[FFmpeg-user] No video during playback after concatinating files

[Garrick Mason]

I am trying to combine a set of approximately 40 videos into a single 
file with as little re-encoding as possible. Most of these videos are 
from the same source (an Elgato capture card), but 4 clips are from a 
different source and therefore have different video stream properties 
such as video codec, aspect ratio, frame rate, etc. (these clips are 
meant to signify scene transitions in the final file, so I will refer to 
them as the transition clips)
I have made copies of the transition clips that are just the raw AV data 
(pcm for audio and an x264 video stream with 0 crf and keyframe 
interval) in an .mov container. After doing that, I transcoded the raw 
files using settings that match those of the video clips coming from the 
capture card. So the result is that I have all 40 videos in the same 
container format with the same audio and the same video settings 
(although I noticed something peculiar about the capture card clips 
which I'll point out when I get to ffmpeg's output), meaning 
concatinating them should be no issue.
After all the files have been combined, I tested playback using MPC-HC 
and VLC and while I do recieve video for the clips coming from the 
capture card, no video appears for the transition clips. I eventually 
found that the issue is present even when combining only two capture 
card clips and a transition clip between them, and since that's easier 
to deal with than a single 5 hour video and the 40 videos making it up, 
those are what I will be providing.

    DL for video clips (and ffmpeg's report when merging):
    https://mega.nz/#F!EV1j2KYK!Rfyj9uhCltqfXU6b_pRP2w
    06.mp4 & 07.mp4 - Clips originating from the capture card
    20mins.mp4 - The transition clip
    out.mp4 - The file generated from ffmpeg
    This is the command I ran to merge the files:
    ffmpeg -report -f concat -i "New Text Document.txt" -c:v copy -c:a
    copy out.mp4

Below is ffmpeg's output on the metadata of each clip when they're 
passed as input

    D:\home\Desktop\vidstuff\New folder\tst>ffmpeg -i 06.mp4 -i 07.mp4
    -i 20mins.mp4

    ffmpeg version N-79209-gb3eda69 Copyright (c) 2000-2016 the FFmpeg
    developers
       built with gcc 5.3.0 (GCC)
                 ...
    Input #0, mov,mp4,m4a,3gp,3g2,mj2, from '06.mp4':
       Metadata:
         major_brand     : isom
         minor_version   : 512
         compatible_brands: isomiso2avc1mp41
         title           : Stealth Who?
         date            : 2014-07-05
         encoder         : Lavf57.29.101
       Duration: 00:00:07.20, start: 0.142159, bitrate: 6055 kb/s
         Stream #0:0(eng): Video: h264 (High) (avc1 / 0x31637661),
    yuv420p(tv, smpte1
    70m/smpte170m/bt709), 720x480 [SAR 40:33 DAR 20:11], 5888 kb/s, SAR
    32:27 DAR 16
    :9, 59.94 fps, 59.94 tbr, 59940 tbn, 119.88 tbc (default)
         Metadata:
           handler_name    : VideoHandler
         Stream #0:1(eng): Audio: aac (LC) (mp4a / 0x6134706D), 48000
    Hz, stereo, flt
    p, 219 kb/s (default)
         Metadata:
           handler_name    : SoundHandler
    Input #1, mov,mp4,m4a,3gp,3g2,mj2, from '07.mp4':
       Metadata:
         major_brand     : isom
         minor_version   : 512
         compatible_brands: isomiso2avc1mp41
         title           : Stealth Who?
         date            : 2014-07-05
         encoder         : Lavf57.29.101
       Duration: 00:00:05.31, start: 0.245245, bitrate: 6363 kb/s
         Stream #1:0(eng): Video: h264 (High) (avc1 / 0x31637661),
    yuv420p(tv, smpte1
    70m/smpte170m/bt709), 720x480 [SAR 40:33 DAR 20:11], 6135 kb/s, SAR
    32:27 DAR 16
    :9, 59.94 fps, 59.94 tbr, 59940 tbn, 119.88 tbc (default)
         Metadata:
           handler_name    : VideoHandler
         Stream #1:1(eng): Audio: aac (LC) (mp4a / 0x6134706D), 48000
    Hz, stereo, flt
    p, 221 kb/s (default)
         Metadata:
           handler_name    : SoundHandler
    Input #2, mov,mp4,m4a,3gp,3g2,mj2, from '20mins.mp4':
       Metadata:
         major_brand     : isom
         minor_version   : 512
         compatible_brands: isomiso2avc1mp41
         encoder         : Lavf57.29.101
       Duration: 00:00:01.92, start: 0.000000, bitrate: 554 kb/s
         Stream #2:0(eng): Video: h264 (High) (avc1 / 0x31637661),
    yuv420p, 720x480 [
    SAR 32:27 DAR 16:9], 407 kb/s, 59.94 fps, 59.94 tbr, 11988 tbn,
    119.88 tbc (defa
    ult)
         Metadata:
           handler_name    : VideoHandler
         Stream #2:1(eng): Audio: aac (LC) (mp4a / 0x6134706D), 48000
    Hz, stereo, flt
    p, 133 kb/s (default)
         Metadata:
           handler_name    : SoundHandler
    At least one output file must be specified

