django-video-encoding

django-video-encoding helps to convert your videos into different formats and resolutions.

Requirements

• Python 3.10 or newer
• Django 5.2 or newer (5.2 LTS, 6.0 and 6.1 are tested)
• ffmpeg and ffprobe

The following combinations are supported and covered by the test matrix:

Installation

1. Install django-video-encoding

sh pip install django-video-encoding

1. Add video_encoding to your INSTALLED_APPS.

Integration

Add a VideoField and a GenericRelation(Format) to your model. You can optionally store the width, height and duration of the video by supplying the corresponding field names to the VideoField.

from django.contrib.contenttypes.fields import GenericRelation
from django.db import models
from video_encoding.fields import VideoField
from video_encoding.models import Format


class Video(models.Model):
    width = models.PositiveIntegerField(editable=False, null=True)
    height = models.PositiveIntegerField(editable=False, null=True)
    duration = models.FloatField(editable=False, null=True)

    file = VideoField(
        width_field='width', height_field='height', duration_field='duration'
    )

    format_set = GenericRelation(Format)

To show all converted videos in the admin, you should add the FormatInline to your ModelAdmin

from django.contrib import admin
from video_encoding.admin import FormatInline

from .models import Video


@admin.register(Video)
class VideoAdmin(admin.ModelAdmin):
    inlines = (FormatInline,)

    list_dispaly = ('get_filename', 'width', 'height', 'duration')
    fields = ('file', 'width', 'height', 'duration')
    readonly_fields = fields

The conversion of the video should be done in a separate process. Typical options are django-rq or celery. We will use django-rq in the following example. The configuration for celery is similar. django-video-encoding already provides a task (convert_all_videos) for converting all videos on a model. This task should be triggered when a video was uploaded. Hence we listen to the post-save signal and enqueue the saved instance for processing.

# signals.py
from django.db.models.signals import post_save
from django.dispatch import receiver
from django_rq import enqueue

from video_encoding import tasks

from .models import Video


@receiver(post_save, sender=Video)
def convert_video(sender, instance, **kwargs):
    enqueue(
        tasks.convert_all_videos,
        instance._meta.app_label,
        instance._meta.model_name,
        instance.pk,
    )

After a while You can access the converted videos using

video = Video.objects.get(...)
for format in video.format_set.complete().all():
   # do something

Generate a video thumbnail

The backend provides a get_thumbnail() method to extract a thumbnail from a video. Here is a basic example on how to generate the thumbnail and store it in the model.

# models.py
from django.db import models


class Video(models.Model):
    width = models.PositiveIntegerField(editable=False, null=True)
    height = models.PositiveIntegerField(editable=False, null=True)
    duration = models.FloatField(editable=False, null=True)

    thumbnail = ImageField(blank=True)
    file = VideoField(
        width_field='width', height_field='height', duration_field='duration'
    )

    format_set = GenericRelation(Format)


# tasks.py
from django.core.files import File
from video_encoding.backends import get_backend

from .models import Video


def create_thumbnail(video_pk):
    video = Video.objects.get(pk=video_pk)
    if not video.file:
        # no video file attached
        return

    if video.thumbnail:
        # thumbnail has already been generated
        return

    encoding_backend = get_backend()
    thumbnail_path = encoding_backend.get_thumbnail(video.file.path)
    filename = (os.path.basename(self.url),)

    try:
        with open(thumbnail_path, 'rb') as file_handler:
            django_file = File(file_handler)
            video.thumbnail.save(filename, django_file)
        video.save()
    finally:
        os.unlink(thumbnail_path)

You should run this method in a separate process by using django-rq, celery or similar) and enqueue execution from within a post_save signal.

# signals.py
from django.db.models.signals import post_save
from django.dispatch import receiver
from django_rq import enqueue

from . import tasks
from .models import Video


@receiver(post_save, sender=Video)
def create_thumbnail(sender, instance, **kwargs):
    enqueue(tasks.create_thumbnail, instance.pk)

Signals

During the encoding multiple signals are emitted to report the progress. You can register to the signals as described in the Django documentation.

This simple example demonstrates, on how to update the "video model" once the convertion is finished.

