FAQ¶

Differences Between ffmpeg-python and typed-ffmpeg¶

typed-ffmpeg is designed with inspiration from ffmpeg-python and aims to follow good design principles. However, due to differences in approach and static typing requirements, there are some key distinctions. Below are the major differences between ffmpeg-python and typed-ffmpeg. If you notice any discrepancies or have suggestions for improvements, feel free to open an issue or submit a pull request.

Defining Custom Filters for single output¶

There are two major differences when defining custom filters in typed-ffmpeg:

Explicit Type Declaration: You must specify whether the filter’s output is video or audio, as typed-ffmpeg requires this information for static type checking.
Keyword-Only Arguments: All filter options must be specified using keyword arguments (kwargs); positional arguments (args) are not allowed.

For example, in ffmpeg-python:

(
    ffmpeg.input("in.mp4")
    .filter('crop', 'in_w-2*10', 'in_h-2*20')
)

The equivalent in typed-ffmpeg would be:

(
    ffmpeg.input("in.mp4")
    .vfilter("crop", out_w="in_w-2*10", out_h="in_h-2*20")
)

For more details: please read Customizing Filters

Additional Notes¶

Most of FFmpeg’s built-in filters are already available in typed-ffmpeg, so the need to define custom filters should be minimal in most cases.