This tutorial deals with video resizing using GPU accelerated libraries supported by FFMPEG in Ubuntu 16.04. The code here belongs to "Resize a Video is Unbelievably Fast by using FFMPEG and GPU Acceleration – A Tutorial" post.
FFmpeg is one of the most famous multimedia frameworks wich is
widely used for processing videos. In order to encode the video,
certainly a video encoder must be used. The popular
x264
is the one which is widely used however it is not
super fast! The latest NVIDIA GPUs
contain a
hardware-based video encoder called NVENC
which is much
faster than traditional ones. In order to be able to utilize this
gpu-accelerated encoder, FFmpeg must be installed with NVENC support.
The full documentation of FFmpeg integrated with NVIDIA can be fount at
here. documentation on NVENC can
be found
here.
Moreover The NVENC programming guide can be found
here.
In this tutorial the main goal is to show how to do resize a video with
GPU-accelerated libraries in Linux. In this tutorial we do not use the
terminal commands directly for employing the FFmpeg with NVENC support.
Instead the python interface is being used to run commands in the
terminal. This can be done using subprocess
python
module. This module is employed for execution and dealing external
commands, intended to supersede the os.sys
module. The
trivial method os its usage will be explained in this tutorial. Please
refer to this
documentation for
further details.
The assumption of this tutorial is that the FFmpeg is already installed with NVENC support. The installation guide can be found in FFMPEG WITH NVIDIA ACCELERATION ON UBUNTU LINUX documentation provided by NVIDIA.
This tutorial is customized for processing multiple videos. The
assumption is that the full path of each video is stored in a
.txt
file in line-by-line format. The example of the
".txt" file is as below:
Figure 1: The format of .txt file.
As a guidance if a recursive search for specific files in a directory
and its subdirectories with extension ".mov"
is
desired, the following method in command line is useful and it saves the
output as a ".txt" file:
find /absolute/path/to/directory/to/be/search -type f -name "\*.mov" > /absolute/path/to/save/the/output/textfile.txt
From now on the assumption is that the ".txt" file is ready and well-formatted. The python script for processing videos is as below:
Resizing a video using FFmpeg with NVENC encoder
import subprocess
import os
import sys
# Pre...
textfile_path = 'videos.txt'
output_dir_base = 'PATH/TO/OUTPUT'
# Read the text file
with open(textfile_path) as f:
content = f.readlines()
# you may also want to remove whitespace characters like `\n` at the end of each line
files_list = [x.strip() for x in content]
# Transpose 90 degree & Clockwise
# It already save the video file using the named defined by output_name.
for file_num, file_path_input in enumerate(files_list, start=1):
# Get the file name without extension
file_name = os.path.basename(file_path_input)
ID = file_name.split('_')[1]
raw_file_name = os.path.basename(file_name).split('.')[0]
file_dir_input = os.path.dirname(file_path_input)
file_dir_output = output_dir_base + '/' + ID
if not os.path.exists(file_dir_output):
os.makedirs(file_dir_output)
file_path_output = file_dir_output + '/' + raw_file_name + '.mkv'
print('processing file: %s' % file_path_input)
subprocess.call(
['ffmpeg', '-y', '-i', file_path_input, '-filter_complex',
'nvresize=1:s=540x960:readback=0[out0]', '-map', '[out0]',
'-acodec', 'copy', '-r', '30', '-vcodec', 'nvenc', '-b:v', '3M', file_path_output])
print('file %s saved' % file_path_output)
The videos.txt
file is saved in the absolute path.
Lines 8-12 of the code reads the ".txt" file and stores each line as
an item of a list called files_list
. The loop starts
at line 16 process each file with the
subprocess.call
command. In each loop the folder of the
input file is found and the output file will be stored in the same
directory but with different naming convention which is explained by the
comments in the code. Each ,
in the subprocess.call
command in the python is correspondent
to an empty space
in the terminal. As an example the
correspondent shell command is as below:
ffmpeg -i file\_path -filter:v transpose=-1 -vcodec nvenc -preset slow -b:v 5M -acodec copy output\_file\_path
The command executed by FFmpeg needs to be described. Each of the
elements started by -
are calling specific operations
and the command follows by them execute the desired operation. For
example -vcodec
indicator will specify the codec to
be used by FFmpeg and nvenc which follows by that point to the
codec. More details can be found at FFmpeg Filters
Documentation. The following
Table, summarize the indicators:
Attribute | Description | Argument | Description |
---|---|---|---|
-i | input argument | file_path_input | path to the input file |
-filter_complex | Multiple resize option | nvresize=1:s=540x960 | custom size of 540x960 |
-acodec | Set the audio codec | copy | get the audio stream as is |
-vcodec | Set the video codec | nvenc | Nvidia GPU accelerated library |
-r | video frame rate | 30 | output rate is 30 f/s |
-b:v | set the video bitrate | 3M | Set to 3M |
-acodec | set the audio codec | copy | only copied & no encoding |
The -vf
is the main command which its full documentation is available at here and it has the filter options.
In order to run the python file we go to the terminal and execute the following:
python /absolute/path/to/python/file
As a consideration, if we are working on any specific virtual environment it has to be activated at first.
A similar approach can be employed in the terminal too. Running commands in Terminal can be much faster than Python. So it is useful to have and idea of how to do it. Consider the following commands:
for i in \*\*/\*.mov; do
> base=\${i%.mov};
> ffmpeg -y -i file\_path\_input -filter\_complex
> nvresize=1:s=540x960:readback=0\[out0\] -map \[out0\] -acodec copy
-r 30 -vcodec nvenc -b:v 3M base=\${i%.mkv};
> done
By the assumption that the globstar
is enabled(in order
to make sure about that, the command of shopt -s globstar
can be executed in the terminal.), the above
command find all the files with the ".mov" extension in the current
directory recursively and after desired processing save them in place
with the ".mkv" extension. The advantage of globstar is that it is
recursive and robust in cases that there is white space in a file or
directory name. The command can be altered arbitrary to read the files
from a text file and do the processing.
This tutorial demonstrated how to process a video and specifically resizing that using FFmpeg and Nvidia GPU accelerated library called NVENC. The advantage of using python interface is to easily parse the .txt file and looping through all files. Moreover it enables the user with options which are more complex to be directly employed in the terminal environment.