1 | Software usage example |
---|
2 | Contact: Woo-Jin Han, wjhan.han@samsung.com |
---|
3 | |
---|
4 | 1. Installation and compilation |
---|
5 | 1.1 Source tree |
---|
6 | - Root |
---|
7 | |--- bin |
---|
8 | |--- build |
---|
9 | |--- linux |
---|
10 | |--- cfg |
---|
11 | |--- cfp |
---|
12 | |--- doc |
---|
13 | |--- source |
---|
14 | |--- App |
---|
15 | |--- TAppDecoder |
---|
16 | |--- TAppEncoder |
---|
17 | |--- Lib |
---|
18 | |--- TLibCommon |
---|
19 | |--- TLibDecoder |
---|
20 | |--- TLibEncoder |
---|
21 | |--- TLibVideoIO |
---|
22 | |
---|
23 | 1.2 Windows using MS Visual Studio |
---|
24 | - Workspaces of VC6 and VC2008 are included in Root/build directory |
---|
25 | 1.3 Linux |
---|
26 | - Makefile is included in Root/build/linux directory |
---|
27 | |
---|
28 | 2. Encoder option |
---|
29 | 2.1 Parameters |
---|
30 | - TAppEncoder.exe -c config.cfg [options] |
---|
31 | - Options |
---|
32 | -c configuration file name |
---|
33 | -i original YUV input file name |
---|
34 | -o decoded YUV output file name |
---|
35 | -b bitstream file name |
---|
36 | -m dQP file name |
---|
37 | -f number of frames to be encoded (default EOS) |
---|
38 | -q Qp value, if value is float, QP is switched once during encoding |
---|
39 | -g GOP size of temporal structure |
---|
40 | -rg GOP size of hierarchical QP assignment |
---|
41 | -s max CU size |
---|
42 | -h CU depth |
---|
43 | -r Number of reference (P) |
---|
44 | -rb0 Number of reference (B_L0) |
---|
45 | -rb1 Number of reference (B_L1) |
---|
46 | -ip intra period in frames, (-1: only first frame) |
---|
47 | -ldm recommended low-delay setting (with LDC), (0=slow sequence, 1=fast sequence) |
---|
48 | -d max dQp offset for block |
---|
49 | -dqr max dQp offset for slice |
---|
50 | -t max transform size |
---|
51 | -ltd min transform depth |
---|
52 | -utd max transform depth |
---|
53 | -v additional reference for weighted prediction (w: scale+offset, o: offset) |
---|
54 | -tap number of interpolation filter taps (luma) |
---|
55 | -tq0 QP offset of temporal layer 0 |
---|
56 | -tq1 QP offset of temporal layer 1 |
---|
57 | -tq2 QP offset of temporal layer 2 |
---|
58 | -tq3 QP offset of temporal layer 3 |
---|
59 | -pdx horizontal source padding size |
---|
60 | -pdy vertical source padding size |
---|
61 | -1/0 <name>: turn on/off <name> |
---|
62 | <name> = ALF - adaptive loop filter |
---|
63 | IBD - bit-depth increasement |
---|
64 | GPB - generalized B instead of P in low-delay mode |
---|
65 | HAD - hadamard ME for fractional-pel |
---|
66 | SRD - SBAC based RD estimation |
---|
67 | RDQ - RDOQ |
---|
68 | LDC - low-delay mode |
---|
69 | NRF - non-reference frame marking in last layer |
---|
70 | BQP - hier-P style QP assignment in low-delay mode |
---|
71 | PAD - automatic source padding of multiple of 16 |
---|
72 | QBO - skip refers highest quality picture |
---|
73 | ASR - adaptive motion search range |
---|
74 | 2.2 Config file (example) |
---|
75 | |
---|
76 | #======== File I/O =============== |
---|
77 | InputFile : d:\test\origcfp\RaceHorses_416x240_30.yuv |
---|
78 | BitstreamFile : RaceHorses.bin |
---|
79 | ReconFile : RaceHorses_enc.yuv |
---|
80 | FrameRate : 30 # Frame Rate per second |
---|
81 | FrameSkip : 0 # Number of frames to be skipped in input |
---|
