FFmpeg图像到视频转换错误：[image2]打开要读取的文件

Question

我在一个AWS Lambda实例上运行ffmpeg。Lambda函数获取输入图像，并使用ffmpeg将其转码为视频片段：

ffmpeg -loop 1 -i /tmp/photo-SNRUR7ZS13.jpg -c:v libx264 -t 7.00 -pix_fmt yuv420p -vf scale=1280x720 /tmp/output.mp4

我看到不一致的行为，有时输出的视频比指定的持续时间短。这在随机图像中发生的情况并不一致。相同的图像有时会正确渲染，有时会被缩短。

此行为仅发生在Lambda上。我无法将其复制到本地计算机上，也无法复制到运行在lambda上的环境相同的专用EC2实例上。

我注意到当输出视频很短时，ffmpeg日志是不同的。主要区别在于重复的[image2 @ 0x4b11140] Opening '/tmp/photo-2HD2Z3UN3W.jpg' for reading行。参见下面的ffmpeg日志。

正常执行，输出视频长度正确：

    ffmpeg -loop 1 -i /tmp/photo-SNRUR7ZS13.jpg -c:v libx264 -t 7.00 -pix_fmt yuv420p -vf scale=1280x720 /tmp/video-TMB6RNO0EE.mp4
ffmpeg version N-89773-g7fcbebbeaf-static https://johnvansickle.com/ffmpeg/  Copyright (c) 2000-2018 the FFmpeg developers
  built with gcc 6.4.0 (Debian 6.4.0-11) 20171206
  configuration: --enable-gpl --enable-version3 --enable-static --disable-debug --disable-ffplay --disable-indev=sndio --disable-outdev=sndio --cc=gcc-6 --enable-fontconfig --enable-frei0r --enable-gnutls --enable-gray --enable-libfribidi --enable-libass --enable-libvmaf --enable-libfreetype --enable-libmp3lame --enable-libopencore-amrnb --enable-libopencore-amrwb --enable-libopenjpeg --enable-librubberband --enable-librtmp --enable-libsoxr --enable-libspeex --enable-libvorbis --enable-libopus --enable-libtheora --enable-libvidstab --enable-libvo-amrwbenc --enable-libvpx --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxvid --enable-libzimg
  libavutil      56.  7.100 / 56.  7.100
  libavcodec     58.  9.100 / 58.  9.100
  libavformat    58.  3.100 / 58.  3.100
  libavdevice    58.  0.100 / 58.  0.100
  libavfilter     7. 11.101 /  7. 11.101
  libswscale      5.  0.101 /  5.  0.101
  libswresample   3.  0.101 /  3.  0.101
  libpostproc    55.  0.100 / 55.  0.100
Input #0, image2, from '/tmp/photo-SNRUR7ZS13.jpg':
  Duration: 00:00:00.04, start: 0.000000, bitrate: 18703 kb/s
    Stream #0:0: Video: mjpeg, yuvj444p(pc, bt470bg/unknown/unknown), 687x860 [SAR 200:200 DAR 687:860], 25 fps, 25 tbr, 25 tbn, 25 tbc
Stream mapping:
  Stream #0:0 -> #0:0 (mjpeg (native) -> h264 (libx264))
Press [q] to stop, [?] for help
[swscaler @ 0x5837900] deprecated pixel format used, make sure you did set range correctly
[libx264 @ 0x51c2340] using SAR=1477/3287
[libx264 @ 0x51c2340] using cpu capabilities: MMX2 SSE2Fast SSSE3 SSE4.2 AVX FMA3 BMI2 AVX2
[libx264 @ 0x51c2340] profile High, level 3.1
[libx264 @ 0x51c2340] 264 - core 155 r61 b00bcaf - H.264/MPEG-4 AVC codec - Copyleft 2003-2017 - http://www.videolan.org/x264.html - options: cabac=1 ref=3 deblock=1:0:0 analyse=0x3:0x113 me=hex subme=7 psy=1 psy_rd=1.00:0.00 mixed_ref=1 me_range=16 chroma_me=1 trellis=1 8x8dct=1 cqm=0 deadzone=21,11 fast_pskip=1 chroma_qp_offset=-2 threads=3 lookahead_threads=1 sliced_threads=0 nr=0 decimate=1 interlaced=0 bluray_compat=0 constrained_intra=0 bframes=3 b_pyramid=2 b_adapt=1 b_bias=0 direct=1 weightb=1 open_gop=0 weightp=2 keyint=250 keyint_min=25 scenecut=40 intra_refresh=0 rc_lookahead=40 rc=crf mbtree=1 crf=23.0 qcomp=0.60 qpmin=0 qpmax=69 qpstep=4 ip_ratio=1.40 aq=1:1.00
