Note that the functions with _IP2R descriptor use the following structures:
typedef struct _IppiMBReconstructHigh_32s16u
{
Ipp32s** ppSrcDstCoeff;
Ipp16u* pSrcDstPlane;
Ipp32s srcDstStep;
Ipp32u cbp;
Ipp32s qp;
Ipp16s* pQuantTable;
Ipp32s bypassFlag;
Ipp32s bitDepth;
} IppiReconstructHighMB_32s16u;
typedef struct _IppiReconstructHighMB_16s8u
{
Ipp16s** ppSrcDstCoeff;
Ipp8u* pSrcDstPlane;
Ipp32s srcDstStep;
Ipp32u cbp;
Ipp32s qp;
Ipp16s* pQuantTable;
Ipp32s bypassFlag;
} IppiReconstructHighMB_16s8u;
with parameters:
ppSrcDstCoeff |
Pointer to the residual coefficients; pointer is updated by the function. |
pSrcDstPlane |
Pointer to the macroblock that is reconstructed in the current plane. |
srcDstStep |
Distance in items between starts of the consecutive lines in the source/destination images. |
cbp |
Coded block pattern.
The value of N depends on a function. For Luma, N is 16, for chroma - N is 4, and for Chroma422 - N is 8. Sizes of DC/AC blocks depend on a function. |
qp |
Quantizer with range depending on the function (Luma/Chroma). For _32s16u_IP2R functions, the range depends on bitDepth. It should be [0 .. 39 + (6*((bitDepth) - 8))] for chroma and [0 .. 51 + (6*((bitDepth) - 8))] for luma. For _16s8u_IP2R functions, the range equals [0..39] for chroma and [0..51] for luma. |
pQuantTable |
Quantization table. Pointer to the quantization table for plane (LevelScale( qP%6, i, j) in [ITUH264]) |
bypassFlag |
Flag enabling lossless coding. Lossless coding is enabled only if bypassFlag is 1 and qp is 0. (See qpprime_y_zero_transform_bypass_flag in [ITUH264] |
bitDepth |
Number of bits of the plane sample with range [8.. 14]. |
Copyright © 2000 - 2010, Intel Corporation. All rights reserved.