upsample(jpeg_decompress cinfo, jpeg_component_info compptr, byte[][] input_data, int input_data_offset, byte[][][] output_data_ptr, int output_data_offset)
{
for (int i = 0; i < cinfo.max_v_samp_factor; i++)
{
output_data_ptr[output_data_offset][i] = input_data[input_data_offset++];
}
}
// Fancy processing for the common case of 2:1 horizontal and 1:1 vertical.
//
// The upsampling algorithm is linear interpolation between pixel centers,
// also known as a "triangle filter". This is a good compromise between
// speed and visual quality. The centers of the output pixels are 1/4 and 3/4
// of the way between input pixel centers.
//
// A note about the "bias" calculations: when rounding fractional values to
// integer, we do not want to always round 0.5 up to the next integer.
// If we did that, we'd introduce a noticeable bias towards larger values.
// Instead, this code is arranged so that 0.5 will be rounded up or down at
// alternate pixel locations (a simple ordered dither pattern).
#if !USE_UNSAFE_STUFF
static void h2v1_fancy_upsample(jpeg_decompress cinfo, jpeg_component_info compptr, byte[][] input_data, int input_data_offset, byte[][][] output_data_ptr, int output_data_offset)
{
byte[][] output_data = output_data_ptr[output_data_offset];
for (int inrow = 0; inrow < cinfo.max_v_samp_factor; inrow++)
{
byte[] inptr = input_data[input_data_offset++];
uint inptr_ind = 0;
byte[] outptr = output_data[inrow];
uint outptr_ind = 0;
// Special case for first column
int invalue = inptr[inptr_ind++];
outptr[outptr_ind++] = (byte)invalue;
outptr[outptr_ind++] = (byte)((invalue * 3 + inptr[inptr_ind] + 2) >> 2);
for (uint colctr = compptr.downsampled_width - 2; colctr > 0; colctr--)
{
// General case: 3/4 * nearer pixel + 1/4 * further pixel
invalue = inptr[inptr_ind++] * 3;
outptr[outptr_ind++] = (byte)((invalue + inptr[inptr_ind - 2] + 1) >> 2);
outptr[outptr_ind++] = (byte)((invalue + inptr[inptr_ind] + 2) >> 2);
}
// Special case for last column
invalue = inptr[inptr_ind];
outptr[outptr_ind++] = (byte)((invalue * 3 + inptr[inptr_ind - 1] + 1) >> 2);
outptr[outptr_ind] = (byte)invalue;
}
}
// Fast processing for the common case of 2:1 horizontal and 2:1 vertical.
// It's still a box filter.
static void h2v2_upsample(jpeg_decompress cinfo, jpeg_component_info compptr, byte[][] input_data, int input_data_offset, byte[][][] output_data_ptr, int output_data_offset)
{
byte[][] output_data = output_data_ptr[output_data_offset];
int outrow = 0;
uint outend = cinfo.output_width;
while (outrow < cinfo.max_v_samp_factor)
{
byte[] inptr = input_data[input_data_offset];
uint inptr_ind = 0;
byte[] outptr = output_data[outrow];
uint outptr_ind = 0;
while (outptr_ind < outend)
{
byte invalue = inptr[inptr_ind++];
outptr[outptr_ind++] = invalue;
outptr[outptr_ind++] = invalue;
}
Array.Copy(output_data[outrow], output_data[outrow + 1], cinfo.output_width);
input_data_offset++;
outrow += 2;
}
}
static void h2v2_fancy_upsample(jpeg_decompress cinfo, jpeg_component_info compptr, byte[][] input_data, int input_data_offset, byte[][][] output_data_ptr, int output_data_offset)
{
unsafe
{
if (!compptr.notFirst)
{
input_data[0].CopyTo(input_data[input_data.Length - 1], 0);
compptr.notFirst = true;
}
byte[][] output_data = output_data_ptr[output_data_offset];
int inrow = 0, outrow = 0;
while (outrow < cinfo.max_v_samp_factor)
{
// inptr0 points to nearest input row
fixed(byte *inptr0_ = input_data[input_data_offset + inrow])
{
// inptr1 points to next nearest
int nextnearestrow = input_data_offset == 0?input_data.Length - 1:input_data_offset + inrow - 1; // next nearest is row above
for (int v = 0; v < 2; v++)
{
fixed(byte *inptr1_ = input_data[nextnearestrow], outptr_ = output_data[outrow++])
{
byte *inptr0 = inptr0_, inptr1 = inptr1_, outptr = outptr_;
// Special case for first column
int thiscolsum = *(inptr0++) * 3 + *(inptr1++);
int nextcolsum = *(inptr0++) * 3 + *(inptr1++);
*(outptr++) = (byte)((thiscolsum * 4 + 8) >> 4);
*(outptr++) = (byte)((thiscolsum * 3 + nextcolsum + 7) >> 4);
int lastcolsum = thiscolsum;
thiscolsum = nextcolsum;
for (uint 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 = *(inptr0++) * 3 + *(inptr1++);
*(outptr++) = (byte)((thiscolsum * 3 + lastcolsum + 8) >> 4);
*(outptr++) = (byte)((thiscolsum * 3 + nextcolsum + 7) >> 4);
lastcolsum = thiscolsum; thiscolsum = nextcolsum;
}
// Special case for last column
*(outptr++) = (byte)((thiscolsum * 3 + lastcolsum + 8) >> 4);
*(outptr++) = (byte)((thiscolsum * 4 + 7) >> 4);
}
nextnearestrow = (input_data_offset + inrow + 1) % input_data.Length; // next nearest is row below
}
}
inrow++;
}
}
}
static void h2v1_fancy_upsample(jpeg_decompress cinfo, jpeg_component_info compptr, byte[][] input_data, int input_data_offset, byte[][][] output_data_ptr, int output_data_offset)
{
unsafe
{
byte[][] output_data = output_data_ptr[output_data_offset];
for (int inrow = 0; inrow < cinfo.max_v_samp_factor; inrow++)
{
fixed(byte *inptr_ = input_data[input_data_offset++], outptr_ = output_data[inrow])
{
byte *inptr = inptr_;
byte *outptr = outptr_;
// Special case for first column
int invalue = *(inptr++);
*(outptr++) = (byte)invalue;
*(outptr++) = (byte)((invalue * 3 + *inptr + 2) >> 2);
for (uint colctr = compptr.downsampled_width - 2; colctr > 0; colctr--)
{
// General case: 3/4 * nearer pixel + 1/4 * further pixel
invalue = *(inptr++) * 3;
*(outptr++) = (byte)((invalue + inptr[-2] + 1) >> 2);
*(outptr++) = (byte)((invalue + *inptr + 2) >> 2);
}
// Special case for last column
invalue = *inptr;
*(outptr++) = (byte)((invalue * 3 + inptr[-1] + 1) >> 2);
*outptr = (byte)invalue;
}
}
}
}
// Allocate downsampled-data buffers needed for downsampled I/O.
// We use values computed in jpeg_start_compress or jpeg_start_decompress.
// We use libjpeg's allocator so that buffers will be released automatically
// when done with strip/tile.
// This is also a handy place to compute samplesperclump, bytesperline.
static bool alloc_downsampled_buffers(TIFF tif, jpeg_component_info[] comp_info, int num_components)
{
JPEGState sp=tif.tif_data as JPEGState;
byte[][] buf;
int samples_per_clump=0;
for(int ci=0; ci<num_components; ci++)
{
jpeg_component_info compptr=comp_info[ci];
samples_per_clump+=compptr.h_samp_factor*compptr.v_samp_factor;
try
{
buf=TIFFjpeg_alloc_sarray(sp, compptr.width_in_blocks*libjpeg.DCTSIZE, (uint)compptr.v_samp_factor*libjpeg.DCTSIZE);
}
catch
{
return false;
}
sp.ds_buffer[ci]=buf;
}
sp.samplesperclump=samples_per_clump;
return true;
}
// Create the wrapped-around downsampling input buffer needed for context mode.
static void create_context_buffer(jpeg_compress cinfo)
{
my_prep_controller prep = (my_prep_controller)cinfo.prep;
int rgroup_height = cinfo.max_v_samp_factor;
for (int ci = 0; ci < cinfo.num_components; ci++)
{
jpeg_component_info compptr = cinfo.comp_info[ci];
// Grab enough space for fake row pointers;
// we need five row groups' worth of pointers for each component.
byte[][] fake_buffer = new byte[5 * rgroup_height][];
// Allocate the actual buffer space (3 row groups) for this component.
// We make the buffer wide enough to allow the downsampler to edge-expand
// horizontally within the buffer, if it so chooses.
byte[][] true_buffer = alloc_sarray(cinfo,
(uint)(((int)compptr.width_in_blocks * cinfo.DCT_size * cinfo.max_h_samp_factor) / compptr.h_samp_factor),
(uint)(3 * rgroup_height));
// Copy true buffer row pointers into the middle of the fake row array
Array.Copy(true_buffer, 0, fake_buffer, rgroup_height, 3 * rgroup_height);
// Fill in the above and below wraparound pointers
for (int i = 0; i < rgroup_height; i++)
{
fake_buffer[i] = true_buffer[2 * rgroup_height + i];
fake_buffer[4 * rgroup_height + i] = true_buffer[i];
}
prep.color_buf[ci] = fake_buffer;
}
}
// Downsample pixel values of a single component.