Output #0, mp4, to '/tmp/video-TMB6RNO0EE.mp4':
  Metadata:
    encoder         : Lavf58.3.100
    Stream #0:0: Video: h264 (libx264) (avc1 / 0x31637661), yuv420p, 1280x720 [SAR 6183:13760 DAR 687:860], q=-1--1, 25 fps, 12800 tbn, 25 tbc
    Metadata:
      encoder         : Lavc58.9.100 libx264
    Side data:
      cpb: bitrate max/min/avg: 0/0/0 buffer size: 0 vbv_delay: -1
frame=   49 fps=0.0 q=28.0 size=       0kB time=-00:00:00.03 bitrate=N/A speed=N/A    
frame=   69 fps= 66 q=28.0 size=       0kB time=00:00:00.76 bitrate=   0.5kbits/s speed=0.728x    
frame=   89 fps= 57 q=28.0 size=       0kB time=00:00:01.56 bitrate=   0.2kbits/s speed=0.998x    
frame=  109 fps= 53 q=28.0 size=       0kB time=00:00:02.36 bitrate=   0.2kbits/s speed=1.14x    
frame=  129 fps= 50 q=28.0 size=       0kB time=00:00:03.16 bitrate=   0.1kbits/s speed=1.22x    
frame=  148 fps= 48 q=28.0 size=       0kB time=00:00:03.92 bitrate=   0.1kbits/s speed=1.27x    
frame=  168 fps= 47 q=28.0 size=       0kB time=00:00:04.72 bitrate=   0.1kbits/s speed=1.31x    
No more output streams to write to, finishing.
frame=  175 fps= 39 q=-1.0 Lsize=      94kB time=00:00:06.88 bitrate= 112.2kbits/s speed=1.54x    
video:91kB audio:0kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: 3.161261%
Input file #0 (/tmp/photo-SNRUR7ZS13.jpg):
  Input stream #0:0 (video): 176 packets read (16459168 bytes); 176 frames decoded; 
  Total: 176 packets (16459168 bytes) demuxed
Output file #0 (/tmp/video-TMB6RNO0EE.mp4):
  Output stream #0:0 (video): 175 frames encoded; 175 packets muxed (93507 bytes); 
  Total: 175 packets (93507 bytes) muxed
[libx264 @ 0x51c2340] frame I:1     Avg QP:14.33  size: 73084
[libx264 @ 0x51c2340] frame P:44    Avg QP:14.09  size:   302
[libx264 @ 0x51c2340] frame B:130   Avg QP:23.31  size:    50
[libx264 @ 0x51c2340] consecutive B-frames:  0.6%  1.1%  0.0% 98.3%
[libx264 @ 0x51c2340] mb I  I16..4:  3.3% 84.5% 12.1%
[libx264 @ 0x51c2340] mb P  I16..4:  0.0%  0.0%  0.0%  P16..4:  3.2%  0.1%  0.0%  0.0%  0.0%    skip:96.7%
[libx264 @ 0x51c2340] mb B  I16..4:  0.0%  0.0%  0.0%  B16..8:  0.4%  0.0%  0.0%  direct: 0.0%  skip:99.6%  L0:31.2% L1:68.8% BI: 0.0%
[libx264 @ 0x51c2340] 8x8 transform intra:84.5% inter:98.8%
[libx264 @ 0x51c2340] coded y,uvDC,uvAC intra: 95.1% 63.9% 51.6% inter: 0.1% 0.6% 0.0%
[libx264 @ 0x51c2340] i16 v,h,dc,p: 26% 21%  4% 49%
[libx264 @ 0x51c2340] i8 v,h,dc,ddl,ddr,vr,hd,vl,hu: 20% 27% 21%  3%  5%  6%  6%  4%  9%
[libx264 @ 0x51c2340] i4 v,h,dc,ddl,ddr,vr,hd,vl,hu: 23% 36% 10%  4%  7%  5%  6%  2%  6%
[libx264 @ 0x51c2340] i8c dc,h,v,p: 51% 29% 16%  4%
[libx264 @ 0x51c2340] Weighted P-Frames: Y:0.0% UV:0.0%
[libx264 @ 0x51c2340] ref P L0: 96.5%  0.0%  3.3%  0.2%
[libx264 @ 0x51c2340] ref B L0: 42.4% 57.6%
[libx264 @ 0x51c2340] ref B L1: 97.0%  3.0%
[libx264 @ 0x51c2340] kb/s:106.08

