使用map-reduce机制提升转码效率和资源利用率 按照分割文件 -> 拆分任务到不同机器 -> 合并转码后的分片文件的方式,充分利用集群资源提升转码速度 栗子: 原文件input.mxf 需要转码成 output.mp4
//按60秒一个文件分割
python video-splitter/ffmpeg-split.py -f input.mxf -s 60
分片后的文件
input-1-of-4.mxf
input-2-of-4.mxf
input-3-of-4.mxf
input-4-of-4.mxf
转码后的文件
output-1-of-4.mp4
output-2-of-4.mp4
output-3-of-4.mp4
output-4-of-4.mp4
合并后的文件
output.mp4
注意,因为单机资源限制,拆分成多进程执行不能减少转码耗时
Simple command line Python script that splits video into multi chunks. Under the hood script uses FFMpeg so you will need to have that installed. No transcoding or modification of video happens, it just get's split properly.
Run python ffmpeg-split.py -h
to see the options. Here are few samples of how it could be used:
python ffmpeg-split.py -f big_video_file.mp4 -s 10
This splits big_video_file.mp4
into chunks, and the size of chunk is 10 seconds. Each chunk will be suffixed with numeric index, for example big_video_file-0.mp4
, big_video_file-1.mp4
, etc.
python ffmpeg-split.py -f input.mp4 -s 600 -v libx264 -e '-vf "scale=320:240" -threads 8'
This splits input.mp4
into chunks, and the size of chunk is 600 seconds. With extra option to scale output to 320:240
and use 8 threads to speed up.
In order to create unequal chunks of a video, you'll need to create manifest.json.
manifest.json
[
{
"start_time": 0,
"length": 34,
"rename_to": "video1"
},
{
"start_time": 35,
"length": 22,
"rename_to": "video2.mp4"
}
]
Afterwards run:
python ffmpeg-split.py -f big_video_file.mp4 -m manifest.json
This splits big_video_file.mp4
into 2 video files, video1.mp4 and video2.mp4. The video1.mp4 is a 34 seconds
clip, starting from 0:00 to 0:34 of the big_video_file.mp4
.
Alternatively, you can use a manifest.csv file to accomplish the task above.
manifest.csv:
start_time,length,rename_to
0,34,video1
35,22,video2
- start_time - number of seconds into the video or start time
- length - length of the video in seconds. The end time of the video is calculated by the start_time plus the length of the video.
- rename_to - name of the video clip to be saved
- end_time - end time of the video
- -v or --vcodec ffmpeg video codec to be used.
- -a or --acodec ffmpeg audio codec to be used.
- -m or --manifest manifest file to control the splitting of videos.
- -f or --file video file to split.
- -s or --split-size seconds to evenly split the videos
- -e or --extra extra optional options for ffmpeg, e.g. '-e -threads 8' to use 8 threads to speed up.
The -s and -m options should not be used together. If they are, -m option takes precedent over the -s option
- There might be some videos that aren't showing properly after splitting the source video with ffmpeg. To resolve
this, use the -v option and pass in the associated video codec for the source video or video format. For example, mp4 videos
use h264 video codec. Therefore, running the command
python ffmpeg-split.py -f example.mp4 -s -v h264
, may resolve this issue.
See FFmpeg installation guide for details.