// This version handles the special case of a full-size component,
// without smoothing.
static void fullsize_downsample(jpeg_compress cinfo, jpeg_component_info compptr, byte[][] input_data, uint in_row_index, byte[][] output_data, uint out_row_index)
{
// Copy the data
jcopy_sample_rows(input_data, (int)in_row_index, output_data, (int)out_row_index, cinfo.max_v_samp_factor, cinfo.image_width);
// Edge-expand
expand_right_edge(output_data, out_row_index, cinfo.max_v_samp_factor, cinfo.image_width, compptr.width_in_blocks * cinfo.DCT_size);
}
// Fast processing for the common case of 2:1 horizontal and 2:1 vertical.
// It's still a box filter.
#if !USE_UNSAFE_STUFF
static void h2v2_fancy_upsample(jpeg_decompress cinfo, jpeg_component_info compptr, byte[][] input_data, int input_data_offset, byte[][][] output_data_ptr, int output_data_offset)
{
if (!compptr.notFirst)
{
input_data[0].CopyTo(input_data[input_data.Length - 1], 0);
compptr.notFirst = true;
}
byte[][] output_data = output_data_ptr[output_data_offset];
int inrow = 0, outrow = 0;
while (outrow < cinfo.max_v_samp_factor)
{
for (int v = 0; v < 2; v++)
{
// inptr0 points to nearest input row, inptr1 points to next nearest
byte[] inptr0 = input_data[input_data_offset + inrow];
byte[] inptr1 = null;
if (v == 0)
{
inptr1 = input_data[input_data_offset == 0?input_data.Length - 1:input_data_offset + inrow - 1]; // next nearest is row above
}
else
{
inptr1 = input_data[(input_data_offset + inrow + 1) % input_data.Length]; // next nearest is row below
}
uint intptr_ind = 0;
byte[] outptr = output_data[outrow++];
uint outptr_ind = 0;
// Special case for first column
int thiscolsum = inptr0[intptr_ind] * 3 + inptr1[intptr_ind]; intptr_ind++;
int nextcolsum = inptr0[intptr_ind] * 3 + inptr1[intptr_ind]; intptr_ind++;
outptr[outptr_ind++] = (byte)((thiscolsum * 4 + 8) >> 4);
outptr[outptr_ind++] = (byte)((thiscolsum * 3 + nextcolsum + 7) >> 4);
int lastcolsum = thiscolsum;
thiscolsum = nextcolsum;
for (uint 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 = inptr0[intptr_ind] * 3 + inptr1[intptr_ind]; intptr_ind++;
outptr[outptr_ind++] = (byte)((thiscolsum * 3 + lastcolsum + 8) >> 4);
outptr[outptr_ind++] = (byte)((thiscolsum * 3 + nextcolsum + 7) >> 4);
lastcolsum = thiscolsum; thiscolsum = nextcolsum;
}
// Special case for last column
outptr[outptr_ind++] = (byte)((thiscolsum * 3 + lastcolsum + 8) >> 4);
outptr[outptr_ind++] = (byte)((thiscolsum * 4 + 7) >> 4);
}
inrow++;
}
}
// Downsample pixel values of a single component.
// This version handles the special case of a full-size component,
// with smoothing. One row of context is required.
static void fullsize_smooth_downsample(jpeg_compress cinfo, jpeg_component_info compptr, byte[][] input_data, uint in_row_index, byte[][] output_data, uint out_row_index)
{
uint output_cols = compptr.width_in_blocks * cinfo.DCT_size;
// 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, in_row_index - 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.
int memberscale = 65536 - cinfo.smoothing_factor * 512; // scaled 1-8*SF
int neighscale = cinfo.smoothing_factor * 64; // scaled SF
int membersum, neighsum;
int colsum, lastcolsum, nextcolsum;
for (int outrow = 0; outrow < compptr.v_samp_factor; outrow++)
{
byte[] outptr = output_data[out_row_index + outrow];
byte[] inptr = input_data[in_row_index + outrow];
byte[] above_ptr = input_data[in_row_index + outrow - 1];
byte[] below_ptr = input_data[in_row_index + outrow + 1];
int iind = 0, aind = 0, bind = 0, oind = 0;
// Special case for first column
colsum = above_ptr[aind++] + below_ptr[bind++] + inptr[iind];
membersum = inptr[iind++];
nextcolsum = above_ptr[aind] + below_ptr[bind] + inptr[iind];
neighsum = colsum + (colsum - membersum) + nextcolsum;
membersum = membersum * memberscale + neighsum * neighscale;
outptr[oind++] = (byte)((membersum + 32768) >> 16);
lastcolsum = colsum; colsum = nextcolsum;
for (uint colctr = output_cols - 2; colctr > 0; colctr--)
{
membersum = inptr[iind++];
nextcolsum = above_ptr[aind++] + below_ptr[bind++] + inptr[iind];
neighsum = lastcolsum + (colsum - membersum) + nextcolsum;
membersum = membersum * memberscale + neighsum * neighscale;
outptr[oind++] = (byte)((membersum + 32768) >> 16);
lastcolsum = colsum; colsum = nextcolsum;
}
// Special case for last column
membersum = inptr[iind];
neighsum = lastcolsum + (colsum - membersum) + colsum;
membersum = membersum * memberscale + neighsum * neighscale;
outptr[oind] = (byte)((membersum + 32768) >> 16);
}
}
static void SET_COMP(jpeg_compress cinfo, int index, int id, int hsamp, int vsamp, int quant, int dctbl, int actbl)
{
jpeg_component_info compptr = cinfo.comp_info[index];
compptr.component_id = id;
compptr.h_samp_factor = hsamp;
compptr.v_samp_factor = vsamp;
compptr.quant_tbl_no = quant;
compptr.dc_tbl_no = dctbl;
compptr.ac_tbl_no = actbl;
}
// Initialize for a Huffman-compressed scan.
static void start_pass_lhuff_decoder(jpeg_decompress cinfo)
{
jpeg_lossless_d_codec losslsd = (jpeg_lossless_d_codec)cinfo.coef;
lhuff_entropy_decoder entropy = (lhuff_entropy_decoder)losslsd.entropy_private;
for (int ci = 0; ci < cinfo.comps_in_scan; ci++)
{
jpeg_component_info compptr = cinfo.cur_comp_info[ci];
int dctbl = compptr.dc_tbl_no;
// Make sure requested tables are present
if (dctbl < 0 || dctbl >= NUM_HUFF_TBLS ||
cinfo.dc_huff_tbl_ptrs[dctbl] == null)
{
ERREXIT1(cinfo, J_MESSAGE_CODE.JERR_NO_HUFF_TABLE, dctbl);
}
// Compute derived values for Huffman tables
// We may do this more than once for a table, but it's not expensive
jpeg_make_d_derived_tbl(cinfo, true, dctbl, ref entropy.derived_tbls[dctbl]);
}
// Precalculate decoding info for each sample in an MCU of this scan
int ptrn = 0;
for (int sampn = 0; sampn < cinfo.blocks_in_MCU;)
{
jpeg_component_info compptr = cinfo.cur_comp_info[cinfo.MCU_membership[sampn]];
int ci = compptr.component_index;
for (int yoffset = 0; yoffset < compptr.MCU_height; yoffset++, ptrn++)
{
// Precalculate the setup info for each output pointer
entropy.output_ptr_info[ptrn].ci = ci;
entropy.output_ptr_info[ptrn].yoffset = yoffset;
entropy.output_ptr_info[ptrn].MCU_width = compptr.MCU_width;
for (int xoffset = 0; xoffset < compptr.MCU_width; xoffset++, sampn++)
{
// Precalculate the output pointer index for each sample
entropy.output_ptr_index[sampn] = ptrn;
// Precalculate which table to use for each sample
entropy.cur_tbls[sampn] = entropy.derived_tbls[compptr.dc_tbl_no];
}
}
}
entropy.num_output_ptrs = ptrn;
// Initialize bitread state variables
entropy.bitstate.bits_left = 0;
entropy.bitstate.get_buffer = 0; // unnecessary, but keeps Purify quiet
entropy.insufficient_data = false;
}
// Control routine to do upsampling (and color conversion).
//
// In this version we upsample each component independently.
// We upsample one row group into the conversion buffer, then apply
// color conversion a row at a time.
static void sep_upsample(jpeg_decompress cinfo, byte[][][] input_buf, ref uint in_row_group_ctr, uint in_row_groups_avail, byte[][] output_buf, uint output_buf_offset, ref uint out_row_ctr, uint out_rows_avail)
{
my_upsampler upsample = (my_upsampler)cinfo.upsample;
// Fill the conversion buffer, if it's empty
if (upsample.next_row_out >= cinfo.max_v_samp_factor)
{
for (int ci = 0; ci < cinfo.num_components; ci++)
{
jpeg_component_info compptr = cinfo.comp_info[ci];
// Invoke per-component upsample method. Notice we pass a POINTER
// to color_buf[ci], so that fullsize_upsample can change it.
upsample.methods[ci](cinfo, compptr, input_buf[ci], (int)(in_row_group_ctr * upsample.rowgroup_height[ci]), upsample.color_buf, ci);
}
upsample.next_row_out = 0;
}
// Color-convert and emit rows
// How many we have in the buffer:
uint num_rows = (uint)(cinfo.max_v_samp_factor - upsample.next_row_out);
// Not more than the distance to the end of the image. Need this test
// in case the image height is not a multiple of max_v_samp_factor:
if (num_rows > upsample.rows_to_go)
{
num_rows = upsample.rows_to_go;
}
// And not more than what the client can accept:
out_rows_avail -= out_row_ctr;
if (num_rows > out_rows_avail)
{
num_rows = out_rows_avail;
}
cinfo.cconvert.color_convert(cinfo, upsample.color_buf, (uint)upsample.next_row_out, output_buf, output_buf_offset + out_row_ctr, (int)num_rows);
// Adjust counts
out_row_ctr += num_rows;
upsample.rows_to_go -= num_rows;
upsample.next_row_out += (int)num_rows;
// When the buffer is emptied, declare this input row group consumed
if (upsample.next_row_out >= cinfo.max_v_samp_factor)
{
in_row_group_ctr++;
}
}
// Output some data from the full-image buffer sample in the multi-pass case.
