现在的位置: 首页 > 综合 > 正文

HEVC配置文件解读

2013年10月14日 ⁄ 综合 ⁄ 共 5856字 ⁄ 字号 评论关闭

#======== File I/O =====================

BitstreamFile[I1]      
          :  F:\\HEVCres\\BasketballDrill_832x480_50_str.bin

ReconFile [I2]        
          :  F:\\HEVCres\\BasketballDrill_832x480_50_rec.yuv

#======== Unit definition ================

MaxCUWidth               : 64     # Maximum coding unit width in pixel

MaxCUHeight               : 64[I3]    
# Maximum coding unit height in pixel

MaxPartitionDepth            : 4[I4]   
   # Maximum coding unit depth

QuadtreeTULog2MaxSize      : 5     # Log2 of maximum transform size for

                                  # quadtree-based TU coding (2...6)

QuadtreeTULog2MinSize      : 2     # Log2 of minimum transform size for

                                  # quadtree-based TU coding (2...6)[I5] 

QuadtreeTUMaxDepthInter     : 3[I6] 

QuadtreeTUMaxDepthIntra     : 3

 

 

#======== Coding Structure =============

IntraPeriod               : 1           # Period of I-Frame
( -1 = only first)

DecodingRefreshType      : 0           # Random Accesss[I7] 
0:none, 1:CDR, 2:IDR

GOPSize                 : 1           # GOP Size (number of B slice = GOPSize-1)

#Type POC QPoffset
QPfactor tcOffsetDiv2
betaOffsetDiv2
temporal_id

#ref_pics_active

#ref_pics reference pictures

#=========== Motion Search =============

FastSearch [I8]        
: 1           # 0:Full search  1:TZ search

SearchRange       : 64          #
(0: Search range is a Full frame
)

HadamardME       : 1          # Use of hadamard measure for fractional ME[I9] 

FEN              : 1           #
Fast encoder decision

FDM              : 1          #
Fast Decision for Merge RD cost

 

#======== Quantization =============

QP               : 32     # Quantization parameter(0-51)[I10] 

MaxDeltaQP       : 0      # CU-based
multi-QP optimization[I11] 

MaxCuDQPDepth   : 0      #
Max depth of a minimum CuDQP
for sub-LCU-level delta QP

DeltaQpRD        : 0      # Slice-based multi-QP optimization[I12] 

RDOQ            : 1      #
RDOQ
[I13] 

RDOQTS         : 1       # RDOQ for
transform skip

#=========== Deblock Filter ============

DeblockingFilterControlPresent   : 0   # Dbl control params present
[I14] (0=not
present, 1=present)

LoopFilterOffsetInPPS[I15]      
    : 0   # Dbl params:

0=varying params in SliceHeader,   param = base_param + GOP_offset_param;

1=constant params in PPS,         param = base_param

LoopFilterDisable              : 0     # Disable deblocking filter (0=Filter, 1=No Filter)

LoopFilterBetaOffset_div2       : 0     # base_param: -13 ~ 13

LoopFilterTcOffset_div2         : 0     # base_param: -13 ~ 13

#=========== Misc. ============

InternalBitDepth      : 8      # codec operating bit-depth[I16] 

 

#=========== Coding Tools =================

SAO                : 1   # Sample adaptive offset[I17] 
(0: OFF, 1: ON)

AMP                : 1   # Asymmetric motion partitions[I18] 
(0: OFF, 1: ON)

TransformSkip        : 1   # Transform skipping[I19] 
(0: OFF, 1: ON)

TransformSkipFast     : 1   # Fast Transform skipping[I20] 
(0: OFF, 1: ON)

SAOLcuBoundary     : 0   # SAOLcuBoundary using non-deblocked pixels[I21] 
(0: OFF, 1: ON)

#============ Slices ================

SliceMode        : 0         

                 # 0: Disable all slice options.

                 # 1: Enforce maximum number of LCU in an slice,

                 # 2: Enforce maximum number of bytes in an 'slice'

                 # 3: Enforce maximum number of tiles in a slice

SliceArgument    : 1500      

               # Argument for 'SliceMode'.

               # If SliceMode==1 it represents max.
SliceGranularity-sized blocks per slice.

               # If SliceMode==2 it represents max. bytes per slice.

               # If SliceMode==3 it represents max. tiles per slice.

LFCrossSliceBoundaryFlag   : 1          

         # In-loop filtering, including ALF and DB, is across or not across slice boundary.[I22] 

         # 0:not across, 1: across

#============ PCM ================

PCMEnabledFlag        : 0       # 0: No PCM mode

PCMLog2MaxSize       : 5[I23]       
# Log2 of maximum PCM block size.

