Readme
ffzap ⚡ is a simple, fast command-line tool for processing media files with ffmpeg. As it's multithreaded and can run as
many tasks in parallel as your system can handle, it's perfect for converting, compressing, or editing audio and video
files quickly and efficiently.
Because it uses ffmpeg under the hood, it supports any media file processing that ffmpeg can handle.
Installation
Usage
Speed comparison to alternatives
Requirements
All available options
Migrating to 1.0.0
License
Installation
brew tap CodeF0x/formulae
brew install ffzap
For macOS, both x86_64 and aarch64 builds are provided. For Linux, only x86_64.
Winget (Windows 10 / 11)
winget install CodeF0x.ffzap
Only x86_64.
Cargo (Universal -- 32bit and 64bit)
(Install Rust first if you haven't already)
cargo install ffzap
Every architecture Rust has compile-targets for.
Building from source
(Install Rust first if you haven't already)
git clone https://github.com/CodeF0x/ffzap
cd ffzap
cargo build -- release
Every architecture Rust has compile-targets for.
The ffzap executable will be under target/release/
Or Download a prebuilt binary
and add it to your path.
Usage
ffzap's usage is almost identical to ffmpeg, consider this simple example:
ffzap - i vids/test-1.webm - f " -c:v libx264 -b:v 1000k" - o transcoded.mp4
Mind that the ffmpeg processing options go into the - f argument (short for --ffmpeg-options ), need to be passed
as a string and without the file name.
With a single file it doesn't really make sense to use ffzap, so consider this more advanced example:
ffzap - i vids/* * /* .{ mp4, mkv} - f " -c:v libx264 -b:v 1000k" - o transcoded/{ { name} } .mp4 - t 4
This command takes all videos in vids and its subfolders ending in . mp4 and . mkv, processes them using the
options provided by - f and saves them to a (new) directory called transcoded , keeping the original filename and
changing the file extension to . mp4 while processing 4 files in parallel.
For more info on the - o syntax, run ffzap - - help . For more ffmpeg options,
visit ffmpeg's documentation .
More practical examples:
Re-encode multiple videos to H265 and the audio to opus
ffzap -- file-list files.txt - f " -c:v libx265 -preset medium -crf 23 -c:a libopus -b:a 128k" - o " Output/{{name}}.mp4" - t 2
Keypoints:
use --file-list to pass a list of file names to process
re-encode the video to H265 using - c: v libx265
- preset medium to balance out speed and file size
- crf 23 to achieve good quality with reasonable file size
re-encode the audio to opus using - c: a libopus
- b: a 128k sets the audio bitrate to 128k for a good trade-off between file size and audio quality
- t 2 runs two ffmpeg processes simultaneously to re-encode two files at once
adjust this number according to your system specs. Most system should be able to handle two instances comfortably
Convert PNG images to JPG
ffzap -- file-list files.txt - f " -c:v mjpeg -q:v 2" - o " Output/{{name}}.jpg" - t 6
Keypoints:
use --file-list to pass a list of file names to process
convert the image to JPG using - c: v mjpeg
- q: v 2 to set very high quality
- t 6 runs six processes in parallel, converting six files at once
adjust this number according to your system specs. Six shouldn't be too taxing on a modern CPU
Add a watermark to multiple videos
ffzap -- file-list files.txt - f " -i watermark.png -filter_complex [1]format=rgba,lut=a=val*0.3[watermark];[0][watermark]overlay=(main_w-overlay_w)/2:(main_h-overlay_h)/2 -c:a copy" - o " {{name}}_watermark.mp4" - t 2
(Note that this command may not work in Windows Powershell as it requires a different escaping format)
Keypoints:
use --file-list to pass a list of file names to process (these are the files the watermark gets added to)
select to watermark file with - i watermark. png inside - f
- filter_complex applies the watermark with 70% opacity to the center of each video
- c: a copy copies the audio
- t 2 processes two files in parallel
adjust this number according to your system specs. Two should be good on most modern systems
Resize multiple videos
ffzap -- file-list files.txt - f " -vf scale=1280:720 -c:a copy" - o " {{name}}_resized.mp4" - t 2
Keypoints:
use --file-list to pass a list of file names to process
- vf scale= 1280 : 720 sets the video resolution to HD
- c: a copy copies the audio
- t 2 processes two files in parallel
adjust this number according to your system specs. Two should be good on most modern systems
Swap video containers
ffzap -- file-list files.txt - o " {{name}}.mkv" - t 2
(It is assumed the source files have a container that's interchangeable with MKV)
Keypoints:
use --file-list to pass a list of file names to process
- o " {{name}}.<desired file extension> to swap all files to the desired container format (in this case MKV)
No - f because it's not needed
- t 2 processes two files in parallel
adjust this number according to your system specs. Two should be good on most modern systems
Speed comparison to commonly used alternatives (based on GitHub stars)
Preface: I want to point out that with this comparison, I am not saying the mentioned tools are bad or inferior to ffzap. They're achieving a great job and are used by many.