// Always attempts to emit one fully interleaved MCU row ("iMCU" row).
// Return value is JPEG_ROW_COMPLETED, JPEG_SCAN_COMPLETED, or JPEG_SUSPENDED.
//
// NB: output_buf contains a plane for each component in image.
static CONSUME_INPUT output_data_d_diff(jpeg_decompress cinfo, byte[][][] output_buf)
{
jpeg_lossless_d_codec losslsd = (jpeg_lossless_d_codec)cinfo.coef;
d_diff_controller diff = (d_diff_controller)losslsd.diff_private;
uint last_iMCU_row = cinfo.total_iMCU_rows - 1;
// Force some input to be done if we are getting ahead of the input.
while (cinfo.input_scan_number < cinfo.output_scan_number || (cinfo.input_scan_number == cinfo.output_scan_number && cinfo.input_iMCU_row <= cinfo.output_iMCU_row))
{
if (cinfo.inputctl.consume_input(cinfo) == CONSUME_INPUT.JPEG_SUSPENDED)
{
return(CONSUME_INPUT.JPEG_SUSPENDED);
}
}
// OK, output from the arrays.
for (int ci = 0; ci < cinfo.num_components; ci++)
{
jpeg_component_info compptr = cinfo.comp_info[ci];
// Align the buffer for this component.
byte[][] buffer = diff.whole_image[ci];
int samp_rows;
if (cinfo.output_iMCU_row < last_iMCU_row)
{
samp_rows = compptr.v_samp_factor;
}
else
{
// NB: can't use last_row_height here; it is input-side-dependent!
samp_rows = (int)(compptr.height_in_blocks % compptr.v_samp_factor);
if (samp_rows == 0)
{
samp_rows = compptr.v_samp_factor;
}
}
for (int row = 0; row < samp_rows; row++)
{
Array.Copy(buffer[cinfo.output_iMCU_row * compptr.v_samp_factor + row], output_buf[ci][row], compptr.width_in_blocks);
}
}
if (++(cinfo.output_iMCU_row) < cinfo.total_iMCU_rows)
{
return(CONSUME_INPUT.JPEG_ROW_COMPLETED);
}
return(CONSUME_INPUT.JPEG_SCAN_COMPLETED);
}
// Initialize difference buffer controller.
static void jinit_d_diff_controller(jpeg_decompress cinfo, bool need_full_buffer)
{
jpeg_lossless_d_codec losslsd = (jpeg_lossless_d_codec)cinfo.coef;
d_diff_controller diff = null;
try
{
diff = new d_diff_controller();
}
catch
{
ERREXIT1(cinfo, J_MESSAGE_CODE.JERR_OUT_OF_MEMORY, 4);
}
losslsd.diff_private = diff;
losslsd.diff_start_input_pass = start_input_pass_d_diff;
losslsd.start_output_pass = start_output_pass_d_diff;
// Create the [un]difference buffers.
for (int ci = 0; ci < cinfo.num_components; ci++)
{
jpeg_component_info compptr = cinfo.comp_info[ci];
diff.diff_buf[ci] = alloc_darray(cinfo, (uint)jround_up(compptr.width_in_blocks, compptr.h_samp_factor), (uint)compptr.v_samp_factor);
diff.undiff_buf[ci] = alloc_darray(cinfo, (uint)jround_up(compptr.width_in_blocks, compptr.h_samp_factor), (uint)compptr.v_samp_factor);
}
if (need_full_buffer)
{
#if D_MULTISCAN_FILES_SUPPORTED
// Allocate a full-image array for each component.
for (int ci = 0; ci < cinfo.num_components; ci++)
{
jpeg_component_info compptr = cinfo.comp_info[ci];
diff.whole_image[ci] = alloc_sarray(cinfo, (uint)jround_up(compptr.width_in_blocks, compptr.h_samp_factor), (uint)jround_up(compptr.height_in_blocks, compptr.v_samp_factor));
}
losslsd.consume_data = consume_data_d_diff;
losslsd.decompress_data = output_data_d_diff;
#else
ERREXIT(cinfo, J_MESSAGE_CODE.JERR_NOT_COMPILED);
#endif
}
else
{
losslsd.consume_data = dummy_consume_data_d_diff;
losslsd.decompress_data = decompress_data_d_diff;
diff.whole_image[0] = null; // flag for no arrays
}
}
// Initialize preprocessing controller.
static void jinit_c_prep_controller(jpeg_compress cinfo, bool need_full_buffer)
{
if (need_full_buffer)
{
ERREXIT(cinfo, J_MESSAGE_CODE.JERR_BAD_BUFFER_MODE); // safety check
}
my_prep_controller prep = null;
try
{
prep = new my_prep_controller();
}
catch
{
ERREXIT1(cinfo, J_MESSAGE_CODE.JERR_OUT_OF_MEMORY, 4);
}
cinfo.prep = prep;
prep.start_pass = start_pass_prep;
// Allocate the color conversion buffer.
// We make the buffer wide enough to allow the downsampler to edge-expand
// horizontally within the buffer, if it so chooses.
if (cinfo.downsample.need_context_rows)
{
// Set up to provide context rows
#if CONTEXT_ROWS_SUPPORTED
prep.pre_process_data = pre_process_context;
create_context_buffer(cinfo);
#else
ERREXIT(cinfo, J_MESSAGE_CODE.JERR_NOT_COMPILED);
#endif
}
else
{
// No context, just make it tall enough for one row group
prep.pre_process_data = pre_process_data;
for (int ci = 0; ci < cinfo.num_components; ci++)
{
jpeg_component_info compptr = cinfo.comp_info[ci];
prep.color_buf[ci] = alloc_sarray(cinfo,
(uint)(((int)compptr.width_in_blocks * cinfo.DCT_size * cinfo.max_h_samp_factor) / compptr.h_samp_factor),
(uint)cinfo.max_v_samp_factor);
}
}
}
// Process some data in the simple no-context case.
//
// Preprocessor output data is counted in "row groups". A row group
// is defined to be v_samp_factor sample rows of each component.
// Downsampling will produce this much data from each max_v_samp_factor input rows.
static void pre_process_data(jpeg_compress cinfo, byte[][] input_buf, ref uint in_row_ctr, uint in_rows_avail, byte[][][] output_buf, ref uint out_row_group_ctr, uint out_row_groups_avail)
{
my_prep_controller prep = (my_prep_controller)cinfo.prep;
while (in_row_ctr < in_rows_avail && out_row_group_ctr < out_row_groups_avail)
{
// Do color conversion to fill the conversion buffer.
uint inrows = in_rows_avail - in_row_ctr;
int numrows = cinfo.max_v_samp_factor - prep.next_buf_row;
numrows = (int)Math.Min((uint)numrows, inrows);
cinfo.cconvert.color_convert(cinfo, input_buf, in_row_ctr, prep.color_buf, (uint)prep.next_buf_row, numrows);
in_row_ctr += (uint)numrows;
prep.next_buf_row += numrows;
prep.rows_to_go -= (uint)numrows;
// If at bottom of image, pad to fill the conversion buffer.
if (prep.rows_to_go == 0 && prep.next_buf_row < cinfo.max_v_samp_factor)
{
for (int ci = 0; ci < cinfo.num_components; ci++)
{
expand_bottom_edge(prep.color_buf[ci], cinfo.image_width, prep.next_buf_row, cinfo.max_v_samp_factor);
}
prep.next_buf_row = cinfo.max_v_samp_factor;
}
// If we've filled the conversion buffer, empty it.
if (prep.next_buf_row == cinfo.max_v_samp_factor)
{
cinfo.downsample.downsample(cinfo, prep.color_buf, 0, output_buf, out_row_group_ctr);
prep.next_buf_row = 0;
out_row_group_ctr++;
}
// If at bottom of image, pad the output to a full iMCU height.
// Note we assume the caller is providing a one-iMCU-height output buffer!
if (prep.rows_to_go == 0 && out_row_group_ctr < out_row_groups_avail)
{
for (int ci = 0; ci < cinfo.num_components; ci++)
{
jpeg_component_info compptr = cinfo.comp_info[ci];
expand_bottom_edge(output_buf[ci], compptr.width_in_blocks * cinfo.DCT_size, (int)(out_row_group_ctr * compptr.v_samp_factor), (int)(out_row_groups_avail * compptr.v_samp_factor));
}
out_row_group_ctr = out_row_groups_avail;
break; // can exit outer loop without test
}
} // while(...)
}
// Process some data in the first pass of a multi-pass case.
// We process the equivalent of one fully interleaved MCU row ("iMCU" row)
// per call, ie, v_samp_factor rows for each component in the image.
// This amount of data is read from the source buffer and saved into the arrays.
//
// We must also emit the data to the compressor. This is conveniently
// done by calling compress_output_diff() after we've loaded the current strip
// of the arrays.
