h264编解码器知识点

缩写：
1．DTS, PTS
对于一个ES来说，他又许多I,P,B帧，而P, B帧都是以I，P帧作为参考。
由于B帧是前向后向参考，因此要对B帧作decode的话，
就必须先decode该B帧后面的P,或者I帧，
于是，decode的时间与帧的真正的present的时间就不一致了，
按照DTS一次对各个帧进行decode，

然后再按照PTS对各个帧进行展现

2.转自http://legendbaby.blogbus.com/logs/22585281.html

Z=ROUND(Y/Qstep)其中Z为量化值，Y为输入系数值，Qstep为量化步长，量化步长有52种，每个宏块的量化步长由量化参数QP决定，QP每增加1，量化步长增加12.5%。

 H.264将一部分变换系数和量化和在一起实现，即在Zij=Yij/Qstep的基础上进行，变成Zij=(Yij*PF/Qstep),这里PF就是变换时提取出的系数a2,ab/2,b2/4......

实 际实现时，用移位的方法实现除法，即Zij=Yij.MF/2qbit;这样就有MF/2qbit=PF/Qstep，则MF=2qbitPF /Qstep，qbit=15+(QP/6),所以有QP相差6，MF增加一倍。QP=4时，Qstep=1，求出对应ij的MF，就是量化矩阵了。

3.
H.264 基本语法结构简介

H.264 相对以往的视频压缩编码标准来说,在语法结构上有很大的改变,其中最大的改 变体现在以下两个方面:

取消帧级语法单元

H.264 语法中没有 frame_header 之类的语法单元,帧信息全部放在 slice_header、 SPS 和 PPS 中,这样可增强 slice 单元解码的独立性,提高码流的抗丢包、抗误码 能力。但由于一帧图像可以对应多个 slice,因此解码器无法通过解析类似 frame_header 的语法来识别码流中的一帧数据。

引入 SPS、PPS 等参数集概念

• ?  将一个视频序列(从 IDR 帧开始到下一个 IDR 帧之前的数据称为一个视频序

  列)全部图像的共同特征抽取出来,放在 SPS 语法单元中。

• ?  将各个图像的典型特征抽取出来,放在 PPS 语法单元中。

• ?  只有视频序列之间才能切换 SPS,即只有 IDR 帧的第一个 slice 可以切换 SPS。

• ?  只有图像之间才能切换 PPS,即只有每帧图像的第一个 slice 才能切换 PPS。

  从宏观上来说,SPS、PPS、IDR 帧(包含一个或多个 I-Slice)、P 帧(包含一个或多个 P-Slice)、B 帧(包含一个或多个 B-Slice)共同构成典型的 H.264 码流结构。除上述典 型语法结构外,为了方便传输私有信息,H.264 还定义了 SEI 语法结构。除非编码器和 解码器进行特定的语法协商,解码器一般不对 SEI 包进行解析 

4.


5.每个 SPS 或者 PPS 仅对应一个 nal    


6.解码视频通道切换   

一般来说,在视频点播或者视频通道切换时,仅搜索 IDR 帧送给解码器是不够的,还 必须将对应的 SPS 和 PPS 送给解码器。H.264 并没有要求 SPS、PPS 以及 IDR 帧捆绑 传输,只是规定一定要在 IDR 帧之前传输相应的 SPS 和 PPS。因此,最严谨的做法是 从视频码流的起始位置开始搜索,将获取的全部 SPS 和 PPS 依次送给解码器,直到搜 索到一个 IDR 帧。

为了降低应用的复杂性,在视频点播或者实时编解码应用场景中,编码器一般连续传 递 SPS、PPS 和 IDR 帧,即严格在 IDR 帧之前传递当前视频序列所需要的 SPS 参数和 若干个 PPS 参数。Hi3507 编码器遵循这一准则,因此可以在每个 IDR 帧的第一个 nalu 前面找到 4 个尺寸较小的 nalu,它们即为当前视频序列解码所需要的全部参数,其中 第一个 nalu 为 SPS。简单地说,对于 Hi3507 码流,只要找到 SPS,并从此位置开始送 给解码器解码,那么一定可以得到完整的视频图像。

