Coverage Report

Created: 2022-08-24 06:15

/src/x265/source/common/quant.h
Line
Count
Source
1
/*****************************************************************************
2
 * Copyright (C) 2013-2020 MulticoreWare, Inc
3
 *
4
 * Authors: Steve Borho <steve@borho.org>
5
 *          Min Chen <chenm003@163.com>
6
 *
7
 * This program is free software; you can redistribute it and/or modify
8
 * it under the terms of the GNU General Public License as published by
9
 * the Free Software Foundation; either version 2 of the License, or
10
 * (at your option) any later version.
11
 *
12
 * This program is distributed in the hope that it will be useful,
13
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
15
 * GNU General Public License for more details.
16
 *
17
 * You should have received a copy of the GNU General Public License
18
 * along with this program; if not, write to the Free Software
19
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
20
 *
21
 * This program is also available under a commercial proprietary license.
22
 * For more information, contact us at license @ x265.com.
23
 *****************************************************************************/
24
25
#ifndef X265_QUANT_H
26
#define X265_QUANT_H
27
28
#include "common.h"
29
#include "scalinglist.h"
30
#include "contexts.h"
31
32
namespace X265_NS {
33
// private namespace
34
35
class CUData;
36
class Entropy;
37
struct TUEntropyCodingParameters;
38
39
struct QpParam
40
{
41
    int rem;
42
    int per;
43
    int qp;
44
    int64_t lambda2; /* FIX8 */
45
    int32_t lambda;  /* FIX8, dynamic range is 18-bits in Main and 20-bits in Main10 */
46
47
67.9k
    QpParam() : qp(MAX_INT) {}
48
49
    void setQpParam(int qpScaled)
50
84.5k
    {
51
84.5k
        if (qp != qpScaled)
52
4.81k
        {
53
4.81k
            rem = qpScaled % 6;
54
4.81k
            per = qpScaled / 6;
55
4.81k
            qp  = qpScaled;
56
4.81k
            lambda2 = (int64_t)(x265_lambda2_tab[qp - QP_BD_OFFSET] * 256. + 0.5);
57
4.81k
            lambda  = (int32_t)(x265_lambda_tab[qp - QP_BD_OFFSET] * 256. + 0.5);
58
4.81k
            X265_CHECK((x265_lambda_tab[qp - QP_BD_OFFSET] * 256. + 0.5) < (double)MAX_INT, "x265_lambda_tab[] value too large\n");
59
4.81k
        }
60
84.5k
    }
61
};
62
63
// NOTE: MUST be 16-byte aligned for asm code
64
struct NoiseReduction
65
{
66
    /* 0 = luma 4x4,   1 = luma 8x8,   2 = luma 16x16,   3 = luma 32x32
67
     * 4 = chroma 4x4, 5 = chroma 8x8, 6 = chroma 16x16, 7 = chroma 32x32
68
     * Intra 0..7 - Inter 8..15 */
69
    ALIGN_VAR_16(uint32_t, nrResidualSum[MAX_NUM_TR_CATEGORIES][MAX_NUM_TR_COEFFS]);
70
    uint32_t nrCount[MAX_NUM_TR_CATEGORIES];
71
    uint16_t nrOffsetDenoise[MAX_NUM_TR_CATEGORIES][MAX_NUM_TR_COEFFS];
72
    uint16_t (*offset)[MAX_NUM_TR_COEFFS];
73
    uint32_t (*residualSum)[MAX_NUM_TR_COEFFS];
74
    uint32_t *count;
75
};
76
77
class Quant
78
{
79
protected:
80
81
    const ScalingList* m_scalingList;
82
    Entropy*           m_entropyCoder;
83
84
    QpParam            m_qpParam[3];
85
86
    int                m_rdoqLevel;
87
    int32_t            m_psyRdoqScale;  // dynamic range [0,50] * 256 = 14-bits
88
    int16_t*           m_resiDctCoeff;
89
    int16_t*           m_fencDctCoeff;
90
    int16_t*           m_fencShortBuf;
91
92
    enum { IEP_RATE = 32768 }; /* FIX15 cost of an equal probable bit */
93
94
public:
95
96
    NoiseReduction*    m_nr;
97
    NoiseReduction*    m_frameNr; // Array of NR structures, one for each frameEncoder
98
99
    Quant();
100
    ~Quant();
101
102
    /* one-time setup */
103
    bool init(double psyScale, const ScalingList& scalingList, Entropy& entropy);
104