//
// NB: input_buf contains a plane for each component in image. All components
// are loaded into the arrays in this pass. However, it may be that
// only a subset of the components are emitted to the compressor during
// this first pass; be careful about looking at the scan-dependent variables
// (MCU dimensions, etc).
static bool compress_first_pass_diff(jpeg_compress cinfo, byte[][][] input_buf)
{
jpeg_lossless_c_codec losslsc = (jpeg_lossless_c_codec)cinfo.coef;
c_diff_controller diff = (c_diff_controller)losslsc.diff_private;
uint last_iMCU_row = cinfo.total_iMCU_rows - 1;
for (int ci = 0; ci < cinfo.num_components; ci++)
{
jpeg_component_info compptr = cinfo.comp_info[ci];
// Count non-dummy sample rows in this iMCU row.
int samp_rows;
if (diff.iMCU_row_num < last_iMCU_row)
{
samp_rows = compptr.v_samp_factor;
}
else
{
// NB: can't use last_row_height here, since may not be set!
samp_rows = (int)(compptr.height_in_blocks % compptr.v_samp_factor);
if (samp_rows == 0)
{
samp_rows = compptr.v_samp_factor;
}
}
uint samps_across = compptr.width_in_blocks;
// Perform point transform scaling and prediction/differencing for all
// non-dummy rows in this iMCU row. Each call on these functions
// process a complete row of samples.
for (int samp_row = 0; samp_row < samp_rows; samp_row++)
{
Array.Copy(input_buf[ci][samp_row], diff.whole_image[ci][samp_row + diff.iMCU_row_num * compptr.v_samp_factor], samps_across);
}
}
// NB: compress_output will increment iMCU_row_num if successful.
// A suspension return will result in redoing all the work above next time.
// Emit data to the compressor, sharing code with subsequent passes
return(compress_output_diff(cinfo, input_buf));
}
// Consume input data and store it in the full-image sample buffer.
// We read as much as one fully interleaved MCU row ("iMCU" row) per call,
// ie, v_samp_factor rows for each component in the scan.
// Return value is JPEG_ROW_COMPLETED, JPEG_SCAN_COMPLETED, or JPEG_SUSPENDED.
static CONSUME_INPUT consume_data_d_diff(jpeg_decompress cinfo)
{
jpeg_lossless_d_codec losslsd = (jpeg_lossless_d_codec)cinfo.coef;
d_diff_controller diff = (d_diff_controller)losslsd.diff_private;
uint last_iMCU_row = cinfo.total_iMCU_rows - 1;
byte[][][] buffer = new byte[MAX_COMPS_IN_SCAN][][];
int[] buffer_ind = new int[MAX_COMPS_IN_SCAN];
// Align the buffers for the components used in this scan.
for (int comp = 0; comp < cinfo.comps_in_scan; comp++)
{
jpeg_component_info compptr = cinfo.cur_comp_info[comp];
int ci = compptr.component_index;
buffer[ci] = diff.whole_image[ci];
buffer_ind[ci] = (int)cinfo.input_iMCU_row * compptr.v_samp_factor;
}
return(decompress_data_d_diff(cinfo, buffer, buffer_ind));
}
// Fast processing for the common case of 2:1 horizontal and 1:1 vertical.
// It's still a box filter.
static void h2v1_upsample(jpeg_decompress cinfo, jpeg_component_info compptr, byte[][] input_data, int input_data_offset, byte[][][] output_data_ptr, int output_data_offset)
{
byte[][] output_data = output_data_ptr[output_data_offset];
uint outend = cinfo.output_width;
for (int outrow = 0; outrow < cinfo.max_v_samp_factor; outrow++)
{
byte[] inptr = input_data[input_data_offset++];
uint inptr_ind = 0;
byte[] outptr = output_data[outrow];
uint outptr_ind = 0;
while (outptr_ind < outend)
{
byte invalue = inptr[inptr_ind++];
outptr[outptr_ind++] = invalue;
outptr[outptr_ind++] = invalue;
}
}
}
// Process some data in subsequent passes of a multi-pass case.
// We process the equivalent of one fully interleaved MCU row ("iMCU" row)
// per call, ie, v_samp_factor rows for each component in the scan.
// The data is obtained from the arrays and fed to the compressor.
// Returns true if the iMCU row is completed, false if suspended.
//
// NB: input_buf is ignored; it is likely to be a null pointer.
static bool compress_output_diff(jpeg_compress cinfo, byte[][][] input_buf)
{
jpeg_lossless_c_codec losslsc = (jpeg_lossless_c_codec)cinfo.coef;
c_diff_controller diff = (c_diff_controller)losslsc.diff_private;
byte[][][] buffer = new byte[MAX_COMPONENTS][][];
int[] buffer_ind = new int[MAX_COMPONENTS];
// Align the buffers for the components used in this scan.
// NB: during first pass, this is safe only because the buffers will
// already be aligned properly, so jmemmgr.cs won't need to do any I/O.
for (int comp = 0; comp < cinfo.comps_in_scan; comp++)
{
jpeg_component_info compptr = cinfo.cur_comp_info[comp];
int ci = compptr.component_index;
buffer[ci] = diff.whole_image[ci];
buffer_ind[ci] = (int)diff.iMCU_row_num * compptr.v_samp_factor;
}
return(compress_data_diff(cinfo, buffer, buffer_ind));
}
// This version handles any integral sampling ratios.
// This is not used for typical JPEG files, so it need not be fast.
// Nor, for that matter, is it particularly accurate: the algorithm is
// simple replication of the input pixel onto the corresponding output
// pixels. The hi-falutin sampling literature refers to this as a
// "box filter". A box filter tends to introduce visible artifacts,
// so if you are actually going to use 3:1 or 4:1 sampling ratios
// you would be well advised to improve this code.
static void int_upsample(jpeg_decompress cinfo, jpeg_component_info compptr, byte[][] input_data, int input_data_offset, byte[][][] output_data_ptr, int output_data_offset)
{
my_upsampler upsample = (my_upsampler)cinfo.upsample;
byte[][] output_data = output_data_ptr[output_data_offset];
int h_expand = upsample.h_expand[compptr.component_index];
int v_expand = upsample.v_expand[compptr.component_index];
int outrow = 0;
uint outend = cinfo.output_width;
while (outrow < cinfo.max_v_samp_factor)
{
// Generate one output row with proper horizontal expansion
byte[] inptr = input_data[input_data_offset];
uint inptr_ind = 0;
byte[] outptr = output_data[outrow];
uint outptr_ind = 0;
while (outptr_ind < outend)
{
byte invalue = inptr[inptr_ind++];
for (int h = h_expand; h > 0; h--)
{
outptr[outptr_ind++] = invalue;
}
}
// Generate any additional output rows by duplicating the first one
if (v_expand > 1)
{
jcopy_sample_rows(output_data, outrow, output_data, outrow + 1, v_expand - 1, cinfo.output_width);
}
input_data_offset++;
outrow += v_expand;
}
}
// Downsample pixel values of a single component.
// One row group is processed per call.
// This version handles arbitrary integral sampling ratios, without smoothing.
// Note that this version is not actually used for customary sampling ratios.
static void int_downsample(jpeg_compress cinfo, jpeg_component_info compptr, byte[][] input_data, uint in_row_index, byte[][] output_data, uint out_row_index)
{
uint output_cols=compptr.width_in_blocks*cinfo.DCT_size;
int h_expand=cinfo.max_h_samp_factor/compptr.h_samp_factor;
int v_expand=cinfo.max_v_samp_factor/compptr.v_samp_factor;
int numpix=h_expand*v_expand;
int numpix2=numpix/2;
// 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, in_row_index, cinfo.max_v_samp_factor, cinfo.image_width, (uint)(output_cols*h_expand));
int inrow=(int)in_row_index;
for(int outrow=0; outrow<compptr.v_samp_factor; outrow++)
{
byte[] outptr=output_data[out_row_index+outrow];
for(uint outcol=0, outcol_h=0; outcol<output_cols; outcol++, outcol_h+=(uint)h_expand)
{
byte[] inptr;
int outvalue=0;
for(int v=0; v<v_expand; v++)
{
inptr=input_data[inrow+v];
for(int h=0; h<h_expand; h++) outvalue+=(int)inptr[outcol_h+h];
}
outptr[outcol]=(byte)((outvalue+numpix2)/numpix);
}
inrow+=v_expand;
}
}
// Routines to calculate various quantities related to the size of the image.
// Called once, when first SOS marker is reached
static void initial_setup_d_input(jpeg_decompress cinfo)
{
// Make sure image isn't bigger than I can handle
if ((int)cinfo.image_height > JPEG_MAX_DIMENSION || (int)cinfo.image_width > JPEG_MAX_DIMENSION)
{
ERREXIT1(cinfo, J_MESSAGE_CODE.JERR_IMAGE_TOO_BIG, JPEG_MAX_DIMENSION);
}
if (cinfo.process == J_CODEC_PROCESS.JPROC_LOSSLESS)
{
// If precision > compiled-in value, we must downscale
if (cinfo.data_precision > BITS_IN_JSAMPLE)
{
WARNMS2(cinfo, J_MESSAGE_CODE.JWRN_MUST_DOWNSCALE, cinfo.data_precision, BITS_IN_JSAMPLE);
}
}
else
{ // Lossy processes
// For now, precision must match compiled-in value...
if (cinfo.data_precision != BITS_IN_JSAMPLE)
{
ERREXIT1(cinfo, J_MESSAGE_CODE.JERR_BAD_PRECISION, cinfo.data_precision);
}
}
// Check that number of components won't exceed internal array sizes
if (cinfo.num_components > MAX_COMPONENTS)
{
ERREXIT2(cinfo, J_MESSAGE_CODE.JERR_COMPONENT_COUNT, cinfo.num_components, MAX_COMPONENTS);
}
// Compute maximum sampling factors; check factor validity
cinfo.max_h_samp_factor = 1;
cinfo.max_v_samp_factor = 1;
for (int ci = 0; ci < cinfo.num_components; ci++)
{
jpeg_component_info compptr = cinfo.comp_info[ci];
if (compptr.h_samp_factor <= 0 || compptr.h_samp_factor > MAX_SAMP_FACTOR || compptr.v_samp_factor <= 0 || compptr.v_samp_factor > MAX_SAMP_FACTOR)
{
ERREXIT(cinfo, J_MESSAGE_CODE.JERR_BAD_SAMPLING);
}
cinfo.max_h_samp_factor = Math.Max(cinfo.max_h_samp_factor, compptr.h_samp_factor);
cinfo.max_v_samp_factor = Math.Max(cinfo.max_v_samp_factor, compptr.v_samp_factor);
}
// We initialize DCT_scaled_size and min_DCT_scaled_size to DCTSIZE.