在视频输入通道不断切换的应用场景中(如单通道解码器用时间片的方式点播多个编 码通道码流的场景),需要不断地切换解码器的输入视频源通道。基于上述描述的解码 完整性原则,切换不是从任意位置开始的。也就是说,如果解码器需要切换到某个视 频通道,那么必须从获取这个视频通道的 SPS 的位置开始切换,否则在切换点,视频 图像会出现异常。 

7.流媒体网络传输的方式   

基于 MPEG-4 的方案在网络传输时通常采用以下两种方案:

•  用尺寸小于网络层 MTU 的定长数据段来切分线性的原始码流。

•  直接传输增加私有格式头的一整帧数据,网络层对整帧数据进行切分,而应用层

  则忽略由于网络拥塞而丢弃的包。
  H.264 提出比 MPEG-4 更优秀的网络传输抗误码解决方案。H.264 包含 VCL 和 NAL,

  VCL 和 NAL 的特点如下:

•  VCL 包括核心压缩引擎和块/宏块/slice 的语法级别定义,应该尽量将 VCL 独

  立于网络之外。

•  NAL 将 VCL 产生的比特字符串适配到各种各样的网络环境中。NAL 覆盖所有

  slice 级以上的语法级别,并包含以下机制:

? 提供每一个 slice 解码时所需要的参数数据。 ? 预防起始码冲突。

• ?  支持 SEI。

• ?  提供将编码 slice 的比特字符串在基于比特流的网络上进行传送。

  H.264 将 NAL 与 VCL 分离的主要原因如下:

•  定义一个 VCL 信号处理与 NAL 传输的接口,这样允许 VCL 和 NAL 工作在完全

  不同的处理器平台上。

•  VCL 和 NAL 都被设计成适应于异质传输环境,当网络环境不同时,网关不需要

  对 VCL 比特流进行重构和重编码。 

IP 网络的类型可分为以下三种:

•  不可控 IP 网络(如 Internet)。

•  可控 IP 网络(如广域网)。

•  无线 IP 网络(如 3G 网络)。

  这三种 IP 网络有不同的 MTU、比特出错概率和 TCP 使用标记。两个 IP 节点之间的 MTU 是动态变化的,通常假定有线 IP 网络的 MTU 为 1.5KB,无线 IP 网络 MTU 的范 围从 100byte 到 500byte。

  在以 UDP 发包时建议以 nalu 为基本单元,这样便于在应用层进行封装。推荐使用 RTP 进行组包传输,一个 RTP 分组里放一个 nalu,将 nalu(包括同步头)放入 RTP 的载荷 中,并设置 RTP 的头信息。由于包传送的路径不同,接收端需要重新对 slice 分组进行 排序,RTP 包含的次序信息可以用来解决这一问题。

  不要随意丢弃尺寸大于 MTU 的 UDP 分组,可直接发送,由网络驱动层进行拆包和组 包,对应用层不会产生影响。即使网络出现 UDP 丢包,丢弃也是一个完整的 nalu。在 H.264 中,nalu 是基于 VCL slice 的网络层封装,每个 slice 对应于多个宏块(不同于 MPEG-4)。H.264 在编码时不同 slice 的宏块之间语法是非相关的,丢掉少量的 slice 不 会对其他 slice 的解码产生影响。

  H.264 的应用越来越广泛,在实际的应用方案中将存在 MPEG-4 与 H.264 并存的情 形。为了解码端的系统稳定,避免软件崩溃,建议以 nalu 为基本单元放入分组包里传 送,这样可以做到 H.264 与 MPEG-4 互不干扰,稳定并存于一套系统中 

解码前的丢包策略

0. 播放器在下列情况下需要进行丢包处理:

•  计算机 CPU 负荷过重或者网络抖动等原因引起播放器前端码流缓冲区即将溢出。

•  应用层检测到当前网络丢包现象严重。

•  播放器调度需要进行丢包处理。

   目前通用的主动丢包处理方法如下: 

•  尽可能将收到的所有码流信息送至解码器,不要轻易丢弃已经收到的码流信息。

•  如果播放器在上述几种情况下需要主动丢包,则播放器应从当前帧开始丢弃,一 直丢弃到下一个 IDR 帧。 

6.

