从ffmpeg获取错误的输出分辨率或FPS

Question

我正在从安全摄像头捕获RTSP流，并使用OSX作为编码平台将其转码为(直播)到iphone。我让它正常工作，我正在调优它。但是，它似乎没有输出所请求的分辨率。这是我的脚本

/Applications/SecurityCamera/openRTSP -v -c -t rtsp://10.0.1.118/ch1-s1 | \
    /Applications/SecurityCamera/ffmpeg \
    -r 10 -i - \
    -y -an -ab 64000 -f mpegts -vcodec copy -s 960x640 \
    -flags +loop -cmp +chroma -partitions +parti4x4+partp8x8+partb8x8 \
    -subq 5 -trellis 1 -refs 1 -coder 0 -me_range 16  -keyint_min 25 \
    -sc_threshold 40 -i_qfactor 0.71 -bt 400k -maxrate 524288 -bufsize 524288 \
    -qcomp 0.6 -qmin 10 -qmax 51 -qdiff 4 -level 30 \
    -aspect 960:640 -r 10 -g 10 -async 2 -\
    |/Applications/SecurityCamera/mediastreamsegmenter -b http://localhost:8080/\
      -f /Library/WebServer/Documents/ -i stream.m3u8 -t 10 -s 4 -D

这是状态报告：

Input #0, h264, from 'pipe:':

  Duration: N/A, bitrate: N/A
  Stream #0.0: Video: h264, yuv420p, 1600x1200, 10 fps, 10 tbr, 1200k tbn, 20 tbc
  [mpegts @ 0x10100c200] muxrate VBR, pcr every 1 pkts, sdt every 200, pat/pmt every 40 pkts
  Output #0, mpegts, to 'pipe:':
  Metadata:
    encoder         : Lavf52.93.0
    Stream #0.0: Video: libx264, yuv420p, 1600x1200 [PAR 1:1 DAR 4:3], q=2-31, 90k tbn, 10 tbc
Stream mapping:
  Stream #0.0 -> #0.0

您可以看到它正在工作，但由于某种原因，它正在输出1600x1200。( -vcodec copy可能会复制所有编解码器参数，而不仅仅是编解码器类型？)

如果我将-vcodec copy更改为-vcodec libx264，则会得到正确的状态报告(说明为960x640，正确)，但流切换为2 FPS (为什么？I‘m强制输入和输出！)并且它在54帧后停止(见下面的输出)

Seems stream 0 codec frame rate differs from container frame rate: 20.00 (20/1) -> 10.00 (20/2)
Input #0, h264, from 'pipe:':
  Duration: N/A, bitrate: N/A
    Stream #0.0: Video: h264, yuv420p, 1600x1200, 10 fps, 10 tbr, 1200k tbn, 20 tbc
[buffer @ 0x100d02420] w:1600 h:1200 pixfmt:yuv420p
[scale @ 0x100d026f0] w:1600 h:1200 fmt:yuv420p -> w:960 h:640 fmt:yuv420p flags:0x4
[libx264 @ 0x10100d400] using SAR=1/1
[libx264 @ 0x10100d400] frame MB size (60x40) > level limit (1620)
[libx264 @ 0x10100d400] using cpu capabilities: MMX2 SSE2Fast SSSE3 Cache64 SlowCTZ SlowAtom
[libx264 @ 0x10100d400] profile Constrained Baseline, level 3.0
[mpegts @ 0x10100c200] muxrate VBR, pcr every 1 pkts, sdt every 200, pat/pmt every 40 pkts
Output #0, mpegts, to 'pipe:':
  Metadata:
    encoder         : Lavf52.93.0
    Stream #0.0: Video: libx264, yuv420p, 960x640 [PAR 1:1 DAR 3:2], q=10-51, 200 kb/s, 90k tbn, 10 tbc
Stream mapping:
  Stream #0.0 -> #0.0
read pmap fffps=  3 q=37.0 size=      37kB time=0.10 bitrate=3008.0kbits/s    bits/s    
video pid set at 100
found sequence start
  next segment value 1026000
written bytes 376 skipped 0
frame=   54 fps=  2 q=-1.0 Lsize=     160kB time=5.40 bitrate= 242.0kbits/s    
video:141kB audio:0kB global headers:0kB muxing overhead 12.872737%
frame I:6     Avg QP:34.68  size: 23524
[libx264 @ 0x10100d400] frame P:48    Avg QP:41.53  size:    75
[libx264 @ 0x10100d400] mb I  I16..4: 63.9%  0.0% 36.1%
[libx264 @ 0x10100d400] mb P  I16..4:  0.1%  0.0%  0.0%  P16..4:  0.8%  0.1%  0.0%  0.0%  0.0%    skip:99.0%
[libx264 @ 0x10100d400] final ratefactor: 38.54
[libx264 @ 0x10100d400] coded y,uvDC,uvAC intra: 57.7% 22.3% 2.0% inter: 0.0% 0.1% 0.0%
[libx264 @ 0x10100d400] i16 v,h,dc,p: 23% 35% 27% 15%
[libx264 @ 0x10100d400] i4 v,h,dc,ddl,ddr,vr,hd,vl,hu: 23% 32% 16%  4%  3%  3%  7%  4%  8%
[libx264 @ 0x10100d400] i8c dc,h,v,p: 83% 11%  5%  0%
[libx264 @ 0x10100d400] kb/s:214.43

Answer 1

正如你所猜到的，-vcodec copy并不意味着“使用相同类型的编码”，它意味着“原封不动地传递数据”。您正在设置的无数其他选项都不会对设置为copy的编解码器执行任何操作。

我假设将它设置为libx264，从而实际对视频进行转码，会减慢它的速度，因为现在它必须做很多工作。特别是当输入视频为1600x1200 h264时。

根据您的需求，在显示点伸缩肯定比转码快，否则请继续调优。