// In the full decompressor, this will be overridden by jdmaster.cs;
// but in the transcoder, jdmaster.cs is not used, so we must do it here.
cinfo.min_DCT_scaled_size = cinfo.DCT_size;
// Compute dimensions of components
for (int ci = 0; ci < cinfo.num_components; ci++)
{
jpeg_component_info compptr = cinfo.comp_info[ci];
compptr.DCT_scaled_size = (uint)cinfo.DCT_size;
// Size in data units
compptr.width_in_blocks = (uint)jdiv_round_up(cinfo.image_width * compptr.h_samp_factor, cinfo.max_h_samp_factor * cinfo.DCT_size);
compptr.height_in_blocks = (uint)jdiv_round_up(cinfo.image_height * compptr.v_samp_factor, cinfo.max_v_samp_factor * cinfo.DCT_size);
// downsampled_width and downsampled_height will also be overridden by
// jdmaster.cs if we are doing full decompression. The transcoder library
// doesn't use these values, but the calling application might.
// Size in samples
compptr.downsampled_width = (uint)jdiv_round_up(cinfo.image_width * compptr.h_samp_factor, cinfo.max_h_samp_factor);
compptr.downsampled_height = (uint)jdiv_round_up(cinfo.image_height * compptr.v_samp_factor, cinfo.max_v_samp_factor);
// Mark component needed, until color conversion says otherwise
compptr.component_needed = true;
// Mark no quantization table yet saved for component
compptr.quant_table = null;
}
// Compute number of fully interleaved MCU rows.
cinfo.total_iMCU_rows = (uint)jdiv_round_up(cinfo.image_height, cinfo.max_v_samp_factor * cinfo.DCT_size);
// Decide whether file contains multiple scans
if (cinfo.comps_in_scan < cinfo.num_components || cinfo.process == J_CODEC_PROCESS.JPROC_PROGRESSIVE)
{
cinfo.inputctl.has_multiple_scans = true;
}
else
{
cinfo.inputctl.has_multiple_scans = false;
}
}
// Downsample pixel values of a single component.
// This version handles the standard case of 2:1 horizontal and 2:1 vertical,
// without smoothing.
static void h2v2_downsample(jpeg_compress cinfo, jpeg_component_info compptr, byte[][] input_data, uint in_row_index, byte[][] output_data, uint out_row_index)
{
uint output_cols=compptr.width_in_blocks*cinfo.DCT_size;
// 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, in_row_index, cinfo.max_v_samp_factor, cinfo.image_width, output_cols*2);
int inrow=(int)in_row_index;
for(int outrow=0; outrow<compptr.v_samp_factor; outrow++)
{
byte[] outptr=output_data[out_row_index+outrow];
byte[] inptr0=input_data[inrow];
byte[] inptr1=input_data[inrow+1];
int bias=1; // bias = 1,2,1,2,... for successive samples
for(uint outcol=0, ind=0; outcol<output_cols; outcol++, ind+=2)
{
outptr[outcol]=(byte)((inptr0[ind]+inptr0[ind+1]+inptr1[ind]+inptr1[ind+1]+bias)>>2);
bias^=3; // 1=>2, 2=>1
}
inrow+=2;
}
}
// Downsample pixel values of a single component.
// This version handles the special case of a full-size component,
// without smoothing.
static void fullsize_downsample(jpeg_compress cinfo, jpeg_component_info compptr, byte[][] input_data, uint in_row_index, byte[][] output_data, uint out_row_index)
{
// Copy the data
jcopy_sample_rows(input_data, (int)in_row_index, output_data, (int)out_row_index, cinfo.max_v_samp_factor, cinfo.image_width);
// Edge-expand
expand_right_edge(output_data, out_row_index, cinfo.max_v_samp_factor, cinfo.image_width, compptr.width_in_blocks*cinfo.DCT_size);
}
// Fast processing for the common case of 2:1 horizontal and 2:1 vertical.
// It's still a box filter.
#if! USE_UNSAFE_STUFF
static void h2v2_fancy_upsample(jpeg_decompress cinfo, jpeg_component_info compptr, byte[][] input_data, int input_data_offset, byte[][][] output_data_ptr, int output_data_offset)
{
if(!compptr.notFirst)
{
input_data[0].CopyTo(input_data[input_data.Length-1], 0);
compptr.notFirst=true;
}
byte[][] output_data=output_data_ptr[output_data_offset];
int inrow=0, outrow=0;
while(outrow<cinfo.max_v_samp_factor)
{
for(int v=0; v<2; v++)
{
// inptr0 points to nearest input row, inptr1 points to next nearest
byte[] inptr0=input_data[input_data_offset+inrow];
byte[] inptr1=null;
if(v==0)
{
inptr1=input_data[input_data_offset==0?input_data.Length-1:input_data_offset+inrow-1]; // next nearest is row above
}
else inptr1=input_data[(input_data_offset+inrow+1)%input_data.Length]; // next nearest is row below
uint intptr_ind=0;
byte[] outptr=output_data[outrow++];
uint outptr_ind=0;
// Special case for first column
int thiscolsum=inptr0[intptr_ind]*3+inptr1[intptr_ind]; intptr_ind++;
int nextcolsum=inptr0[intptr_ind]*3+inptr1[intptr_ind]; intptr_ind++;
outptr[outptr_ind++]=(byte)((thiscolsum*4+8)>>4);
outptr[outptr_ind++]=(byte)((thiscolsum*3+nextcolsum+7)>>4);
int lastcolsum=thiscolsum;
thiscolsum=nextcolsum;
for(uint 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=inptr0[intptr_ind]*3+inptr1[intptr_ind]; intptr_ind++;
outptr[outptr_ind++]=(byte)((thiscolsum*3+lastcolsum+8)>>4);
outptr[outptr_ind++]=(byte)((thiscolsum*3+nextcolsum+7)>>4);
lastcolsum=thiscolsum; thiscolsum=nextcolsum;
}
// Special case for last column
outptr[outptr_ind++]=(byte)((thiscolsum*3+lastcolsum+8)>>4);
outptr[outptr_ind++]=(byte)((thiscolsum*4+7)>>4);
}
inrow++;
}
}
// Downsample pixel values of a single component.
// This version handles the standard case of 2:1 horizontal and 2:1 vertical,
// with smoothing. One row of context is required.
static void h2v2_smooth_downsample(jpeg_compress cinfo, jpeg_component_info compptr, byte[][] input_data, uint in_row_index, byte[][] output_data, uint out_row_index)
{
uint output_cols=compptr.width_in_blocks*cinfo.DCT_size;
// 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, in_row_index-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.
int memberscale=16384-cinfo.smoothing_factor*80; // scaled (1-5*SF)/4
int neighscale=cinfo.smoothing_factor*16; // scaled SF/4
int inrow=(int)in_row_index;
for(int outrow=0; outrow<compptr.v_samp_factor; outrow++)
{
byte[] outptr=output_data[out_row_index+outrow];
byte[] inptr0=input_data[inrow];
byte[] inptr1=input_data[inrow+1];
byte[] above_ptr=input_data[inrow-1];
byte[] below_ptr=input_data[inrow+2];
// Special case for first column: pretend column -1 is same as column 0
int membersum=inptr0[0]+inptr0[1]+inptr1[0]+inptr1[1];
int neighsum=above_ptr[0]+above_ptr[1]+below_ptr[0]+below_ptr[1]+inptr0[0]+inptr0[2]+inptr1[0]+inptr1[2];
neighsum+=neighsum;
neighsum+=above_ptr[0]+above_ptr[2]+below_ptr[0]+below_ptr[2];
membersum=membersum*memberscale+neighsum*neighscale;
outptr[0]=(byte)((membersum+32768)>>16);
int iind=2, oind=1;
for(uint colctr=output_cols-2; colctr>0; colctr--)
{
// sum of pixels directly mapped to this output element
membersum=inptr0[iind]+inptr0[iind+1]+inptr1[iind]+inptr1[iind+1];
// sum of edge-neighbor pixels
neighsum=above_ptr[iind]+above_ptr[iind+1]+below_ptr[iind]+below_ptr[iind+1]+inptr0[iind-1]+inptr0[iind+2]+inptr1[iind-1]+inptr1[iind+2];
// The edge-neighbors count twice as much as corner-neighbors
neighsum+=neighsum;
// Add in the corner-neighbors
neighsum+=above_ptr[iind-1]+above_ptr[iind+2]+below_ptr[iind-1]+below_ptr[iind+2];
// form final output scaled up by 2^16
membersum=membersum*memberscale+neighsum*neighscale;
// round, descale and output it
outptr[oind]=(byte)((membersum+32768)>>16);
iind+=2;
oind++;
}
// Special case for last column
membersum=inptr0[iind]+inptr0[iind+1]+inptr1[iind]+inptr1[iind+1];
neighsum=above_ptr[iind]+above_ptr[iind+1]+below_ptr[iind]+below_ptr[iind+1]+inptr0[iind-1]+inptr0[iind+1]+inptr1[iind-1]+inptr1[iind+1];
neighsum+=neighsum;
neighsum+=above_ptr[iind-1]+above_ptr[iind+1]+below_ptr[iind-1]+below_ptr[iind+1];
membersum=membersum*memberscale+neighsum*neighscale;
outptr[oind]=(byte)((membersum+32768)>>16);
inrow+=2;
}
}
static CONSUME_INPUT decompress_data_d_diff(jpeg_decompress cinfo, byte[][][] output_buf, int[] output_buf_ind)
{
jpeg_lossless_d_codec losslsd = (jpeg_lossless_d_codec)cinfo.coef;
d_diff_controller diff = (d_diff_controller)losslsd.diff_private;
// Loop to process as much as one whole iMCU row
for (uint yoffset = diff.MCU_vert_offset; yoffset < diff.MCU_rows_per_iMCU_row; yoffset++)
{
// Process restart marker if needed; may have to suspend
if (cinfo.restart_interval != 0)
{
if (diff.restart_rows_to_go == 0)
{
if (!process_restart_d_diff(cinfo))
{
return(CONSUME_INPUT.JPEG_SUSPENDED);
}
}
}
uint MCU_col_num = diff.MCU_ctr; // index of current MCU within row
// Try to fetch an MCU-row (or remaining portion of suspended MCU-row).
