My personal scratch pad for ffmpeg tasks that I use. Seems like everytime I need to convert some video…I end up googling for the same thing over and over again.
When creating a H.264 video file, there are a ton of options available - but this is a quick and dirty MP4 version.
ffmpeg -i "input-video" -c:v libx264 -crf 18 -pix_fmt yuv420p "output.mp4"
Pretty much the same as #1, but use an MOV container and mark it as ‘fast start’. (faststart moves the MOOV atom to the beginning of the file versus. the end - better for streaming)
ffmpeg -i "input-video" -c:v libx264 -crf 18 -pix_fmt yuv420p -movflags +faststart "output.mov"
Not nearly as useful as the H.264 version, but there are times when I need a video in an lossless format.
ffmpeg -i "input-video" -c:v rawvideo -pix_fmt bgr24 "output.avi"
If the source has an alpha channel - use a pixel format such as bgra to preserve it.
ffmpeg -i "input-video" -c:v rawvideo -pix_fmt bgra "output.avi"
Another example if you want to use QTRLE (Quicktime Animation).
ffmpeg -i "input-video" -c:v qtrle -pix_fmt argb "output.mov"
The following searches for all .avi files in a directory and converts them to H.264 in an MP4 container. (Windows .BAT file)
@echo off
for /r %%f in (*.avi) do (
ffmpeg -i "%%~nxf" -c:v libx264 -crf 18 -pix_fmt yuv420p "%%~nf.mp4"
)
To specify both width and height:
ffmpeg -i "input-video" -filter:v scale=640:320 -c copy -y "output-video"
To specify one dimension and preserve aspect-ratio (use -1):
ffmpeg -i "input-video" -filter:v scale=640:-1 -c copy -y "output-video"
For example, to slice out part of an existing mp4 video - starting at the 1 minute mark and grab 30 seconds:
ffmpeg -ss 00:01:00 -i "input-video.mp4" -t 30 -c copy -avoid_negative_ts 1 -y output.mp4
The following will dump all frames from the input video file into a subdirectory called ‘frames’ (in this example PNG is used; but JPG would also work)
ffmpeg -i "input-video" -an -f image2 "frames/frame_%05d.png"
Note: if using a Windows .bat file, you will need to escape the %, so the above pattern would be frame_%%05d.png in a .bat file
To create a video from a directory of images, the image files will need to be named with a pattern.
The following will create a video file at 60fps from a series of still images:
ffmpeg -r 60 -f image2 -i "frames/frame_%%05d.png" -c:v libx264 -crf 18 -pix_fmt yuv420p "output.mp4"
If you wish to grab just the audio track from a source and encode it as MP3:
ffmpeg -i "input-video" -vn -codec:a libmp3lame -b:a 256K "output.mp3"
Remove the audio track from a file while retaining the video stream as is.
ffmpeg -i "input-video" -an -codec:v copy "output-video"
Got a bunch of videos from an phone or iPad? I use this one for kids sports … frequent hitting of start & stop record during a game will create several video files. I often just want 1 larger video file of the game to upload to YouTube or archive.
To give ffmpeg the list of videos to combine, create a text file in the same directory as the individual video files - something like the following:
file 'IMG_0051.MOV'
file 'IMG_0052.MOV'
file 'IMG_0053.MOV'
file 'IMG_0054.MOV'
file 'IMG_0055.MOV'
file 'IMG_0056.MOV'
file 'IMG_0057.MOV'
file 'IMG_0058.MOV'
If the above text file is named 04MAR2017_BB_Game.txt, then the following ffmpeg command can be used to combine the individual MOV files into a single (larger) video file:
ffmpeg -f concat -safe 0 -i 04MAR2017_BB_Game.txt -c copy 04MAR2017_BB_Game.mov