  4 Infile can be raw (in which case resolution is required),
  5   or YUV4MPEG (*.y4m),
  6   or Avisynth if compiled with support (no).
  7   or libav* formats if compiled with lavf support (no) or ffms support (no).
  8 Outfile type is selected by filename:
  9  .264 -> Raw bytestream
 10  .mkv -> Matroska
 11  .flv -> Flash Video
 12  .mp4 -> MP4 if compiled with GPAC support (yes)
 13 Output bit depth: 8 (configured at compile time)
 21 Example usage:
 23       Constant quality mode:
 24             x264 --crf 24 -o

 25

crf

--crf 23 （默认）

一种根据片子质量自动分配码率的vbr码率控制方式。一遍编码，如果对码率没要求请尽量使用crf模式。

可用的值从1到51，越小编码质量越好，码率越高。一般使用16到24，可以为浮点。

crf并不是恒定质量的方式，同一片子同一crf值，其他参数不同可能码率和质量能差比较大，不同的片子之间就更没有可比性了。

CRF压制算法设置

crf就是constant ratefactor，就是保证“一定质量”，智能分配码率。智能分配码率包含两个意思：

同 一帧内分配码率。就是在细节的地方分配更多的字节（bits）。实际效果举例：前景细节保留完整，背景压缩得很厉害。帧间分配码率。crf会智能分析哪些 是重要帧，哪些是次要帧。

重要帧会得到更多的字节。实际效果举例：帧排列：清晰-模糊-模糊-清晰。但是给人的感觉会是比较清晰的，因为看电影的时候一般 注意焦点，不会去辨认背景。次要帧一般都一闪而过，也不容易注意到。

x264参数设置：

main里面的crf调节是对上面提到的“一定质量”的调节。值18基本为无损，19-21.5为高质量，22-26为中等质量。

在rate control里面还有一个adaptive quantizers，是对智能分配码率的调节。数值越小，智能分配越极端，就是说会把更多的码率给细节和关键帧。

大片颜色区域和次要帧会压缩得更厉害。

这两个参数是crf最重要的参数。

  26       Two-pass with a bitrate of 1000kbps:
 27             x264 --pass 1 --bitrate 1000 -o

 28             x264 --pass 2 --bitrate 1000 -o

bitrate=

Sets the average bitrate to be used in kbits/second (default: off). Since local bitrate may vary, this average may be inaccurate for very short videos (see ratetol). Constant bitrate can be achieved by combining this with vbv_maxrate, at significant reduction in quality.

 29
 30       Lossless:
 31             x264 --qp 0 -o

 32

x264有4种码率控制方式，分别是1pass bitrate、crf、qp和2pass bitrate。其中2pass bitrate包含pass bitrate。

1pass bitrate和qp（恒定量化值）一般不推荐使用。

  33       Maximum PSNR at the cost of speed and visual quality:
 34             x264 --preset placebo --tune psnr -o

 36       Constant bitrate at 1000kbps with a 2 second-buffer:
 37             x264 --vbv-bufsize 2000 --bitrate 1000 -o

 39 Presets:
 41       --profile       Force the limits of an H.264 profile
 42                                   Overrides all settings.
 43                                   - baseline:
 44                                     --no-8x8dct --bframes 0 --no-cabac
 45                                     --cqm flat --weightp 0

46                                     No interlaced.
 47                                     No lossless.
 48                                   - main:
 49                                     --no-8x8dct --cqm flat
 50                                     No lossless.
 51                                   - high:
 52                                     No lossless.
 53                                   - high10:
 54                                     No lossless.
 55                                     Support for bit depth 8-10.
 56                                   - high422:
 57                                     No lossless.
 58                                     Support for bit depth 8-10.
 59                                     Support for 4:2:0/4:2:2 chroma subsampling.
 60                                   - high444:
 61                                     Support for bit depth 8-10.
 62                                     Support for 4:2:0/4:2:2/4:4:4 chroma subsampling.
 63       --preset        Use a preset to select encoding settings [medium]
 64                                   Overridden by user settings.
 65                                   - ultrafast:
 66                                     --no-8x8dct --aq-mode 0 --b-adapt 0
 67                                     --bframes 0 --no-cabac --no-deblock
 68                                     --no-mbtree --me dia --no-mixed-refs
 69                                     --partitions none --rc-lookahead 0 --ref 1
 70                                     --scenecut 0 --subme 0 --trellis 0
 71                                     --no-weightb --weightp 0
 72                                   - superfast:
 73                                     --no-mbtree --me dia --no-mixed-refs
 74                                     --partitions i8x8,i4x4 --rc-lookahead 0
 75                                     --ref 1 --subme 1 --trellis 0 --weightp 1
 76                                   - veryfast:
 77                                     --no-mixed-refs --rc-lookahead 10
 78                                     --ref 1 --subme 2 --trellis 0 --weightp 1
 79                                   - faster:
 80                                     --no-mixed-refs --rc-lookahead 20
 81                                     --ref 2 --subme 4 --weightp 1
 82                                   - fast:
 83                                     --rc-lookahead 30 --ref 2 --subme 6
 84                                     --weightp 1
 85                                   - medium:
 86                                     Default settings apply.
 87                                   - slow:
 88                                     --b-adapt 2 --direct auto --me umh
 89                                     --rc-lookahead 50 --ref 5 --subme 8