# apps.py
from django.apps import AppConfig


class MyAppConfig(AppConfig):
    # ...

    def ready(self) -> None:
        from . import signals  # register signals


# signals.py
from typing import Type

from django.dispatch import receiver
from video_encoding import signals

from myapp.models import Video


@receiver(signals.encoding_finished, sender=Video)
def mark_as_finished(sender: Type[Video], instance: Video) -> None:
    """
    Mark video as "convertion has been finished".
    """
    video.processed = True
    video.save(update_fields=['processed'])

signals.encoding_started

This is sent before the encoding starts.

Arguments
sender: Type[models.Model]: Model which contains the VideoField.
instance: models.Model): Instance of the model containing the VideoField.

signals.encoding_finished

Like encoding_started(), but sent after the file had been converted into all formats.

Arguments
sender: Type[models.Model]: Model which contains the VideoField.
instance: models.Model): Instance of the model containing the VideoField.

signals.format_started

This is sent before the video is converted to one of the configured formats.

Arguments
sender: Type[models.Model]: Model which contains the VideoField.
instance: models.Model): Instance of the model containing the VideoField.
format: Format: The format instance, which will reference the encoded video file.

signals.format_finished

Like format_finished, but sent after the video encoding process and includes whether the encoding was succesful or not.

Arguments
sender: Type[models.Model]: Model which contains the VideoField.
instance: models.Model): Instance of the model containing the VideoField.
format: Format: The format instance, which will reference the encoded video file.
result: ConversionResult: Instance of video_encoding.signals.ConversionResult and indicates whether the convertion FAILED, SUCCEEDED or was SKIPPED.

Configuration

VIDEO_ENCODING_THREADS (default: 1)
Defines how many threads should be used for encoding. This may not be supported by every backend.

VIDEO_ENCODING_BACKEND (default: 'video_encoding.backends.ffmpeg.FFmpegBackend')
Choose the backend for encoding. django-video-encoding only supports ffmpeg, but you can implement your own backend. Feel free to pulish your plugin and submit a pull request.

VIDEO_ENCODING_BACKEND_PARAMS (default: {})
If your backend requires some special configuration, you can specify them here as dict.

VIDEO_ENCODING_FORMATS (for defaults see video_encoding/config.py)
This dictionary defines all required encodings and has some resonable defaults. The default scale filters only downscale (min(height, ih)), so smaller videos are never upscaled. To explicitly allow upscaling, override the scale filter in your own formats, e.g. scale=-1:480. If you want to customize the formats, you have to specify name, extension and params for each format. For example

VIDEO_ENCODING_FORMATS = {
    'FFmpeg': [
        {
            'name': 'webm_sd',
            'extension': 'webm',
            'params': [
                '-b:v', '1000k', '-maxrate', '1000k', '-bufsize', '2000k',
                '-codec:v', 'libvpx', '-r', '30',
                '-vf', 'scale=-1:480', '-qmin', '10', '-qmax', '42',
                '-codec:a', 'libvorbis', '-b:a', '128k', '-f', 'webm',
           ],
        },
     ]

Encoding Backends

video_encoding.backends.ffmpeg.FFmpegBackend (default)

Backend for using ffmpeg and ffprobe to convert your videos.

Options

VIDEO_ENCODING_FFMPEG_PATH
Path to ffmpeg. If no path is provided, the backend uses which to locate it. VIDEO_ENCODING_FFPROBE_PATH
Path to ffprobe. If no path is provided, the backend uses which to locate it.

Custom Backend

You can implement a custom encoding backend. Create a new class which inherits from video_encoding.backends.base.BaseEncodingBackend. You must set the property name and implement the methods encode, get_media_info and get_thumbnail. For further details see the reference implementation: video_encoding.backends.ffmpeg.FFmpegBackend.

If you want to open source your backend, follow these steps.

1. create a packages named django-video-encoding-BACKENDNAME
2. publish your package to pypi

3. Submit a pull requests with the following changes:

4. add the package to extra_requires

5. provide reasonable defaults for VIDEO_ENCODING_FORMATS

Development

This project uses uv for packaging and managing all dependencies, hatchling as build backend, ruff for linting and formatting, mypy for type checking and pytest for testing.

Additionally, install ffmpeg and ffprobe, because the test suite encodes real videos.

Clone this repository and run

uv sync --group dev

to create a virtual environment containing all dependencies. Afterwards, you can run the test suite using

uv run pytest

The test project in test_proj can be started with

uv run python -m django runserver --settings=test_proj.settings

Before committing, run the same checks as CI:

uv run ruff check .
uv run ruff format --check .
uv run mypy video_encoding test_proj

This repository follows the Conventional Commits style.