uint MCU_count = losslsd.entropy_decode_mcus(cinfo, diff.diff_buf, yoffset, MCU_col_num, cinfo.MCUs_per_row - MCU_col_num);
if (MCU_count != cinfo.MCUs_per_row - MCU_col_num)
{
// Suspension forced; update state counters and exit
diff.MCU_vert_offset = yoffset;
diff.MCU_ctr += MCU_count;
return(CONSUME_INPUT.JPEG_SUSPENDED);
}
// Account for restart interval (no-op if not using restarts)
diff.restart_rows_to_go--;
// Completed an MCU row, but perhaps not an iMCU row
diff.MCU_ctr = 0;
}
uint last_iMCU_row = cinfo.total_iMCU_rows - 1;
// Undifference and scale each scanline of the disassembled MCU-row
// separately. We do not process dummy samples at the end of a scanline
// or dummy rows at the end of the image.
for (int comp = 0; comp < cinfo.comps_in_scan; comp++)
{
jpeg_component_info compptr = cinfo.cur_comp_info[comp];
int ci = compptr.component_index;
int stop = cinfo.input_iMCU_row == last_iMCU_row?compptr.last_row_height:compptr.v_samp_factor;
for (int row = 0, prev_row = compptr.v_samp_factor - 1; row < stop; prev_row = row, row++)
{
losslsd.predict_undifference[ci](cinfo, ci, diff.diff_buf[ci][row], diff.undiff_buf[ci][prev_row], diff.undiff_buf[ci][row], compptr.width_in_blocks);
losslsd.scaler_scale(cinfo, diff.undiff_buf[ci][row], output_buf[ci][output_buf_ind[ci] + row], compptr.width_in_blocks);
}
}
// Completed the iMCU row, advance counters for next one.
//
// NB: output_data will increment output_iMCU_row.
// This counter is not needed for the single-pass case
// or the input side of the multi-pass case.
if (++(cinfo.input_iMCU_row) < cinfo.total_iMCU_rows)
{
start_iMCU_row_d_diff(cinfo);
return(CONSUME_INPUT.JPEG_ROW_COMPLETED);
}
// Completed the scan
cinfo.inputctl.finish_input_pass(cinfo);
return(CONSUME_INPUT.JPEG_SCAN_COMPLETED);
}
static void h2v2_fancy_upsample(jpeg_decompress cinfo, jpeg_component_info compptr, byte[][] input_data, int input_data_offset, byte[][][] output_data_ptr, int output_data_offset)
{
unsafe
{
if(!compptr.notFirst)
{
input_data[0].CopyTo(input_data[input_data.Length-1], 0);
compptr.notFirst=true;
}
byte[][] output_data=output_data_ptr[output_data_offset];
int inrow=0, outrow=0;
while(outrow<cinfo.max_v_samp_factor)
{
// inptr0 points to nearest input row
fixed(byte* inptr0_=input_data[input_data_offset+inrow])
{
// inptr1 points to next nearest
int nextnearestrow=input_data_offset==0?input_data.Length-1:input_data_offset+inrow-1; // next nearest is row above
for(int v=0; v<2; v++)
{
fixed(byte* inptr1_=input_data[nextnearestrow], outptr_=output_data[outrow++])
{
byte* inptr0=inptr0_, inptr1=inptr1_, outptr=outptr_;
// Special case for first column
int thiscolsum=*(inptr0++)*3+*(inptr1++);
int nextcolsum=*(inptr0++)*3+*(inptr1++);
*(outptr++)=(byte)((thiscolsum*4+8)>>4);
*(outptr++)=(byte)((thiscolsum*3+nextcolsum+7)>>4);
int lastcolsum=thiscolsum;
thiscolsum=nextcolsum;
for(uint 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=*(inptr0++)*3+*(inptr1++);
*(outptr++)=(byte)((thiscolsum*3+lastcolsum+8)>>4);
*(outptr++)=(byte)((thiscolsum*3+nextcolsum+7)>>4);
lastcolsum=thiscolsum; thiscolsum=nextcolsum;
}
// Special case for last column
*(outptr++)=(byte)((thiscolsum*3+lastcolsum+8)>>4);
*(outptr++)=(byte)((thiscolsum*4+7)>>4);
}
nextnearestrow=(input_data_offset+inrow+1)%input_data.Length; // next nearest is row below
}
}
inrow++;
}
}
}
// Module initialization routine for downsampling.
// Note that we must select a routine for each component.
static void jinit_downsampler(jpeg_compress cinfo)
{
my_downsampler downsample = null;
try
{
downsample = new my_downsampler();
}
catch
{
ERREXIT1(cinfo, J_MESSAGE_CODE.JERR_OUT_OF_MEMORY, 4);
}
cinfo.downsample = downsample;
downsample.start_pass = start_pass_downsample;
downsample.downsample = sep_downsample;
downsample.need_context_rows = false;
if (cinfo.CCIR601_sampling)
{
ERREXIT(cinfo, J_MESSAGE_CODE.JERR_CCIR601_NOTIMPL);
}
bool smoothok = true;
// Verify we can handle the sampling factors, and set up method pointers
for (int ci = 0; ci < cinfo.num_components; ci++)
{
jpeg_component_info compptr = cinfo.comp_info[ci];
if (compptr.h_samp_factor == cinfo.max_h_samp_factor && compptr.v_samp_factor == cinfo.max_v_samp_factor)
{
#if INPUT_SMOOTHING_SUPPORTED
if (cinfo.smoothing_factor != 0)
{
downsample.methods[ci] = fullsize_smooth_downsample;
downsample.need_context_rows = true;
}
else
#endif
downsample.methods[ci] = fullsize_downsample;
}
else if (compptr.h_samp_factor * 2 == cinfo.max_h_samp_factor && compptr.v_samp_factor == cinfo.max_v_samp_factor)
{
smoothok = false;
downsample.methods[ci] = h2v1_downsample;
}
else if (compptr.h_samp_factor * 2 == cinfo.max_h_samp_factor && compptr.v_samp_factor * 2 == cinfo.max_v_samp_factor)
{
#if INPUT_SMOOTHING_SUPPORTED
if (cinfo.smoothing_factor != 0)
{
downsample.methods[ci] = h2v2_smooth_downsample;
downsample.need_context_rows = true;
}
else
#endif
downsample.methods[ci] = h2v2_downsample;
}
else if ((cinfo.max_h_samp_factor % compptr.h_samp_factor) == 0 && (cinfo.max_v_samp_factor % compptr.v_samp_factor) == 0)
{
smoothok = false;
downsample.methods[ci] = int_downsample;
}
else
{
ERREXIT(cinfo, J_MESSAGE_CODE.JERR_FRACT_SAMPLE_NOTIMPL);
}
}
#if INPUT_SMOOTHING_SUPPORTED
if (cinfo.smoothing_factor != 0 && !smoothok)
{
TRACEMS(cinfo, 0, J_MESSAGE_CODE.JTRC_SMOOTH_NOTIMPL);
}
#endif
}
// Module initialization routine for upsampling.
public static void jinit_upsampler(jpeg_decompress cinfo)
{
my_upsampler upsample = null;
bool need_buffer;
int h_in_group, v_in_group, h_out_group, v_out_group;
try
{
upsample = new my_upsampler();
}
catch
{
ERREXIT1(cinfo, J_MESSAGE_CODE.JERR_OUT_OF_MEMORY, 4);
}
cinfo.upsample = upsample;
upsample.start_pass = start_pass_upsample;
upsample.upsample = sep_upsample;
#if UPSCALING_CONTEXT
upsample.need_context_rows = false; // until we find out differently
#endif
if (cinfo.CCIR601_sampling)
{
ERREXIT(cinfo, J_MESSAGE_CODE.JERR_CCIR601_NOTIMPL); // this isn't supported
}
// jdmainct.cs doesn't support context rows when min_DCT_scaled_size = 1,
// so don't ask for it.
bool do_fancy = cinfo.do_fancy_upsampling && cinfo.min_DCT_scaled_size > 1;
// Verify we can handle the sampling factors, select per-component methods,
// and create storage as needed.
for (int ci = 0; ci < cinfo.num_components; ci++)
{
jpeg_component_info compptr = cinfo.comp_info[ci];
// Compute size of an "input group" after IDCT scaling. This many samples
// are to be converted to max_h_samp_factor * max_v_samp_factor pixels.