 90                                   - slower:
 91                                     --b-adapt 2 --direct auto --me umh
 92                                     --partitions all --rc-lookahead 60
 93                                     --ref 8 --subme 9 --trellis 2
 94                                   - veryslow:
 95                                     --b-adapt 2 --bframes 8 --direct auto
 96                                     --me umh --merange 24 --partitions all
 97                                     --ref 16 --subme 10 --trellis 2
 98                                     --rc-lookahead 60
 99                                   - placebo:
100                                     --bframes 16 --b-adapt 2 --direct auto
101                                     --slow-firstpass --no-fast-pskip
102                                     --me tesa --merange 24 --partitions all
103                                     --rc-lookahead 60 --ref 16 --subme 11
104                                     --trellis 2
105       --tune          Tune the settings for a particular type of source
106                               or situation
107                                   Overridden by user settings.
108                                   Multiple tunings are separated by commas.
109                                   Only one psy tuning can be used at a time.
110                                   - film (psy tuning):
111                                     --deblock -1:-1 --psy-rd :0.15
112                                   - animation (psy tuning):
113                                     --bframes {+2} --deblock 1:1
114                                     --psy-rd 0.4: --aq-strength 0.6
115                                     --ref {Double if >1 else 1}
116                                   - grain (psy tuning):
117                                     --aq-strength 0.5 --no-dct-decimate
118                                     --deadzone-inter 6 --deadzone-intra 6
119                                     --deblock -2:-2 --ipratio 1.1
120                                     --pbratio 1.1 --psy-rd :0.25
121                                     --qcomp 0.8
122                                   - stillimage (psy tuning):
123                                     --aq-strength 1.2 --deblock -3:-3
124                                     --psy-rd 2.0:0.7
125                                   - psnr (psy tuning):
126                                     --aq-mode 0 --no-psy
127                                   - ssim (psy tuning):
128                                     --aq-mode 2 --no-psy
129                                   - fastdecode:
130                                     --no-cabac --no-deblock --no-weightb
131                                     --weightp 0
132                                   - zerolatency:
133                                     --bframes 0 --force-cfr --no-mbtree
134                                     --sync-lookahead 0 --sliced-threads

135                                     --rc-lookahead 0
136       --slow-firstpass        Don't force these faster settings with --pass 1:
137                                   --no-8x8dct --me dia --partitions none
138                                   --ref 1 --subme {2 if >2 else unchanged}
139                                   --trellis 0 --fast-pskip
140
141 Frame-type options:
143   -I, --keyint Maximum GOP size [250]

Sets maximum interval between IDR-frames (default: 250). Larger values save bits, thus improve quality, at the cost of seeking precision. Unlike MPEG-1/2/4, H.264 does not suffer from DCT drift with large values of keyint.

最长相隔多少帧出现一个IDR帧
144   -i, --min-keyint   Minimum GOP size [auto]

Sets minimum interval between IDR-frames (default: 25). If scenecuts appear within this interval, they are still encoded as I-frames, but do not start a new GOP. In H.264, I-frames do not necessarily bound a closed GOP because it is allowable for a P-frame to be predicted from more frames than just the one frame before it (also see frameref). Therefore, I-frames are not necessarily seekable. IDR-frames restrict subsequent P-frames from referring to any frame prior to the IDR-frame.

