ffmpeg-python VS FFmpeg

Even given an option it can be difficult to find the corresponding documentation, if only because of the many different submodules and encoders and decoders and filters that have o-so-slightly different options. That said, I've just switched from pydub to ffmpeg-python (due to memory issues of the former[1]) and judging from the Jupiter notebook[2] it seems a much more intuitive method of constructing ffmpeg pipelines.

[1] https://github.com/jiaaro/pydub/issues/135

[2] https://github.com/kkroening/ffmpeg-python/tree/master/examp...

You might want to look into ffmpeg-python as well

This bug on github sounds like the same issue you have

take a look at https://github.com/kkroening/ffmpeg-python

The ffmpeg-python python package still makes use of the “future” python package (search for “past” in the code base). These should be removed and updated as necessary.

I've had a look through these examples, but I can't get the numpy conversion to work (I get broken pipe errors, but if any of you can get it working this solution would be ideal)

Your going to be wanting to enumerate all the files in a folder, shuffle them, and then use a library for splicing them together. A quick google search brought me to moviePy, which seems pretty solid. ffmpeg is the industry standard for programmatically working with video, but it's...complicated to say the least. Looks like there are some good python bindings for it, but it may be overkill for this project. I'm sure there are others out there too.

stitching [0], [1], and [2]

[0]https://github.com/kkroening/ffmpeg-python#quickstart

[1]https://kkroening.github.io/ffmpeg-python/#ffmpeg.zoompan

[2]https://ffmpeg.org/ffmpeg-filters.html#Examples-133

I've been using ffmpeg-python recently, and it will do a lot of the crazy complex transforms for you, and you can write sane python code vs. trying to deal with the crazy default ffmpeg syntax.

https://github.com/kkroening/ffmpeg-python

The lessons yes, but the repo contains assembly for the 5-6 architectures in wide use in consumer hardware today. Separate files of course. https://github.com/FFmpeg/FFmpeg/tree/master/libavcodec

Another way to frame this is that even though the FFmpeg repository clearly indicates it's a mirror repository and has had a single open pull request titled "WARNING: PULL REQUESTS ON THIS REPOSITORY ARE IGNORED" [0] sitting in the pull requests page for 10 years, they've still had to close hundreds of pull requests.

[0] https://github.com/FFmpeg/FFmpeg/pull/153

To accomplish this task, I decided to try out the new Strands Agents SDK. It's a fairly new framework for building agents that has a simple way to define tools that the agent can use to assist in responding to prompts. For this solution, we'll need FFMPEG and Whisper installed on the machine where the agent runs. I'll be working locally, but this could easily be converted to a server-based solution using FastAPI or another web framework and deployed to the cloud in a Docker/Podman container.

Extraer fotogramas: Una clase VideoProcessor que utiliza la biblioteca ffmpeg libavcodec para procesar video y crear fotogramas a intervalos de un segundo (personalizable a través de la configuración de FPS).

Extract frames: A VideoProcessor class that uses the ffmpeg libavcodec library to process video and create frames at one-second intervals (customizable through FPS settings).

Intrinsics have the disadvantages of asm (non-portable) but also don't reliably have the advantages of them (compilers are pretty unpredictable about optimizing with them) and they're ugly (especially x86 with its weird Hungarian stuff).

There is just a little bit of intrinsics code in ffmpeg, which I wrote, that does memory copies.

https://github.com/FFmpeg/FFmpeg/blob/master/libavutil/x86/i...

It's like this because we didn't want to hide the memory accesses from the compiler, because that hurts optimization, as well as memory tools like ASan.