h_in_group = (int)(compptr.h_samp_factor * compptr.DCT_scaled_size) / cinfo.min_DCT_scaled_size;
v_in_group = (int)(compptr.v_samp_factor * compptr.DCT_scaled_size) / cinfo.min_DCT_scaled_size;
h_out_group = cinfo.max_h_samp_factor;
v_out_group = cinfo.max_v_samp_factor;
upsample.rowgroup_height[ci] = v_in_group; // save for use later
need_buffer = true;
if (!compptr.component_needed)
{
// Don't bother to upsample an uninteresting component.
upsample.methods[ci] = noop_upsample;
need_buffer = false;
}
else if (h_in_group == h_out_group && v_in_group == v_out_group)
{
// Fullsize components can be processed without any work.
upsample.methods[ci] = fullsize_upsample;
need_buffer = false;
upsample.color_buf[ci] = new byte[cinfo.max_v_samp_factor][];
}
else if (h_in_group * 2 == h_out_group && v_in_group == v_out_group)
{
// Special cases for 2h1v upsampling
if (do_fancy && compptr.downsampled_width > 2)
{
upsample.methods[ci] = h2v1_fancy_upsample;
}
else
{
upsample.methods[ci] = h2v1_upsample;
}
}
else if (h_in_group * 2 == h_out_group && v_in_group * 2 == v_out_group)
{
// Special cases for 2h2v upsampling
#if UPSCALING_CONTEXT
if (do_fancy && compptr.downsampled_width > 2)
{
upsample.methods[ci] = h2v2_fancy_upsample;
upsample.need_context_rows = true;
compptr.doContext = true;
}
else
#endif
upsample.methods[ci] = h2v2_upsample;
}
else if ((h_out_group % h_in_group) == 0 && (v_out_group % v_in_group) == 0)
{
// Generic integral-factors upsampling method
upsample.methods[ci] = int_upsample;
upsample.h_expand[ci] = (byte)(h_out_group / h_in_group);
upsample.v_expand[ci] = (byte)(v_out_group / v_in_group);
}
else
{
ERREXIT(cinfo, J_MESSAGE_CODE.JERR_FRACT_SAMPLE_NOTIMPL);
}
if (need_buffer)
{
upsample.color_buf[ci] = alloc_sarray(cinfo, (uint)jround_up((int)cinfo.output_width, (int)cinfo.max_h_samp_factor), (uint)cinfo.max_v_samp_factor);
}
}
}
// These are the routines invoked by sep_upsample to upsample pixel values
// of a single component. One row group is processed per call.
// For full-size components, we just make color_buf[ci] point at the
// input buffer, and thus avoid copying any data. Note that this is
// safe only because sep_upsample doesn't declare the input row group
// "consumed" until we are done color converting and emitting it.
static void fullsize_upsample(jpeg_decompress cinfo, jpeg_component_info compptr, byte[][] input_data, int input_data_offset, byte[][][] output_data_ptr, int output_data_offset)
{
for(int i=0; i<cinfo.max_v_samp_factor; i++) output_data_ptr[output_data_offset][i]=input_data[input_data_offset++];
}
// Downsample pixel values of a single component.
// This version handles the standard case of 2:1 horizontal and 2:1 vertical,
// with smoothing. One row of context is required.
static void h2v2_smooth_downsample(jpeg_compress cinfo, jpeg_component_info compptr, byte[][] input_data, uint in_row_index, byte[][] output_data, uint out_row_index)
{
uint output_cols = compptr.width_in_blocks * cinfo.DCT_size;
// 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, in_row_index - 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.
int memberscale = 16384 - cinfo.smoothing_factor * 80; // scaled (1-5*SF)/4
int neighscale = cinfo.smoothing_factor * 16; // scaled SF/4
int inrow = (int)in_row_index;
for (int outrow = 0; outrow < compptr.v_samp_factor; outrow++)
{
byte[] outptr = output_data[out_row_index + outrow];
byte[] inptr0 = input_data[inrow];
byte[] inptr1 = input_data[inrow + 1];
byte[] above_ptr = input_data[inrow - 1];
byte[] below_ptr = input_data[inrow + 2];
// Special case for first column: pretend column -1 is same as column 0
int membersum = inptr0[0] + inptr0[1] + inptr1[0] + inptr1[1];
int neighsum = above_ptr[0] + above_ptr[1] + below_ptr[0] + below_ptr[1] + inptr0[0] + inptr0[2] + inptr1[0] + inptr1[2];
neighsum += neighsum;
neighsum += above_ptr[0] + above_ptr[2] + below_ptr[0] + below_ptr[2];
membersum = membersum * memberscale + neighsum * neighscale;
outptr[0] = (byte)((membersum + 32768) >> 16);
int iind = 2, oind = 1;
for (uint colctr = output_cols - 2; colctr > 0; colctr--)
{
// sum of pixels directly mapped to this output element
membersum = inptr0[iind] + inptr0[iind + 1] + inptr1[iind] + inptr1[iind + 1];
// sum of edge-neighbor pixels
neighsum = above_ptr[iind] + above_ptr[iind + 1] + below_ptr[iind] + below_ptr[iind + 1] + inptr0[iind - 1] + inptr0[iind + 2] + inptr1[iind - 1] + inptr1[iind + 2];
// The edge-neighbors count twice as much as corner-neighbors
neighsum += neighsum;
// Add in the corner-neighbors
neighsum += above_ptr[iind - 1] + above_ptr[iind + 2] + below_ptr[iind - 1] + below_ptr[iind + 2];
// form final output scaled up by 2^16
membersum = membersum * memberscale + neighsum * neighscale;
// round, descale and output it
outptr[oind] = (byte)((membersum + 32768) >> 16);
iind += 2;
oind++;
}
// Special case for last column
membersum = inptr0[iind] + inptr0[iind + 1] + inptr1[iind] + inptr1[iind + 1];
neighsum = above_ptr[iind] + above_ptr[iind + 1] + below_ptr[iind] + below_ptr[iind + 1] + inptr0[iind - 1] + inptr0[iind + 1] + inptr1[iind - 1] + inptr1[iind + 1];
neighsum += neighsum;
neighsum += above_ptr[iind - 1] + above_ptr[iind + 1] + below_ptr[iind - 1] + below_ptr[iind + 1];
membersum = membersum * memberscale + neighsum * neighscale;
outptr[oind] = (byte)((membersum + 32768) >> 16);
inrow += 2;
}
}
// This is a no-op version used for "uninteresting" components.
// These components will not be referenced by color conversion.
static void noop_upsample(jpeg_decompress cinfo, jpeg_component_info compptr, byte[][] input_data, int input_data_offset, byte[][][] output_data_ptr, int output_data_offset)
{
output_data_ptr[output_data_offset]=null; // safety check
}
// This version handles any integral sampling ratios.
// This is not used for typical JPEG files, so it need not be fast.
// Nor, for that matter, is it particularly accurate: the algorithm is
// simple replication of the input pixel onto the corresponding output
// pixels. The hi-falutin sampling literature refers to this as a
// "box filter". A box filter tends to introduce visible artifacts,
// so if you are actually going to use 3:1 or 4:1 sampling ratios
// you would be well advised to improve this code.
static void int_upsample(jpeg_decompress cinfo, jpeg_component_info compptr, byte[][] input_data, int input_data_offset, byte[][][] output_data_ptr, int output_data_offset)
{
my_upsampler upsample=(my_upsampler)cinfo.upsample;
byte[][] output_data=output_data_ptr[output_data_offset];
int h_expand=upsample.h_expand[compptr.component_index];
int v_expand=upsample.v_expand[compptr.component_index];
int outrow=0;
uint outend=cinfo.output_width;
while(outrow<cinfo.max_v_samp_factor)
{
// Generate one output row with proper horizontal expansion
byte[] inptr=input_data[input_data_offset];
uint inptr_ind=0;
byte[] outptr=output_data[outrow];
uint outptr_ind=0;
while(outptr_ind<outend)
{
byte invalue=inptr[inptr_ind++];
for(int h=h_expand; h>0; h--) outptr[outptr_ind++]=invalue;
}
// Generate any additional output rows by duplicating the first one
if(v_expand>1) jcopy_sample_rows(output_data, outrow, output_data, outrow+1, v_expand-1, cinfo.output_width);
input_data_offset++;
outrow+=v_expand;
}
}
// Fast processing for the common case of 2:1 horizontal and 1:1 vertical.