PCMLog2MinSize       : 3[I24]       
# Log2 of minimum PCM block size.

PCMInputBitDepthFlag   : 1 

# 0: PCM bit-depth is internal bit-depth.
1: PCM bit-depth is input bit-depth.

PCMFilterDisableFlag    : 0 

# 0: Enable loop filtering on I_PCM samples. 1: Disable loop filtering on I_PCM samples.

#============ Tiles ================

UniformSpacingIdc          : 0  

# 0: the column boundaries are indicated by ColumnWidth array[I25] ,
the row boundaries are indicated by RowHeight array

# 1: the column and row boundaries are distributed uniformly

 

NumTileColumnsMinus1    : 0     #
Number of columns in a picture minus 1

ColumnWidthArray        : 2 3    

# Array containing ColumnWidth values in units of LCU (from left to right in picture)

 

NumTileRowsMinus1      : 0      #
Number of rows in a picture minus 1

RowHeightArray          : 2   

# Array containing RowHeight values in units of LCU (from top to bottom in picture)

LFCrossTileBoundaryFlag   : 1     # In-loop filtering is across or not across tile boundary.

                               # 0:not across, 1: across

#============ WaveFront[I26] 
================

WaveFrontSynchro[I27]     
    : 0   

#  0:  No WaveFront synchronisation (WaveFrontSubstreams must be 1 in this
case
).

# >0:  WaveFront synchronises with the LCU above and to the right by this many LCUs.

#=========== Quantization Matrix =================

ScalingList             : 0              #
ScalingList 0 : off, 1 : default, 2 : file read

ScalingListFile          : scaling_list.txt  

# Scaling[I28] 
List file name. If file is not exist, use Default Matrix.

#============ Lossless ================

TransquantBypassEnableFlag     : 0   #
Value of PPS flag[I29] .

CUTransquantBypassFlagValue   : 0

#Constant lossless-value signaling per CU, if TransquantBypassEnableFlag is 1.

                                                      

### DO NOT ADD ANYTHING BELOW THIS LINE ###

### DO NOT DELETE THE EMPTY LINE BELOW ###


 [I1]编码后获得的码流

 [I2]重建文件,即获得的编码视频文件

 [I3]最大编码单元TCU(LCU)

 [I4]最大深度4,深度范围[0,1,2,3]其中0表示64x64,1表示32x32,2表示16x16,3表示8x8.

 [I5]log2为底4的对数2;

log2为底8的对数3;

log2为底16的对数4;

log2为底32的对数5;

log2为底64的对数6;

 [I6]3表示8x8

 [I7]解码更新类型,即随机接入类型

 [I8]快速搜索类型,0:全搜索,1:TZ搜索

 [I9]分数(像素位置)运动估计

 [I10]量化参数范围0-51

 [I11]QP最优化

 [I12]基于片的多QP最优化

 [I13]率失真优化Q

 [I14]Dbl控制是否使用去块滤波器

 [I15]PPS: picture parameter
set

 [I16]编解码控制比特深度,这个应该表示主要档次,即MAIN

 [I17]采样自适应偏移

 [I18]非对称运动分割

 [I19]变换过程跳过

 [I20]快速变换过程跳过

 [I21]SAOLCU边界像素用未经过去块滤波操作的像素值表示

 [I22]ALFDB两个滤波器在片边界是否相交

 [I23]PCM块最大为32x32

 [I24]PCM块最小为8x8

 [I25]Tiles的列边界由列宽数组决定

 [I26]WPP并行处理同步可以参考overview

 [I27]WPP并行处理同步

 [I28]这里用的是默认配置

 [I29]PPS: Picture parameter
set

抱歉!评论已关闭.