    bool allocNoiseReduction(const x265_param& param);
105
106
    /* CU setup */
107
    void setQPforQuant(const CUData& ctu, int qp);
108
109
    uint32_t transformNxN(const CUData& cu, const pixel* fenc, uint32_t fencStride, const int16_t* residual, uint32_t resiStride, coeff_t* coeff,
110
                          uint32_t log2TrSize, TextType ttype, uint32_t absPartIdx, bool useTransformSkip);
111
112
    void invtransformNxN(const CUData& cu, int16_t* residual, uint32_t resiStride, const coeff_t* coeff,
113
                         uint32_t log2TrSize, TextType ttype, bool bIntra, bool useTransformSkip, uint32_t numSig);
114
    uint64_t ssimDistortion(const CUData& cu, const pixel* fenc, uint32_t fStride, const pixel* recon, intptr_t rstride,
115
                            uint32_t log2TrSize, TextType ttype, uint32_t absPartIdx);
116
117
    /* Pattern decision for context derivation process of significant_coeff_flag */
118
    static uint32_t calcPatternSigCtx(uint64_t sigCoeffGroupFlag64, uint32_t cgPosX, uint32_t cgPosY, uint32_t cgBlkPos, uint32_t trSizeCG)
119
191k
    {
120
191k
        if (trSizeCG == 1)
121
33.4k
            return 0;
122
123
158k
        X265_CHECK(trSizeCG <= 8, "transform CG is too large\n");
124
158k
        X265_CHECK(cgBlkPos < 64, "cgBlkPos is too large\n");
125
        // NOTE: cgBlkPos+1 may more than 63, it is invalid for shift,
126
        //       but in this case, both cgPosX and cgPosY equal to (trSizeCG - 1),
127
        //       the sigRight and sigLower will clear value to zero, the final result will be correct
128
158k
        const uint32_t sigPos = (uint32_t)(sigCoeffGroupFlag64 >> (cgBlkPos + 1)); // just need lowest 7-bits valid
129
130
        // TODO: instruction BT is faster, but _bittest64 still generate instruction 'BT m, r' in VS2012
131
158k
        const uint32_t sigRight = (cgPosX != (trSizeCG - 1)) & sigPos;
132
158k
        const uint32_t sigLower = (cgPosY != (trSizeCG - 1)) & (sigPos >> (trSizeCG - 1));
133
158k
        return sigRight + sigLower * 2;
134
191k
    }
135
136
    /* Context derivation process of coeff_abs_significant_flag */
137
    static uint32_t getSigCoeffGroupCtxInc(uint64_t cgGroupMask, uint32_t cgPosX, uint32_t cgPosY, uint32_t cgBlkPos, uint32_t trSizeCG)
138
110k
    {
139
110k
        X265_CHECK(cgBlkPos < 64, "cgBlkPos is too large\n");
140
        // NOTE: unsafe shift operator, see NOTE in calcPatternSigCtx
141
110k
        const uint32_t sigPos = (uint32_t)(cgGroupMask >> (cgBlkPos + 1)); // just need lowest 8-bits valid
142
110k
        const uint32_t sigRight = (cgPosX != (trSizeCG - 1)) & sigPos;
143
110k
        const uint32_t sigLower = (cgPosY != (trSizeCG - 1)) & (sigPos >> (trSizeCG - 1));
144
145
110k
        return (sigRight | sigLower);
146
110k
    }
147
148
    /* static methods shared with entropy.cpp */
149
    static uint32_t getSigCtxInc(uint32_t patternSigCtx, uint32_t log2TrSize, uint32_t trSize, uint32_t blkPos, bool bIsLuma, uint32_t firstSignificanceMapContext);
150
151
protected:
152
153
    void setChromaQP(int qpin, TextType ttype, int chFmt);
154
155
    uint32_t signBitHidingHDQ(int16_t* qcoeff, int32_t* deltaU, uint32_t numSig, const TUEntropyCodingParameters &codingParameters, uint32_t log2TrSize);
156
157
    template<uint32_t log2TrSize>
158
    uint32_t rdoQuant(const CUData& cu, int16_t* dstCoeff, TextType ttype, uint32_t absPartIdx, bool usePsy);
159
160
public:
161
    typedef uint32_t (Quant::*rdoQuant_t)(const CUData& cu, int16_t* dstCoeff, TextType ttype, uint32_t absPartIdx, bool usePsy);
162
163
private:
164
    static rdoQuant_t rdoQuant_func[NUM_CU_DEPTH];
165
};
166
}
167
168
#endif // ifndef X265_QUANT_H