最少相隔多少帧出现一个IDR帧
145       --no-scenecut           Disable adaptive I-frame decision
146       --scenecut     How aggressively to insert extra I-frames [40]

Controls how aggressively to insert extra I-frames (default: 40). With small values of scenecut, the codec often has to force an I-frame when it would exceed keyint.

Good values of scenecut may find a better location for the I-frame.

Large values use more I-frames than necessary, thus wasting bits. -1 disables scene-cut detection,

so I-frames are inserted only once every other keyint frames, even if a scene-cut occurs earlier.

This is not recommended and wastes bitrate as scenecuts encoded as P-frames are just as big as I-frames, but do not reset the "keyint counter"
147       --intra-refresh         Use Periodic Intra Refresh instead of IDR frames
Periodic Intra Refresh can replace keyframes by using a column of intra blocks that move across the video from one side to the other,
thereby "refreshing" the image. In effect, instead of a big keyframe, the keyframe is "spread" over many frames.
The video is still seekable: a special header, called the SEI Recovery Point,
tells the decoder to "start here, decode X frames, and then start displaying the video."

停用IDR帧，作为替代x264会为每隔--的 帧的每个宏区块（macroblock）使用内部编码（intra coding）。

区块是以一个水平卷动的行刷新，称为刷新波（refresh wave）。

这有利于低延迟的资料流，使它有可能比标准的IDR帧达到更加固定的帧大小。

它也增强了视讯资料流对封包遗失的恢复能力。此选项会降低压缩效 率，因此必要时才使用。

148   -b, --bframes      Number of B-frames between I and P [3]

设定x264可以使用的最大并行B帧数。

没有B帧时，一个典型的x264资料流有着像这样的帧类型：IPPPPP...PI。当设了--bframes 2时，最多两个连续的P帧可以被B帧取代，就像：IBPBBPBPPPB...PI。

B帧类似于P帧，除了B帧还能从它之后的帧做动态预测（motion prediction）。就压缩比来说效率会大幅提高。它们的平均品质是由--所控制

149       --b-adapt      Adaptive B-frame decision method [1]
150                                   Higher values may lower threading efficiency.
151                                   - 0: Disabled
152                                   - 1: Fast
153                                   - 2: Optimal (slow with high --bframes)
默认值：1

设定弹性B帧位置决策算法。此设定控制x264如何决定要放置P帧或B帧。

0：停用，总是挑选B帧。这与旧的no-b-adapt设定相同作用。 1：“快速”算法，较快，越大的--值会稍微提高速度。当使用此模式时，基本上建议搭配-- 16使用。 2：“最佳”算法，较慢，越大的--值会大幅降低速度。

注意：对于多重阶段（multi-pass）编码，仅在第一阶段（first pass）才需要此选项，因为帧类型在此时已经决定完了。

154       --b-bias       Influences how often B-frames are used [0]

默认值：0

控制使用B帧而不使用P帧的可能性。大于0的值增加偏向B帧的加权，而小于0的值则相反。范围是从-100到100。
100并不保证全是B帧（要全是B帧该使用-- 0），而-100也不保证全是P帧。

仅在你认为能比x264做出更好的位元率控制决策时才使用此选项。

155       --b-pyramid     Keep some B-frames as references [normal]
156                                   - none: Disabled
157                                   - strict: Strictly hierarchical pyramid
158                                   - normal: Non-strict (not Blu-ray compatible)

默认值：normal

允许B帧作为其他帧的参照帧。没有此设定时，帧只能参照I/P帧。虽然I/P帧因其较高的品质作为参照帧更有价值，
但B帧也是很有用的。作为参照帧的B帧会得到一个介于P帧和普通B帧之间的量化值。
b-pyramid需要至少两个以上的--才会运作。

如果是在为蓝光编码，须使用none或strict。

none：不允许B帧作为参照帧。 strict：每minigop允许一个B帧作为参照帧，这是蓝光标准强制执行的限制。 normal：每minigop允许多个B帧作为参照帧。 159       --open-gop              Use recovery points to close GOPs
160                               Only available with b-frames

默认值：none

open-gop是一个提高效率的编码技术。有三种模式：

none：停用open-gop。 normal：启用open-gop。 bluray：启用open-gop。一个效率较低的open-gop版本，因为normal模式无法用于蓝光编码。