This is the command I used to transcode the transition clip from the raw 
mov format to mp4

    ffmpeg -ss 2.6 -i RAW20mins.mov -filter:v
    "scale=w=720:h=480,setsar=32:27" -r 59.94 -t 1.9 -r 59.94 "20mins.mp4"

What's interesting is that for both capture card clips, ffmpeg reports 
two different SAR & DAR pairs for one video stream. Am I correct in 
assuming the first pair (in the square brackets) are just inferred 
values that aren't actually found within the input file?

    Input #0, mov,mp4,m4a,3gp,3g2,mj2, from '06.mp4':
         Stream #0:0(eng): Video: h264 (High) (avc1 / 0x31637661),
    yuv420p(tv, smpte1
    70m/smpte170m/bt709), 720x480 [SAR 40:33 DAR 20:11], 5888 kb/s, SAR
    32:27 DAR 16
    :9, 59.94 fps, 59.94 tbr, 59940 tbn, 119.88 tbc (default)
                 ...
    Input #1, mov,mp4,m4a,3gp,3g2,mj2, from '07.mp4':
         Stream #1:0(eng): Video: h264 (High) (avc1 / 0x31637661),
    yuv420p(tv, smpte1
    70m/smpte170m/bt709), 720x480 [SAR 40:33 DAR 20:11], 6135 kb/s, SAR
    32:27 DAR 16
    :9, 59.94 fps, 59.94 tbr, 59940 tbn, 119.88 tbc (default)

I have tried using the concat _filter_ instead of the demuxer to 
generate the final file and originally I generated the transcition clip 
using the first SAR (40:33). This however generated the following error:

    D:\home\Desktop\vidstuff\New folder\tst\New folder>ffmpeg -i 06.mp4
    -i 20mins.mp
    4 -i 07.mp4 -filter_complex
    "[0:0][0:1][1:0][1:1][2:0][2:1]concat=n=3:v=1:a=1[v]
    [a]" -map "[v]" -map "[a]" out.mp4
                 ....
    [Parsed_concat_0 @ 0000000002a43640] Input link in1:v0 parameters
    (size 720x480,
      SAR 40:33) do not match the corresponding output link in0:v0
    parameters (720x48
    0, SAR 32:27)
    [Parsed_concat_0 @ 0000000002a43640] Failed to configure output pad
    on Parsed_co
    ncat_0
    Error configuring complex filters.
    Invalid argument

When I generated the transition clip using the second SAR (32:27), there 
was no such error and ffmpeg went on it's merry way generating the final 
file.

Using the concat filter as I have described above results in a file that 
/does/ display video during playback, including the transition clips, 
meaning it is a workaround for the issue at hand. However, since this 
will trigger a re-encoding of the entire video stream, I only want to 
use that as a last resort (and also because I am curious about what is 
causing the issue during playback).

P.S. I am on a windows 7 computer using a windows build of ffmpeg

P.S.S. I just noticed during the writing of this that if I seek directly 
to the transition clip portion of the final video during playback, the 
video will appear (this happens for MPC-HC, but not for VLC). It is only 
during normal, uninterrupted playback that the video does not appear. 
I'm just spitballing here, but can I force ffmpeg to set a keyframe in 
the final file at the points matching up with the start of each input 
video stream?