FFMpegCore VS FFmpeg.AutoGen

For getting info about the files, Im cribbing from a dll called FFMpegCore- https://github.com/rosenbjerg/FFMpegCore , its a wrapper around FFmpeg (so you need put a copy of ffmpeg & ffprobe) in your output folder so that works correctly

Working with multimedia is terra incognita for most of the developers since it's something that one rarely encounters while working with usual business applications. So when I was tasked to convert video for the project I'm currently working on I was expecting to deal with some sort of old poorly maintained C++ library. So FFmpegCore was a pleasant surprise since it enables working with .Net Core which is an area of my expertise.

It's been a while since I messed with this stuff, but FFmpeg.AutoGen it's pretty good once you get a few wrappers set up to abstract away all the ugly API calls.

I would recommend you take a look at https://github.com/Ruslan-B/FFmpeg.AutoGen which provide thin wrapper for FFMPEG for dotnet. Because this is thin wrapper, you probably have to teach yourself a bit on FFmpeg APi which is a pain. Maybe you can try other libraries which do not claim mobile support, I think you just need native DLL to make other project works.

I have some problems with outdated methods like "avcodec_decode_video2", currently frame splits to three horizontally and turns into black-white, i can't provide my code now because im not at pc but it really same to https://github.com/Ruslan-B/FFmpeg.AutoGen/blob/master/FFmpeg.AutoGen.Example/VideoStreamDecoder.cs, it not related to C# because i using FFmpeg.Autogen (unsafe bindings without any wrapping)

https://github.com/Ruslan-B/FFmpeg.AutoGen is great, but quite tricky to use. Basically you'll need to create an AVFormatContext with a custom AVIOContext (pb field, see libavformat docs). This IOContext contains the read() and seek() functions, which you'll create using avio_alloc_context(). For the seek() implementation, just make a new request with Content-Range for subsequent reads. After everything is set up, you can just seek based on a time code using avformat_seek_file().

We used libav directly with FFmpeg.Autogen, with some manual tweaking to certain structures that allowed us to inject proprietary ad timing metadata (not my decision, that came from higher up). Gave me a great excuse to learn how .NET interop works.

It actually uses the native FFmpeg C API directly (libav) instead of the command line tool, with FFmpeg.AutoGen.