某些解码器不完全支援open-gop资料流，这就是为什么此选项并未默认为启用。
如果想启用open-gop，应该先测试所有可能用来播放的解码器。

161       --no-cabac              Disable CABAC
162   -r, --ref          Number of reference frames [3]
163       --no-deblock            Disable loop filter
164   -f, --deblock   Loop filter parameters [0:0]
165       --slices       Number of slices per frame; forces rectangular
166                               slices and is overridden by other slicing options
167       --slice-max-size Limit the size of each slice in bytes
168       --slice-max-mbs Limit the size of each slice in macroblocks
169       --tff                   Enable interlaced mode (top field first)
170       --bff                   Enable interlaced mode (bottom field first)
171       --constrained-intra     Enable constrained intra prediction.
172       --pulldown      Use soft pulldown to change frame rate
173                                   - none, 22, 32, 64, double, triple, euro (requires cfr input)
174       --fake-interlaced       Flag stream as interlaced but encode progressive.
175                               Makes it possible to encode 25p and 30p Blu-Ray
176                               streams. Ignored in interlaced mode.
177       --frame-packing For stereoscopic videos define frame arrangement
178                                   - 0: checkerboard - pixels are alternatively from L and R
179                                   - 1: column alternation - L and R are interlaced by column

180                                   - 2: row alternation - L and R are interlaced by row
181                                   - 3: side by side - L is on the left, R on the right
182                                   - 4: top bottom - L is on top, R on bottom
183                                   - 5: frame alternation - one view per frame
184
185 Ratecontrol:
186
187   -q, --qp           Force constant QP (0-69, 0=lossless)
188   -B, --bitrate      Set bitrate (kbit/s)
189       --crf            Quality-based VBR (0-51) [23.0]
190       --rc-lookahead Number of frames for frametype lookahead [40]
191       --vbv-maxrate Max local bitrate (kbit/s) [0]
192       --vbv-bufsize Set size of the VBV buffer (kbit) [0]
193       --vbv-init       Initial VBV buffer occupancy [0.9]
194       --crf-max        With CRF+VBV, limit RF to this value
195                                   May cause VBV underflows!
196       --qpmin        Set min QP [0]
197       --qpmax        Set max QP [69]
198       --qpstep       Set max QP step [4]
199       --ratetol        Tolerance of ABR ratecontrol and VBV [1.0]
200       --ipratio        QP factor between I and P [1.40]
201       --pbratio        QP factor between P and B [1.30]
202       --chroma-qp-offset   QP difference between chroma and luma [0]
203       --aq-mode      AQ method [1]
204                                   - 0: Disabled
205                                   - 1: Variance AQ (complexity mask)
206                                   - 2: Auto-variance AQ (experimental)
207       --aq-strength    Reduces blocking and blurring in flat and
208                               textured areas. [1.0]
209
210   -p, --pass         Enable multipass ratecontrol
211                                   - 1: First pass, creates stats file
212                                   - 2: Last pass, does not overwrite stats file
213                                   - 3: Nth pass, overwrites stats file
214       --stats         Filename for 2 pass stats ["x264_2pass.log"]
215       --no-mbtree             Disable mb-tree ratecontrol.
216       --qcomp          QP curve compression [0.60]
217       --cplxblur       Reduce fluctuations in QP (before curve compression) [20.0]
218       --qblur          Reduce fluctuations in QP (after curve compression) [0.5]
219       --zones //...  Tweak the bitrate of regions of the video
220                               Each zone is of the form
221                                   ,,
222                                   where is either
223                                       q= (force QP)

