private void jpeg_idct_float(int component_index, short[] coef_block, int output_row, int output_col)
{
/* buffers data between passes */
float[] workspace = new float[JpegConstants.DCTSIZE2];
/* Pass 1: process columns from input, store into work array. */
int coefBlockIndex = 0;
int workspaceIndex = 0;
float[] quantTable = m_dctTables[component_index].float_array;
int quantTableIndex = 0;
for (int ctr = JpegConstants.DCTSIZE; ctr > 0; ctr--)
{
/* Due to quantization, we will usually find that many of the input
* coefficients are zero, especially the AC terms. We can exploit this
* by short-circuiting the IDCT calculation for any column in which all
* the AC terms are zero. In that case each output is equal to the
* DC coefficient (with scale factor as needed).
* With typical images and quantization tables, half or more of the
* column DCT calculations can be simplified this way.
*/
if (coef_block[coefBlockIndex + JpegConstants.DCTSIZE * 1] == 0 &&
coef_block[coefBlockIndex + JpegConstants.DCTSIZE * 2] == 0 &&
coef_block[coefBlockIndex + JpegConstants.DCTSIZE * 3] == 0 &&
coef_block[coefBlockIndex + JpegConstants.DCTSIZE * 4] == 0 &&
coef_block[coefBlockIndex + JpegConstants.DCTSIZE * 5] == 0 &&
coef_block[coefBlockIndex + JpegConstants.DCTSIZE * 6] == 0 &&
coef_block[coefBlockIndex + JpegConstants.DCTSIZE * 7] == 0)
{
/* AC terms all zero */
float dcval = FLOAT_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSIZE * 0],
quantTable[quantTableIndex + JpegConstants.DCTSIZE * 0]);
workspace[workspaceIndex + JpegConstants.DCTSIZE * 0] = dcval;
workspace[workspaceIndex + JpegConstants.DCTSIZE * 1] = dcval;
workspace[workspaceIndex + JpegConstants.DCTSIZE * 2] = dcval;
workspace[workspaceIndex + JpegConstants.DCTSIZE * 3] = dcval;
workspace[workspaceIndex + JpegConstants.DCTSIZE * 4] = dcval;
workspace[workspaceIndex + JpegConstants.DCTSIZE * 5] = dcval;
workspace[workspaceIndex + JpegConstants.DCTSIZE * 6] = dcval;
workspace[workspaceIndex + JpegConstants.DCTSIZE * 7] = dcval;
coefBlockIndex++; /* advance pointers to next column */
quantTableIndex++;
workspaceIndex++;
continue;
}
/* Even part */
float tmp0 = FLOAT_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSIZE * 0],
quantTable[quantTableIndex + JpegConstants.DCTSIZE * 0]);
float tmp1 = FLOAT_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSIZE * 2],
quantTable[quantTableIndex + JpegConstants.DCTSIZE * 2]);
float tmp2 = FLOAT_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSIZE * 4],
quantTable[quantTableIndex + JpegConstants.DCTSIZE * 4]);
float tmp3 = FLOAT_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSIZE * 6],
quantTable[quantTableIndex + JpegConstants.DCTSIZE * 6]);
float tmp10 = tmp0 + tmp2; /* phase 3 */
float tmp11 = tmp0 - tmp2;
float tmp13 = tmp1 + tmp3; /* phases 5-3 */
float tmp12 = (tmp1 - tmp3) * 1.414213562f - tmp13; /* 2*c4 */
tmp0 = tmp10 + tmp13; /* phase 2 */
tmp3 = tmp10 - tmp13;
tmp1 = tmp11 + tmp12;
tmp2 = tmp11 - tmp12;
/* Odd part */
float tmp4 = FLOAT_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSIZE * 1],
quantTable[quantTableIndex + JpegConstants.DCTSIZE * 1]);
float tmp5 = FLOAT_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSIZE * 3],
quantTable[quantTableIndex + JpegConstants.DCTSIZE * 3]);
float tmp6 = FLOAT_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSIZE * 5],
quantTable[quantTableIndex + JpegConstants.DCTSIZE * 5]);
float tmp7 = FLOAT_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSIZE * 7],
quantTable[quantTableIndex + JpegConstants.DCTSIZE * 7]);
float z13 = tmp6 + tmp5; /* phase 6 */
float z10 = tmp6 - tmp5;
float z11 = tmp4 + tmp7;
float z12 = tmp4 - tmp7;
tmp7 = z11 + z13; /* phase 5 */