短视频日志：

ffmpeg -framerate 25 -y -loop 1 -i /tmp/photo-2HD2Z3UN3W.jpg -t 15.00 -filter_complex "[0:v]crop=h=ih:w='if(gt(a,16/9),ih*16/9,iw)':y=0:x='if(gt(a,16/9),(ow-iw)/2,0)'[tmp];[tmp]scale=-1:4000,crop=w=iw:h='min(iw*9/16,ih)':x=0:y='0.17*ih-((t/15.00)*min(0.17*ih,(ih-oh)/6))',trim=duration=15.00[tmp1];[tmp1]zoompan=z='if(lte(pzoom,1.0),1.15,max(1.0,pzoom-0.0005))':x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)':d=1,setsar=sar=1:1[animated];[animated]fade=out:st=12.00:d=3.00:c=#000000[animated]" -map "[animated]" -pix_fmt yuv420p -s 1280x720 -y /tmp/video-QB1JCDT021.mp4
ffmpeg version N-89773-g7fcbebbeaf-static https://johnvansickle.com/ffmpeg/ Copyright (c) 2000-2018 the FFmpeg developers
built with gcc 6.4.0 (Debian 6.4.0-11) 20171206
configuration: --enable-gpl --enable-version3 --enable-static --disable-debug --disable-ffplay --disable-indev=sndio --disable-outdev=sndio --cc=gcc-6 --enable-fontconfig --enable-frei0r --enable-gnutls --enable-gray --enable-libfribidi --enable-libass --enable-libvmaf --enable-libfreetype --enable-libmp3lame --enable-libopencore-amrnb --enable-libopencore-amrwb --enable-libopenjpeg --enable-librubberband --enable-librtmp --enable-libsoxr --enable-libspeex --enable-libvorbis --enable-libopus --enable-libtheora --enable-libvidstab --enable-libvo-amrwbenc --enable-libvpx --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxvid --enable-libzimg
libavutil 56. 7.100 / 56. 7.100
libavcodec 58. 9.100 / 58. 9.100
libavformat 58. 3.100 / 58. 3.100
libavdevice 58. 0.100 / 58. 0.100
libavfilter 7. 11.101 / 7. 11.101
libswscale 5. 0.101 / 5. 0.101
libswresample 3. 0.101 / 3. 0.101
libpostproc 55. 0.100 / 55. 0.100
Input #0, image2, from '/tmp/photo-2HD2Z3UN3W.jpg':
Duration: 00:00:00.04, start: 0.000000, bitrate: 373617 kb/s
Stream #0:0: Video: mjpeg, yuvj444p(pc, bt470bg/unknown/unknown), 1936x2592 [SAR 72:72 DAR 121:162], 25 fps, 25 tbr, 25 tbn, 25 tbc
Stream mapping:
Stream #0:0 (mjpeg) -> crop
fade -> Stream #0:0 (libx264)
Press [q] to stop, [?] for help
[swscaler @ 0x4d63b40] deprecated pixel format used, make sure you did set range correctly
[swscaler @ 0x4df7340] deprecated pixel format used, make sure you did set range correctly
[swscaler @ 0x50e97c0] deprecated pixel format used, make sure you did set range correctly
[swscaler @ 0x50e97c0] Warning: data is not aligned! This can lead to a speed loss
[libx264 @ 0x4b17480] using SAR=1/1
[libx264 @ 0x4b17480] using cpu capabilities: MMX2 SSE2Fast SSSE3 SSE4.2 AVX FMA3 BMI2 AVX2
[libx264 @ 0x4b17480] profile High, level 3.1