// It's still a box filter.
static void h2v1_upsample(jpeg_decompress cinfo, jpeg_component_info compptr, byte[][] input_data, int input_data_offset, byte[][][] output_data_ptr, int output_data_offset)
{
byte[][] output_data=output_data_ptr[output_data_offset];
uint outend=cinfo.output_width;
for(int outrow=0; outrow<cinfo.max_v_samp_factor; outrow++)
{
byte[] inptr=input_data[input_data_offset++];
uint inptr_ind=0;
byte[] outptr=output_data[outrow];
uint outptr_ind=0;
while(outptr_ind<outend)
{
byte invalue=inptr[inptr_ind++];
outptr[outptr_ind++]=invalue;
outptr[outptr_ind++]=invalue;
}
}
}
// Do computations that are needed before processing a JPEG scan
// cinfo.comps_in_scan and cinfo.cur_comp_info[] were set from SOS marker
static void per_scan_setup_d_input(jpeg_decompress cinfo)
{
if (cinfo.comps_in_scan == 1)
{
// Noninterleaved (single-component) scan
jpeg_component_info compptr = cinfo.cur_comp_info[0];
// Overall image size in MCUs
cinfo.MCUs_per_row = compptr.width_in_blocks;
cinfo.MCU_rows_in_scan = compptr.height_in_blocks;
// For noninterleaved scan, always one block per MCU
compptr.MCU_width = 1;
compptr.MCU_height = 1;
compptr.MCU_blocks = 1;
compptr.MCU_sample_width = (int)compptr.DCT_scaled_size;
compptr.last_col_width = 1;
// For noninterleaved scans, it is convenient to define last_row_height
// as the number of block rows present in the last iMCU row.
int tmp = (int)(compptr.height_in_blocks % compptr.v_samp_factor);
if (tmp == 0)
{
tmp = compptr.v_samp_factor;
}
compptr.last_row_height = tmp;
// Prepare array describing MCU composition
cinfo.blocks_in_MCU = 1;
cinfo.MCU_membership[0] = 0;
}
else
{
// Interleaved (multi-component) scan
if (cinfo.comps_in_scan <= 0 || cinfo.comps_in_scan > MAX_COMPS_IN_SCAN)
{
ERREXIT2(cinfo, J_MESSAGE_CODE.JERR_COMPONENT_COUNT, cinfo.comps_in_scan, MAX_COMPS_IN_SCAN);
}
// Overall image size in MCUs
cinfo.MCUs_per_row = (uint)jdiv_round_up(cinfo.image_width, cinfo.max_h_samp_factor * cinfo.DCT_size);
cinfo.MCU_rows_in_scan = (uint)jdiv_round_up(cinfo.image_height, cinfo.max_v_samp_factor * cinfo.DCT_size);
cinfo.blocks_in_MCU = 0;
for (int ci = 0; ci < cinfo.comps_in_scan; ci++)
{
jpeg_component_info compptr = cinfo.cur_comp_info[ci];
// Sampling factors give # of blocks of component in each MCU
compptr.MCU_width = compptr.h_samp_factor;
compptr.MCU_height = compptr.v_samp_factor;
compptr.MCU_blocks = (uint)(compptr.MCU_width * compptr.MCU_height);
compptr.MCU_sample_width = (int)(compptr.MCU_width * compptr.DCT_scaled_size);
// Figure number of non-dummy blocks in last MCU column & row
int tmp = (int)(compptr.width_in_blocks % compptr.MCU_width);
if (tmp == 0)
{
tmp = compptr.MCU_width;
}
compptr.last_col_width = tmp;
tmp = (int)(compptr.height_in_blocks % compptr.MCU_height);
if (tmp == 0)
{
tmp = compptr.MCU_height;
}
compptr.last_row_height = tmp;
// Prepare array describing MCU composition
int mcublks = (int)compptr.MCU_blocks;
if (cinfo.blocks_in_MCU + mcublks > D_MAX_BLOCKS_IN_MCU)
{
ERREXIT(cinfo, J_MESSAGE_CODE.JERR_BAD_MCU_SIZE);
}
while ((mcublks--) > 0)
{
cinfo.MCU_membership[cinfo.blocks_in_MCU++] = ci;
}
}
}
}
static void h2v1_fancy_upsample(jpeg_decompress cinfo, jpeg_component_info compptr, byte[][] input_data, int input_data_offset, byte[][][] output_data_ptr, int output_data_offset)
{
unsafe
{
byte[][] output_data=output_data_ptr[output_data_offset];
for(int inrow=0; inrow<cinfo.max_v_samp_factor; inrow++)
{
fixed(byte* inptr_=input_data[input_data_offset++], outptr_=output_data[inrow])
{
byte* inptr=inptr_;
byte* outptr=outptr_;
// Special case for first column
int invalue=*(inptr++);
*(outptr++)=(byte)invalue;
*(outptr++)=(byte)((invalue*3+*inptr+2)>>2);
for(uint colctr=compptr.downsampled_width-2; colctr>0; colctr--)
{
// General case: 3/4 * nearer pixel + 1/4 * further pixel
invalue=*(inptr++)*3;
*(outptr++)=(byte)((invalue+inptr[-2]+1)>>2);
*(outptr++)=(byte)((invalue+*inptr+2)>>2);
}
// Special case for last column
invalue=*inptr;
*(outptr++)=(byte)((invalue*3+inptr[-1]+1)>>2);
*outptr=(byte)invalue;
}
}
}
}
// Downsample pixel values of a single component.
// This version handles the special case of a full-size component,
// with smoothing. One row of context is required.
static void fullsize_smooth_downsample(jpeg_compress cinfo, jpeg_component_info compptr, byte[][] input_data, uint in_row_index, byte[][] output_data, uint out_row_index)
{
uint output_cols=compptr.width_in_blocks*cinfo.DCT_size;
// 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, in_row_index-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.
int memberscale=65536-cinfo.smoothing_factor*512; // scaled 1-8*SF
int neighscale=cinfo.smoothing_factor*64; // scaled SF
int membersum, neighsum;
int colsum, lastcolsum, nextcolsum;
for(int outrow=0; outrow<compptr.v_samp_factor; outrow++)
{
byte[] outptr=output_data[out_row_index+outrow];
byte[] inptr=input_data[in_row_index+outrow];
byte[] above_ptr=input_data[in_row_index+outrow-1];
byte[] below_ptr=input_data[in_row_index+outrow+1];
int iind=0, aind=0, bind=0, oind=0;
// Special case for first column
colsum=above_ptr[aind++]+below_ptr[bind++]+inptr[iind];
membersum=inptr[iind++];
nextcolsum=above_ptr[aind]+below_ptr[bind]+inptr[iind];
neighsum=colsum+(colsum-membersum)+nextcolsum;
membersum=membersum*memberscale+neighsum*neighscale;
outptr[oind++]=(byte)((membersum+32768)>>16);
lastcolsum=colsum; colsum=nextcolsum;
for(uint colctr=output_cols-2; colctr>0; colctr--)
{
membersum=inptr[iind++];
nextcolsum=above_ptr[aind++]+below_ptr[bind++]+inptr[iind];
neighsum=lastcolsum+(colsum-membersum)+nextcolsum;
membersum=membersum*memberscale+neighsum*neighscale;
outptr[oind++]=(byte)((membersum+32768)>>16);
lastcolsum=colsum; colsum=nextcolsum;
}
// Special case for last column
membersum=inptr[iind];
neighsum=lastcolsum+(colsum-membersum)+colsum;
membersum=membersum*memberscale+neighsum*neighscale;
outptr[oind]=(byte)((membersum+32768)>>16);
}
}
// Fast processing for the common case of 2:1 horizontal and 2:1 vertical.
// It's still a box filter.
static void h2v2_upsample(jpeg_decompress cinfo, jpeg_component_info compptr, byte[][] input_data, int input_data_offset, byte[][][] output_data_ptr, int output_data_offset)
{
byte[][] output_data=output_data_ptr[output_data_offset];
int outrow=0;
uint outend=cinfo.output_width;
while(outrow<cinfo.max_v_samp_factor)
{
byte[] inptr=input_data[input_data_offset];
uint inptr_ind=0;
byte[] outptr=output_data[outrow];
uint outptr_ind=0;
while(outptr_ind<outend)
{
byte invalue=inptr[inptr_ind++];
outptr[outptr_ind++]=invalue;
outptr[outptr_ind++]=invalue;
}
Array.Copy(output_data[outrow], output_data[outrow+1], cinfo.output_width);
input_data_offset++;
outrow+=2;
}
}
// This is a no-op version used for "uninteresting" components.
// These components will not be referenced by color conversion.
static void noop_upsample(jpeg_decompress cinfo, jpeg_component_info compptr, byte[][] input_data, int input_data_offset, byte[][][] output_data_ptr, int output_data_offset)
{
output_data_ptr[output_data_offset] = null; // safety check
}
// Fancy processing for the common case of 2:1 horizontal and 1:1 vertical.
//
// The upsampling algorithm is linear interpolation between pixel centers,
// also known as a "triangle filter". This is a good compromise between
// speed and visual quality. The centers of the output pixels are 1/4 and 3/4
// of the way between input pixel centers.
//
// A note about the "bias" calculations: when rounding fractional values to
// integer, we do not want to always round 0.5 up to the next integer.
// If we did that, we'd introduce a noticeable bias towards larger values.
// Instead, this code is arranged so that 0.5 will be rounded up or down at
// alternate pixel locations (a simple ordered dither pattern).
#if! USE_UNSAFE_STUFF
static void h2v1_fancy_upsample(jpeg_decompress cinfo, jpeg_component_info compptr, byte[][] input_data, int input_data_offset, byte[][][] output_data_ptr, int output_data_offset)
{
byte[][] output_data=output_data_ptr[output_data_offset];
for(int inrow=0; inrow<cinfo.max_v_samp_factor; inrow++)
{
byte[] inptr=input_data[input_data_offset++];
uint inptr_ind=0;
byte[] outptr=output_data[inrow];
uint outptr_ind=0;
// Special case for first column
int invalue=inptr[inptr_ind++];
outptr[outptr_ind++]=(byte)invalue;
outptr[outptr_ind++]=(byte)((invalue*3+inptr[inptr_ind]+2)>>2);
for(uint colctr=compptr.downsampled_width-2; colctr>0; colctr--)
{
// General case: 3/4 * nearer pixel + 1/4 * further pixel
invalue=inptr[inptr_ind++]*3;
outptr[outptr_ind++]=(byte)((invalue+inptr[inptr_ind-2]+1)>>2);
outptr[outptr_ind++]=(byte)((invalue+inptr[inptr_ind]+2)>>2);
}
// Special case for last column
invalue=inptr[inptr_ind];
outptr[outptr_ind++]=(byte)((invalue*3+inptr[inptr_ind-1]+1)>>2);
outptr[outptr_ind]=(byte)invalue;
}
}