tmp11 = (z11 - z13) * 1.414213562f; /* 2*c4 */
float z5 = (z10 + z12) * 1.847759065f; /* 2*c2 */
tmp10 = z5 - z12 * 1.082392200f; /* 2*(c2-c6) */
tmp12 = z5 - z10 * 2.613125930f; /* 2*(c2+c6) */
tmp6 = tmp12 - tmp7; /* phase 2 */
tmp5 = tmp11 - tmp6;
tmp4 = tmp10 - tmp5;
workspace[workspaceIndex + JpegConstants.DCTSIZE * 0] = tmp0 + tmp7;
workspace[workspaceIndex + JpegConstants.DCTSIZE * 7] = tmp0 - tmp7;
workspace[workspaceIndex + JpegConstants.DCTSIZE * 1] = tmp1 + tmp6;
workspace[workspaceIndex + JpegConstants.DCTSIZE * 6] = tmp1 - tmp6;
workspace[workspaceIndex + JpegConstants.DCTSIZE * 2] = tmp2 + tmp5;
workspace[workspaceIndex + JpegConstants.DCTSIZE * 5] = tmp2 - tmp5;
workspace[workspaceIndex + JpegConstants.DCTSIZE * 3] = tmp3 + tmp4;
workspace[workspaceIndex + JpegConstants.DCTSIZE * 4] = tmp3 - tmp4;
coefBlockIndex++; /* advance pointers to next column */
quantTableIndex++;
workspaceIndex++;
}
/* Pass 2: process rows from work array, store into output array. */
workspaceIndex = 0;
byte[] limit = m_cinfo.m_sample_range_limit;
int limitOffset = m_cinfo.m_sampleRangeLimitOffset - RANGE_SUBSET;
for (int ctr = 0; ctr < JpegConstants.DCTSIZE; ctr++)
{
/* Rows of zeroes can be exploited in the same way as we did with columns.
* However, the column calculation has created many nonzero AC terms, so
* the simplification applies less often (typically 5% to 10% of the time).
* And testing floats for zero is relatively expensive, so we don't bother.
*/
/* Even part */
/* Prepare range-limit and float->int conversion */
float z5 = workspace[workspaceIndex + 0] + (RANGE_CENTER + 0.5f);
float tmp10 = z5 + workspace[workspaceIndex + 4];
float tmp11 = z5 - workspace[workspaceIndex + 4];
float tmp13 = workspace[workspaceIndex + 2] + workspace[workspaceIndex + 6];
float tmp12 = (workspace[workspaceIndex + 2] - workspace[workspaceIndex + 6]) * 1.414213562f - tmp13; /* 2*c4 */
float tmp0 = tmp10 + tmp13;
float tmp3 = tmp10 - tmp13;
float tmp1 = tmp11 + tmp12;
float tmp2 = tmp11 - tmp12;
/* Odd part */
float z13 = workspace[workspaceIndex + 5] + workspace[workspaceIndex + 3];
float z10 = workspace[workspaceIndex + 5] - workspace[workspaceIndex + 3];
float z11 = workspace[workspaceIndex + 1] + workspace[workspaceIndex + 7];
float z12 = workspace[workspaceIndex + 1] - workspace[workspaceIndex + 7];
float tmp7 = z11 + z13; /* phase 5 */
tmp11 = (z11 - z13) * 1.414213562f; /* 2*c4 */
z5 = (z10 + z12) * 1.847759065f; /* 2*c2 */
tmp10 = z5 - z12 * 1.082392200f; /* 2*(c2-c6) */
tmp12 = z5 - z10 * 2.613125930f; /* 2*(c2+c6) */
float tmp6 = tmp12 - tmp7; /* phase 2 */
float tmp5 = tmp11 - tmp6;
float tmp4 = tmp10 - tmp5;
/* Final output stage: float->int conversion and range-limit */
int currentOutRow = output_row + ctr;
m_componentBuffer[currentOutRow][output_col + 0] = limit[limitOffset + (int)(tmp0 + tmp7) & RANGE_MASK];
m_componentBuffer[currentOutRow][output_col + 7] = limit[limitOffset + (int)(tmp0 - tmp7) & RANGE_MASK];
m_componentBuffer[currentOutRow][output_col + 1] = limit[limitOffset + (int)(tmp1 + tmp6) & RANGE_MASK];
m_componentBuffer[currentOutRow][output_col + 6] = limit[limitOffset + (int)(tmp1 - tmp6) & RANGE_MASK];
m_componentBuffer[currentOutRow][output_col + 2] = limit[limitOffset + (int)(tmp2 + tmp5) & RANGE_MASK];
m_componentBuffer[currentOutRow][output_col + 5] = limit[limitOffset + (int)(tmp2 - tmp5) & RANGE_MASK];
m_componentBuffer[currentOutRow][output_col + 3] = limit[limitOffset + (int)(tmp3 + tmp4) & RANGE_MASK];
m_componentBuffer[currentOutRow][output_col + 4] = limit[limitOffset + (int)(tmp3 - tmp4) & RANGE_MASK];
workspaceIndex += JpegConstants.DCTSIZE; /* advance pointer to next row */
}
}