224                                   or  b= (bitrate multiplier)
225       --qpfile        Force frametypes and QPs for some or all frames
226                               Format of each line: framenumber frametype QP
227                               QP is optional (none lets x264 choose). Frametypes: I,i,K,P,B,b.
228                                   K= depending on open-gop setting
229                               QPs are restricted by qpmin/qpmax.
230
231 Analysis:
232
233   -A, --partitions    Partitions to consider ["p8x8,b8x8,i8x8,i4x4"]
234                                   - p8x8, p4x4, b8x8, i8x8, i4x4
235                                   - none, all
236                                   (p4x4 requires p8x8. i8x8 requires --8x8dct.)
237       --direct        Direct MV prediction mode ["spatial"]
238                                   - none, spatial, temporal, auto
239       --no-weightb            Disable weighted prediction for B-frames
240       --weightp      Weighted prediction for P-frames [2]
241                                   - 0: Disabled
242                                   - 1: Weighted refs
243                                   - 2: Weighted refs + Duplicates
244       --me            Integer pixel motion estimation method ["hex"]
245                                   - dia: diamond search, radius 1 (fast)
246                                   - hex: hexagonal search, radius 2
247                                   - umh: uneven multi-hexagon search
248                                   - esa: exhaustive search
249                                   - tesa: hadamard exhaustive search (slow)
250       --merange      Maximum motion vector search range [16]
251       --mvrange      Maximum motion vector length [-1 (auto)]
252       --mvrange-thread   Minimum buffer between threads [-1 (auto)]
253   -m, --subme        Subpixel motion estimation and mode decision [7]
254                                   - 0: fullpel only (not recommended)
255                                   - 1: SAD mode decision, one qpel iteration
256                                   - 2: SATD mode decision
257                                   - 3-5: Progressively more qpel
258                                   - 6: RD mode decision for I/P-frames
259                                   - 7: RD mode decision for all frames
260                                   - 8: RD refinement for I/P-frames
261                                   - 9: RD refinement for all frames
262                                   - 10: QP-RD - requires trellis=2, aq-mode>0
263                                   - 11: Full RD: disable all early terminations
264       --psy-rd   Strength of psychovisual optimization ["1.0:0.0"]
265                                   #1: RD (requires subme>=6)
266                                   #2: Trellis (requires trellis, experimental)
267       --no-psy                Disable all visual optimizations that worsen
268                               both PSNR and SSIM.

269       --no-mixed-refs         Don't decide references on a per partition basis
270       --no-chroma-me          Ignore chroma in motion estimation
271       --no-8x8dct             Disable adaptive spatial transform size
272   -t, --trellis      Trellis RD quantization. [1]
273                                   - 0: disabled
274                                   - 1: enabled only on the final encode of a MB
275                                   - 2: enabled on all mode decisions
276       --no-fast-pskip         Disables early SKIP detection on P-frames
277       --no-dct-decimate       Disables coefficient thresholding on P-frames
278       --nr           Noise reduction [0]
279
280       --deadzone-inter   Set the size of the inter luma quantization deadzone [21]
281       --deadzone-intra   Set the size of the intra luma quantization deadzone [11]
282                                   Deadzones should be in the range 0 - 32.
283       --cqm           Preset quant matrices ["flat"]
284                                   - jvt, flat
285       --cqmfile       Read custom quant matrices from a JM-compatible file
286                                   Overrides any other --cqm* options.
287       --cqm4            Set all 4x4 quant matrices
288                                   Takes a comma-separated list of 16 integers.
289       --cqm8            Set all 8x8 quant matrices
290                                   Takes a comma-separated list of 64 integers.
291       --cqm4i, --cqm4p, --cqm8i, --cqm8p
292                               Set both luma and chroma quant matrices
293       --cqm4iy, --cqm4ic, --cqm4py, --cqm4pc
294                               Set individual quant matrices
295
296 Video Usability Info (Annex E):
297 The VUI settings are not used by the encoder but are merely suggestions to
298 the playback equipment. See doc/vui.txt for details. Use at your own risk.
299
300       --overscan      Specify crop overscan setting ["undef"]
301                                   - undef, show, crop
302       --videoformat   Specify video format ["undef"]
303                                   - component, pal, ntsc, secam, mac, undef
304       --range         Specify color range ["auto"]
305                                   - auto, tv, pc
306       --colorprim     Specify color primaries ["undef"]
307                                   - undef, bt709, bt470m, bt470bg
308                                     smpte170m, smpte240m, film
309       --transfer      Specify transfer characteristics ["undef"]
310                                   - undef, bt709, bt470m, bt470bg, linear,
311                                     log100, log316, smpte170m, smpte240m
312       --colormatrix   Specify color matrix setting ["???"]
313                                   - undef, bt709, fcc, bt470bg