My personal issue with these tools is that I think they don't use system resources efficiently / don't allow the user to take control and this test aims to show that.
ffzap is not a magical tool that speeds up your batch encoding processes by simply increasing the amount of ffmpeg instances.
You need to be aware of what your system can and can't do and then leverage ffzap to take advantage of that.
Test system:
Intel Core i7 13700K
RTX 3090
32 GB RAM
NVMe SSD
Test Videos (29):
1440p
60 FPS
H264
AAC / 192k
Re-Encoding settings:
Video:
Codec: H265
Quality: CRF 24
Preset: medium
Resolution: 1080p
Audio:
Used ffzap command:
ffzap - i Videos/* * /* .mp4 - f " -c:v libx265 -crf 24 -preset medium -vf scale=1920:1080 -c:a copy" - o " Out/{{name}}.mp4" - t 2
In this example, we're comparing ffzap to other tools on how they utilize the system's CPU when re-encoding a video. An i7 13700K can run two parallel 1080p encodings comfortably when limited to 12 threads per ffmpeg instance, and we achieve that by using - t 2 (--threads 2 ) in ffzap. This of course
slows down each individual process, but because we're running two jobs at once, it makes up for it.
All other mentioned tools (except FFBatch AV Converter) run only one encoding job at any given time, resulting in the following total encoding times:
Software
Version
Total Processing Time
Available Builds
Gui
CLI
ffzap
1.0.0
00:32:16 (100%)
Windows, Linux, macOS
No
Yes
Handbrake
1.9.0
00:48:13 (149.5%)
Windows, Linux, macOS
Yes
Yes
StaxRip
2.44.4
01:43:26 (320.5%)
Windows
Yes
Yes
FFBatch AV Converter
3.2.1
00:30:44 (95.2%)
Windows, Linux (Wine)
Yes
No
ffzap and FFBatch AV Converter are the clear winners here. Given the small difference in total encoding time between the two,
ffzap could just be as fast if not even faster if I would invest more time in finding the sweetspot between ffmpeg instances and threads per instance.
For the sake of simplicity for this example I didn't to that.
Requirements
Available options
Click here to expand section
$ ffzap -- help
⚡ A multithreaded CLI for digital media processing using ffmpeg. If ffmpeg can do it, ffzap can do it - as many files in parallel as your system can handle.
Usage: ffzap [ OPTIONS] -- output < OUTPUT>
Options:
-t, -- thread-count < THREAD_COUNT>
The amount of threads you want to utilize. most systems can handle 2. Go higher if you have a powerful computer. Default is 2. Can' t be lower than 1
[default: 2]
-f, --ffmpeg-options <FFMPEG_OPTIONS>
Options you want to pass to ffmpeg. For the output file name, use --output
-i, --input <INPUT>...
The files you want to process
--file-list <FILE_LIST>
Path to a file containing paths to process. One path per line
--overwrite
If ffmpeg should overwrite files if they already exist. Default is false
--verbose
If verbose logs should be shown while ffzap is running
--delete
Delete the source file after it was successfully processed. If the process fails, the file is kept
-o, --output <OUTPUT>
Specify the output file pattern. Use placeholders to customize file paths:
{{dir}} - Entire specified file path, e.g. ./path/to/file.txt -> ?./path/to/
{{name}} - Original file' s name (without extension )
{{ext } } - Original file' s extension
Example: /destination/{{dir}}/{{name}}_transcoded.{{ext}}
Outputs the file in /destination, mirroring the original structure and keeping both the file extension and name, while adding _transcoded to the name.
-h, --help
Print help (see a summary with ' -h' )
-V, --version
Print version
Migrating to 1.0.0
In version 1. 0 . 0 , the following changes were made:
--input-directory has been deprecated and replaced by --input .
--input-file has been deprecated and replaced by --file-list .
Instead of using --input-directory , you now have to use --input to specify the files you want to process:
ffzap -- input < files here> - f " <options here>" - o " <output pattern here>"
Note: The short form - i remains unaffected by this change.
Instead of --input-file , use --file-list to specify a file containing a list of files to process:
ffzap -- file-list < path to list here> - f " <options here>" - o " <output pattern here>"
For further details and motivation behind these changes, refer to issue 16 .
License
You are free to:
Use : You can use this software for any purpose.
Modify : You can modify the software as you like.
Distribute : You can distribute the original or lightly modified software only if you credit the original author (
Tobias "CodeF0x" Oettl). Selling of the original or lightly modified versions is not allowed.
Sell or redistribute substantially modified versions : If you make significant changes to this software, you may
redistribute or sell the modified version without crediting the original author.
Conditions :
No selling of this software in its original form or with minor modifications.
Credit must be given to the original author (Tobias "CodeF0x" Oettl) if redistributing unmodified or minimally
modified versions.