grey images (and video) when using -ss

[aphid]

I'm splitting mpeg2 video, as captured from a capture card (and will ultimately be applying some libavfilter stuff for trademark obfuscation, but that is peripheral to this report). I've recently upgraded from ffmpeg that was probably 1.5 years or so old, and splitting fine both for still images and video, to git. When creating images with -ss, output is solid grey unless -ss happens to fall exactly on a keyframe. When creating video, it starts gray, transitions to gray+artifacts and then looks fine after the first keyframe.

This appears to be the same issue referenced here: ​https://roundup.libav.org/issue2019

Output for the following command/paste can be found here (content is public domain): ​http://metavidstorage02.ucsc.edu/raw_mpeg2/imagedump/

    ffmpeg -ss 10992 -i senate_proceeding_04-08-11.mpeg -t 10 /video/metavid/raw_mpeg2/imagedump/img%d.jpg
    FFmpeg version git-N-29131-gf4bc923, Copyright (c) 2000-2011 the FFmpeg developers
      built on Apr 18 2011 12:22:16 with gcc 4.4.3
      configuration: --enable-gpl --enable-version3 --enable-nonfree --enable-postproc --enable-libfaac --enable-libopencore-amrnb --enable-libopencore-amrwb --enable-libtheora --enable-libvorbis --enable-libx264 --enable-libxvid --enable-x11grab --enable-libvpx --enable-libfreetype
      libavutil    50. 40. 1 / 50. 40. 1
      libavcodec   52.119. 1 / 52.119. 1
      libavformat  52.108. 0 / 52.108. 0
      libavdevice  52.  4. 0 / 52.  4. 0
      libavfilter   1. 78. 0 /  1. 78. 0
      libswscale    0. 13. 0 /  0. 13. 0
      libpostproc  51.  2. 0 / 51.  2. 0
    [mpeg @ 0x13d0650] max_analyze_duration reached
    Input #0, mpeg, from 'senate_proceeding_04-08-11.mpeg':
      Duration: 18:47:49.66, start: 0.338522, bitrate: 6544 kb/s
        Stream #0.0[0x1e0]: Video: mpeg2video (Main), yuv420p, 720x480 [PAR 8:9 DAR 4:3], 8000 kb/s, 29.97 fps, 29.97 tbr, 90k tbn, 59.94 tbc
        Stream #0.1[0x1c0]: Audio: mp2, 48000 Hz, stereo, s16, 224 kb/s
    Incompatible pixel format 'yuv420p' for codec 'mjpeg', auto-selecting format 'yuvj420p'
    [buffer @ 0x13d5490] w:720 h:480 pixfmt:yuv420p
    [ffsink @ 0x13d22b0] auto-inserting filter 'auto-inserted scaler 0' between the filter 'src' and the filter 'out'
    [scale @ 0x13d2590] w:720 h:480 fmt:yuv420p -> w:720 h:480 fmt:yuvj420p flags:0xa0000004
    Output #0, image2, to '/video/metavid/raw_mpeg2/imagedump/img%d.jpg':
      Metadata:
        encoder         : Lavf52.108.0
        Stream #0.0: Video: mjpeg, yuvj420p, 720x480 [PAR 8:9 DAR 4:3], q=2-31, 200 kb/s, 90k tbn, 29.97 tbc
    Stream mapping:
      Stream #0.0 -> #0.0
    Press [q] to stop encoding
    [mpeg2video @ 0x13d3100] warning: first frame is no keyframe
    frame=  298 fps= 52 q=24.8 Lsize=      -0kB time=9.94 bitrate=  -0.0kbits/s dup=3 drop=0    
    video:5531kB audio:0kB global headers:0kB muxing overhead -100.000388%

[aphid]

fwiw, an earlier version of ffmpeg I found on another one of our servers (FFmpeg version SVN-r0.5.1-4:0.5.1-1ubuntu1.1) behaves to my expectations; using -ss before -i produces split video with the first frame as fully rendered video and produces images which are not gray or full of artifacts.