[libx264 @ 0x4b17480] 264 - core 155 r61 b00bcaf - H.264/MPEG-4 AVC codec - Copyleft 2003-2017 - http://www.videolan.org/x264.html - options: cabac=1 ref=3 deblock=1:0:0 analyse=0x3:0x113 me=hex subme=7 psy=1 psy_rd=1.00:0.00 mixed_ref=1 me_range=16 chroma_me=1 trellis=1 8x8dct=1 cqm=0 deadzone=21,11 fast_pskip=1 chroma_qp_offset=-2 threads=3 lookahead_threads=1 sliced_threads=0 nr=0 decimate=1 interlaced=0 bluray_compat=0 constrained_intra=0 bframes=3 b_pyramid=2 b_adapt=1 b_bias=0 direct=1 weightb=1 open_gop=0 weightp=2 keyint=250 keyint_min=25 scenecut=40 intra_refresh=0 rc_lookahead=40 rc=crf mbtree=1 crf=23.0 qcomp=0.60 qpmin=0 qpmax=69 qpstep=4 ip_ratio=1.40 aq=1:1.00
Output #0, mp4, to '/tmp/video-QB1JCDT021.mp4':
Metadata:
encoder : Lavf58.3.100
Stream #0:0: Video: h264 (libx264) (avc1 / 0x31637661), yuv420p, 1280x720 [SAR 1:1 DAR 16:9], q=-1--1, 25 fps, 12800 tbn, 25 tbc
Metadata:
encoder : Lavc58.9.100 libx264
Side data:
cpb: bitrate max/min/avg: 0/0/0 buffer size: 0 vbv_delay: -1
[swscaler @ 0x5bd0380] deprecated pixel format used, make sure you did set range correctly
debug=1
cur_dts is invalid (this is harmless if it occurs once at the start per stream)
[image2 @ 0x4b11140] Opening '/tmp/photo-2HD2Z3UN3W.jpg' for reading
[AVIOContext @ 0x4b6ecc0] Statistics: 1868086 bytes read, 0 seeks
[mjpeg @ 0x4b14940] marker=d8 avail_size_in_buf=1868084
[mjpeg @ 0x4b14940] marker parser used 0 bytes (0 bits)
[mjpeg @ 0x4b14940] marker=e0 avail_size_in_buf=1868082
[mjpeg @ 0x4b14940] marker parser used 16 bytes (128 bits)
[mjpeg @ 0x4b14940] marker=db avail_size_in_buf=1868064
[mjpeg @ 0x4b14940] index=0
[mjpeg @ 0x4b14940] qscale[0]: 0
[mjpeg @ 0x4b14940] marker parser used 67 bytes (536 bits)
[mjpeg @ 0x4b14940] marker=db avail_size_in_buf=1867995
[mjpeg @ 0x4b14940] index=1
[mjpeg @ 0x4b14940] qscale[1]: 1
[mjpeg @ 0x4b14940] marker parser used 67 bytes (536 bits)
[mjpeg @ 0x4b14940] marker=c0 avail_size_in_buf=1867926
[mjpeg @ 0x4b14940] sof0: picture: 1936x2592
[mjpeg @ 0x4b14940] component 0 1:1 id: 0 quant:0
[mjpeg @ 0x4b14940] component 1 1:1 id: 1 quant:1
[mjpeg @ 0x4b14940] component 2 1:1 id: 2 quant:1
[mjpeg @ 0x4b14940] pix fmt id 11111100
[mjpeg @ 0x4b14940] marker parser used 17 bytes (136 bits)
[mjpeg @ 0x4b14940] marker=c4 avail_size_in_buf=1867907
[mjpeg @ 0x4b14940] class=0 index=0 nb_codes=11
[mjpeg @ 0x4b14940] marker parser used 30 bytes (240 bits)
[mjpeg @ 0x4b14940] marker=c4 avail_size_in_buf=1867875
[mjpeg @ 0x4b14940] class=1 index=0 nb_codes=242
[mjpeg @ 0x4b14940] marker parser used 82 bytes (656 bits)
[mjpeg @ 0x4b14940] marker=c4 avail_size_in_buf=1867791
[mjpeg @ 0x4b14940] class=0 index=1 nb_codes=8
[mjpeg @ 0x4b14940] marker parser used 27 bytes (216 bits)
[mjpeg @ 0x4b14940] marker=c4 avail_size_in_buf=1867762
[mjpeg @ 0x4b14940] class=1 index=1 nb_codes=241
[mjpeg @ 0x4b14940] marker parser used 51 bytes (408 bits)
[mjpeg @ 0x4b14940] escaping removed 7149 bytes
[mjpeg @ 0x4b14940] marker=da avail_size_in_buf=1867709
[mjpeg @ 0x4b14940] component: 0
[mjpeg @ 0x4b14940] component: 1
[mjpeg @ 0x4b14940] component: 2
[mjpeg @ 0x4b14940] marker parser used 1860559 bytes (14884468 bits)
[mjpeg @ 0x4b14940] marker=d9 avail_size_in_buf=0
