[Reuben Martin]

Your video should contain information in it that lets decoders know that the video is anamorphic. Either the software you are using to play it back does not respect that information, or the source file has been altered or never had that info in it to begin with for some reason.

If you re-encode it with ffmpeg you can specify the 16:9 display aspect ratio, or even scale it up to 1920×1080 of you want.

[Reuben Martin]

Read your terminal output. It’s telling you why it won’t work.

[mpeg1video @ 0x367c160] MPEG1/2 does not support 15/1 fps

[Reuben Martin]

It means your source file is corrupt. The frame rate denominator is way out of bounds. Manually specify the input source frame rate to override it.

[Reuben Martin]

Redirect stderr output to whatever you plan to use to parse the text output. Please note that if you are trying to do this on windows, that A) You must enjoy pain and B) You will need more then the standard DOS tools to implement something like that.

If you were doing this on a *nix machine, you would direct stderr to a named pipe, then assign a file descriptor to the pipe to avoid blocking, and then access the file descriptor with a script of some sort to parse and act upon the stream of text.

[Reuben Martin]

ffplay -vf showinfo myvideo.ext

[Reuben Martin]

Aside from possible differences in what code revision was checkout out when building, I would suspect that the difference in decoding speed is entirely determined by build options, compilers used, and differences between the capabilities of the hardware and OS.

i.e. The code is last place you should be looking in this case.

[Reuben Martin]

A video file can an arbitrary number of streams. They are simply separate data streams chopped up into chunks and interleaved together. They do no have to be just audio or video either. For example a stream can hold time code, closed captioning, subtitle or metadata.

A good example you might be familiar with is DVDs. They often have more than one audio stream to support multiple languages. When you switch from english to spanish, it’s just changing with audio stream it is reading.

Likewise, audio streams can contain any number of channels (e.g. surround sound) FFmpeg can do channel mapping also.

You can have multiple video streams that can be in different codecs, different resolutions, and even different frame rates. You are only limited by whatever the limitations are of the format they are muxed into.

The “-map 0:0” means “first file, first stream”. First file being whatever file is listed first in the command line via the “-i” option. First stream is whatever stream is first inside the file, regardless of the type of stream.

[Reuben Martin]

Specifying the video codec as copy, (i.e. “-vcodec copy”) will copy the video stream without re-encoding. If you want to replace the audio, you use more than one input, and use stream mapping to specify which stream sources end up in the output.

https://ffmpeg.org/ffmpeg.html#Advanced-options

[Reuben Martin]

Why don’t you just do :

video4linux2 > VLC stream live RTSP

I’m not totally sure you understand what SDP is. It’s a Session Descriptor Protocol. It doesn’t contain any media, it just describes how to connect to the stream you want to connect to. They’re not needed to connect to an RTSP stream if you just type in the streaming address directly, but if you want one they’re are simple to write using a simple text editor.

[Reuben Martin]

That would require write a shell script if you want to do batch operations. (Or for something more sane, perhaps a ruby, python, or perl script)

But if you are not comfortable with running programs from a shell terminal, you will probably not be too keen on writing a script either.

There are two examples given here of burning timecode: https://ffmpeg.org/trac/ffmpeg/wiki/FilteringGuide