314                                     smpte170m, smpte240m, GBR, YCgCo
315       --chromaloc    Specify chroma sample location (0 to 5) [0]
316       --nal-hrd       Signal HRD information (requires vbv-bufsize)
317                                   - none, vbr, cbr (cbr not allowed in .mp4)
318       --pic-struct            Force pic_struct in Picture Timing SEI
319       --crop-rect     Add 'left,top,right,bottom' to the bitstream-level
320                               cropping rectangle
321
322 Input/Output:
323
324   -o, --output        Specify output file
325       --muxer         Specify output container format ["auto"]
326                                   - auto, raw, mkv, flv, mp4
327       --demuxer       Specify input container format ["auto"]
328                                   - auto, raw, y4m
329       --input-fmt     Specify input file format (requires lavf support)
330       --input-csp     Specify input colorspace format for raw input
331                               - valid csps for `raw' demuxer:
332                                 i420, yv12, nv12, i422, yv16, nv16, i444, yv24, bgr, bgra, rgb
333       --output-csp    Specify output colorspace ["i420"]
334                                   - i420, i422, i444, rgb
335       --input-depth Specify input bit depth for raw input
336       --input-range   Specify input color range ["auto"]
337                                   - auto, tv, pc
338       --input-res    Specify input resolution (width x height)
339       --index         Filename for input index file
340       --sar width:height      Specify Sample Aspect Ratio
341       --fps   Specify framerate
342       --seek         First frame to encode
343       --frames       Maximum number of frames to encode
344       --level         Specify level (as defined by Annex A)
345       --bluray-compat         Enable compatibility hacks for Blu-ray support
346
347   -v, --verbose               Print stats for each frame
348       --no-progress           Don't show the progress indicator while encoding
349       --quiet                 Quiet Mode
350       --log-level     Specify the maximum level of logging ["info"]
351                                   - none, error, warning, info, debug
352       --psnr                  Enable PSNR computation
353       --ssim                  Enable SSIM computation
354       --threads      Force a specific number of threads
355       --lookahead-threads Force a specific number of lookahead threads
356       --sliced-threads        Low-latency but lower-efficiency threading
357       --thread-input          Run Avisynth in its own thread
358       --sync-lookahead Number of buffer frames for threaded lookahead

359       --non-deterministic     Slightly improve quality of SMP, at the cost of repeatability
360       --cpu-independent       Ensure exact reproducibility across different cpus,
361                                   as opposed to letting them select different algorithms
362       --asm          Override CPU detection
363       --no-asm                Disable all CPU optimizations
364       --visualize             Show MB types overlayed on the encoded video
365       --dump-yuv      Save reconstructed frames
366       --sps-id       Set SPS and PPS id numbers [0]
367       --aud                   Use access unit delimiters
368       --force-cfr             Force constant framerate timestamp generation
369       --tcfile-in     Force timestamp generation with timecode file
370       --tcfile-out    Output timecode v2 file from input timestamps
371       --timebase     Specify timebase numerator and denominator
372                      Specify timebase numerator for input timecode file
373                               or specify timebase denominator for other input
374       --dts-compress          Eliminate initial delay with container DTS hack
375
376 Filtering:
377
378       --vf, --video-filter //... Apply video filtering to the input file
379
380       Filter options may be specified in := format.
381
382       Available filters:
383       crop:left,top,right,bottom
384             removes pixels from the edges of the frame
385       resize:[width,height][,sar][,fittobox][,csp][,method]
386             resizes frames based on the given criteria:
387             - resolution only: resizes and adapts sar to avoid stretching
388             - sar only: sets the sar and resizes to avoid stretching
389             - resolution and sar: resizes to given resolution and sets the sar
390             - fittobox: resizes the video based on the desired constraints
391                - width, height, both
392             - fittobox and sar: same as above except with specified sar
393             - csp: convert to the given csp. syntax: [name][:depth]
394                - valid csp names [keep current]: i420, yv12, nv12, i422, yv16, nv16, i444, yv24, bgr, bgra, rgb
395                - depth: 8 or 16 bits per pixel [keep current]
396             note: not all depths are supported by all csps.
397             - method: use resizer method ["bicubic"]
398                - fastbilinear, bilinear, bicubic, experimental, point,
399                - area, bicublin, gauss, sinc, lanczos, spline
400       select_every:step,offset1[,...]
401             apply a selection pattern to input frames
402             step: the number of frames in the pattern
403             offsets: the offset into the step to select a frame