[mjpeg @ 0x4b14940] decode frame unused 0 bytes
[swscaler @ 0x5bd42c0] deprecated pixel format used, make sure you did set range correctly
cur_dts is invalid (this is harmless if it occurs once at the start per stream)
[image2 @ 0x4b11140] Opening '/tmp/photo-2HD2Z3UN3W.jpg' for reading
[AVIOContext @ 0x4b6ecc0] Statistics: 1868086 bytes read, 0 seeks
[mjpeg @ 0x4b14940] marker=d8 avail_size_in_buf=1868084
[mjpeg @ 0x4b14940] marker parser used 0 bytes (0 bits)
[mjpeg @ 0x4b14940] marker=e0 avail_size_in_buf=1868082
[mjpeg @ 0x4b14940] marker parser used 16 bytes (128 bits)
[mjpeg @ 0x4b14940] marker=db avail_size_in_buf=1868064
[mjpeg @ 0x4b14940] index=0
[mjpeg @ 0x4b14940] qscale[0]: 0
[mjpeg @ 0x4b14940] marker parser used 67 bytes (536 bits)
[mjpeg @ 0x4b14940] marker=db avail_size_in_buf=1867995
[mjpeg @ 0x4b14940] index=1
[mjpeg @ 0x4b14940] qscale[1]: 1
[mjpeg @ 0x4b14940] marker parser used 67 bytes (536 bits)
[mjpeg @ 0x4b14940] marker=c0 avail_size_in_buf=1867926
[mjpeg @ 0x4b14940] sof0: picture: 1936x2592
[mjpeg @ 0x4b14940] component 0 1:1 id: 0 quant:0
[mjpeg @ 0x4b14940] component 1 1:1 id: 1 quant:1
[mjpeg @ 0x4b14940] component 2 1:1 id: 2 quant:1
[mjpeg @ 0x4b14940] pix fmt id 11111100
[mjpeg @ 0x4b14940] marker parser used 17 bytes (136 bits)
[mjpeg @ 0x4b14940] marker=c4 avail_size_in_buf=1867907
[mjpeg @ 0x4b14940] class=0 index=0 nb_codes=11
[mjpeg @ 0x4b14940] marker parser used 30 bytes (240 bits)
[mjpeg @ 0x4b14940] marker=c4 avail_size_in_buf=1867875
[mjpeg @ 0x4b14940] class=1 index=0 nb_codes=242
[mjpeg @ 0x4b14940] marker parser used 82 bytes (656 bits)
[mjpeg @ 0x4b14940] marker=c4 avail_size_in_buf=1867791
[mjpeg @ 0x4b14940] class=0 index=1 nb_codes=8
[mjpeg @ 0x4b14940] marker parser used 27 bytes (216 bits)
[mjpeg @ 0x4b14940] marker=c4 avail_size_in_buf=1867762
[mjpeg @ 0x4b14940] class=1 index=1 nb_codes=241
[mjpeg @ 0x4b14940] marker parser used 51 bytes (408 bits)
[mjpeg @ 0x4b14940] escaping removed 7149 bytes
[mjpeg @ 0x4b14940] marker=da avail_size_in_buf=1867709
[mjpeg @ 0x4b14940] component: 0
[mjpeg @ 0x4b14940] component: 1
[mjpeg @ 0x4b14940] component: 2
[mjpeg @ 0x4b14940] marker parser used 1860559 bytes (14884468 bits)
[mjpeg @ 0x4b14940] marker=d9 avail_size_in_buf=0
[mjpeg @ 0x4b14940] decode frame unused 0 bytes
[swscaler @ 0x5bd8200] deprecated pixel format used, make sure you did set range correctly
cur_dts is invalid (this is harmless if it occurs once at the start per stream)
...
...
...

按照要求，这里有一个link to the full log。在此log - ffmpeg中，仅渲染375帧中的323帧。

Opening '/tmp/photo-2HD2Z3UN3W.jpg'片段会重复很多很多次，直到它最终呈现出一个短视频。有没有人知道为什么它一直打开图像文件？这一定与潜在的问题有关。

Answer 1

应用于图像(序列)的-loop选项强制ffmpeg在每次循环输入时重新打开和解码文件。因此，在这种情况下，ffmpeg必须通过网络重复打开该文件。我的猜测是，一些HTTP请求没有得到足够快的响应，而ffmpeg会丢弃那些接收到的帧。

在删除-loop 1并切换到loop filter：loop=NB_LOOPS:1:0,setpts=N/FRAME_RATE/TB之后，ffmpeg只打开图像一次，然后在其缓冲区中克隆解码的帧。这给出了预期的结果。