82 | SourceWidth : 416 # Input frame width |
---|
83 | SourceHeight : 240 # Input frame height |
---|
84 | FrameToBeEncoded : 9 # Number of frames to be coded |
---|
85 | |
---|
86 | #======== Coding Structure ======= |
---|
87 | IntraPeriod : 32 # Period of I-Frame ( -1 = only first) |
---|
88 | GOPSize : 8 # GOP Size (number of B slice = GOPSize-1) |
---|
89 | NumOfReference : 1 # Number of reference frames |
---|
90 | NumOfReferenceB_L0 : 1 # Number of reference frames for L0 for B-slices |
---|
91 | NumOfReferenceB_L1 : 1 # Number of reference frames for L1 for B-slices |
---|
92 | QP : 32 # Quantization parameter(0-51) |
---|
93 | |
---|
94 | #======== New Structure =========== |
---|
95 | MaxCUWidth : 128 # Maximum Coding Unit size in width |
---|
96 | MaxCUHeight : 128 # Maximum Coding Unit size in Height |
---|
97 | MaxPartitionDepth : 5 # Maximum partition depth. ( minimum width = MaxWidth >> (MaxPartitionDepth-1) ) |
---|
98 | |
---|
99 | #=========== B Slice =================== |
---|
100 | HierarchicalCoding : 1 # B hierarchical coding ON/OFF (if OFF, no reference B is used) |
---|
101 | |
---|
102 | #=========== Entropy Coding ============ |
---|
103 | SymbolMode : 1 # CAVLC: 0, SBAC: 1, only 1 supported, CAVLC implementation is not completed |
---|
104 | |
---|
105 | #=========== Loop/Deblock Filter ======= |
---|
106 | LoopFilterDisable : 0 # Disable loop filter in slice header (0=Filter, 1=No Filter) |
---|
107 | LoopFilterAlphaC0Offset : 0 # Range: -26 ~ 26 |
---|
108 | LoopFilterBetaOffset : 0 # Range: -26 ~ 26 |
---|
109 | |
---|
110 | #=========== Motion search ============= |
---|
111 | FastSearch : 1 # 0:Full search 1:Diamond 2:PMVFAST(not supported) |
---|
112 | SearchRange : 128 # (0: Search range is a Full frame) |
---|
113 | MaxDeltaQP : 0 # Absolute delta QP (1:default) |
---|
114 | |
---|
115 | 2.3 Typical example |
---|
116 | |
---|
117 | Example 1) TAppEncoder.exe -c test.cfg -q 32 -g 8 -f 9 -s 64 -h 4 |
---|
118 | (Hier-B) -> QP 32, hierarchical-B GOP 8, 9 frames, 64x64-8x8 CU (~4x4 PU) |
---|
119 | |
---|
120 | Example 2) TAppEncoder.exe -c test.cfg -q 32 -g 4 -f 9 -s 64 -h 4 -1 LDC |
---|
121 | (Hier-P) -> QP 32, hierarchical-P GOP 4, 9 frames, 64x64-8x8 CU (~4x4 PU) |
---|
122 | |
---|
123 | Example 3) TAppEncoder.exe -c test.cfg -q 32 -g 1 -f 9 -s 64 -h 4 -1 LDC |
---|
124 | (IPPP) -> QP 32, hierarchical-P GOP 4, 9 frames, 64x64-8x8 CU (~4x4 PU) |
---|
125 | |
---|
126 | 3. Decoder option |
---|
127 | 3.1 Parameters |
---|
128 | - TAppDecoder.exe -b test.bin -o test.yuv |
---|
129 | . Decode test.bin and make test.yuv as the reconstructed YUV |
---|
130 | - TAppDecoder.exe -b test.bin |
---|
131 | . Decode test.bin but YUV writing is skipped |
---|
132 | |
---|
133 | 4. Contact point |
---|
134 | |
---|
135 | =========================================================== |
---|
136 | Woo-Jin Han, Principal Engineer |
---|
137 | (wjhan.han@samsung.com) |
---|
138 | |
---|
139 | M/M Platform Lab |
---|
140 | |
---|
141 | Digital Media & Communications R&D Center |
---|
142 | Digital Media & Communications Business |
---|
143 | Samsung Electronics Co. Ltd. |
---|
144 | |
---|
145 | Korea phone) |
---|
146 | +82-31-279-8831 (office), +82-10-3329-6393 (cellular) |
---|
147 | =========================================================== |
---|