Shulou(Shulou.com)06/01 Report--

In this issue, the editor will bring you about how to achieve video compression and format conversion in Serverless. The article is rich in content and analyzes and narrates it from a professional point of view. I hope you can get something after reading this article.

Ready to start ffmpeg

You can see this description in the encyclopedia:

FFmpeg is an open source computer program that can be used to record, convert digital audio and video, and convert it into a stream. Use the LGPL or GPL license. It provides a complete solution for recording, converting and streaming audio and video. It contains a very advanced audio / video codec library libavcodec. In order to ensure high portability and codec quality, many code in libavcodec are developed from scratch. FFmpeg is developed under the Linux platform, but it can also be compiled and run in other operating systems, including Windows, Mac OS X and so on. The project was first initiated by Fabrice Bellard and was mainly maintained by Michael Niedermayer from 2004 to 2015. Many FFmpeg developers are from MPlayer projects, and the current FFmpeg is on the server of the MPlayer project team. The name of the project comes from the MPEG video coding standard, and the preceding "FF" stands for "Fast Forward".

In the actual production life, ffmpeg is indeed a very good tool, we can use this tool to carry out image compression / transcoding and other operations.

Through ffmpeg's official website, we can see different operating systems with different files for us to choose from:

In other words, if we want to use this module in the cloud function, we need to have such a module that can be run in the environment where the cloud function is located. According to the cloud function document, you can see:

In other words, we need to have a ffmpeg that can be used under the CentOS operating system, and then we prepare this file:

On the CentOS operating system, download the source package: wget http://www.ffmpeg.org/releases/ffmpeg-3.1.tar.gz

Extract and enter the directory: tar-zxvf ffmpeg-3.1.tar.gz & & cd ffmpeg-3.1

Compile and install:. / configure & & make & & make install

During the. / configure operation, yasm/nasm not found or too old. Use-disable-yasm for a crippledbuild error.

Yasm is an assembly compiler. Ffmpeg uses assembly instructions, such as MMX and SSE, to improve efficiency. So when yasm is not installed on the system, an error will be reported, and you can install the yasm compiler to solve the problem:

Download wget http://www.tortall.net/projects/yasm/releases/yasm-1.3.0.tar.gz

Extract and enter the directory: tar zxvf yasm-1.3.0.tar.gz & & cd yasm-1.3.0

Compile and install:. / configure & & make & & make install

After completing the compilation and installation of ffmpeg, you can see the generated file in the current directory: ffmpeg

At this point, we can save this file and use it in Tencent Cloud's cloud functions.

Serverless facilitates Video Compression

According to the time architecture diagram provided by Tencent Cloud, we can see that the trigger function of object storage is recommended, that is, we store the video in object storage, and then use the trigger trigger function of object storage to process the video, and then send back the operation of object storage.

Code implementation:

Import osimport subprocessfrom qcloud_cos_v5 import CosConfigfrom qcloud_cos_v5 import CosS3Clientsecret_id = os.environ.get ('secret_id') secret_key = os.environ.get (' secret_key') region = os.environ.get ('region') cosClient = CosS3Client (CosConfig (Region=region, SecretId=secret_id, SecretKey=secret_key)) # move ffmpeg to tmp directory And grant permissions to with open (". / ffmpeg", "rb") as rf: with open ("/ tmp/ffmpeg", "wb") as wf: wf.write (rf.read ()) subprocess.run ('chmod 755 / tmp/ffmpeg', shell=True) def main_handler (event) Context): for record in event ['Records']: bucket = record [' cos'] ['cosBucket'] [' name'] +'-'+ record ['cos'] [' cosBucket'] ['appid'] key = "/" .join (record [' cos'] ['cosObject'] [' key'] .split ("/") [3:]) download_path ='/ tmp/ {} '.format (key.split ( '/') [- 1]) upload_path ='/ tmp/new_mp4- {} '.format (key.split (' /') [- 1]) # download picture print ("key" Key) response = cosClient.get_object (Bucket=bucket, Key=key) response ['Body'] .get_stream_to_file (download_path) # execute the ffmpeg instruction to compress the video child = subprocess.run (' / tmp/ffmpeg-I% s-r 10-bpura 32k% s% (download_path, upload_path), stdout=subprocess.PIPE, stderr=subprocess.PIPE, close_fds=True Shell=True) # upload picture cosClient.put_object_from_local_file (Bucket=bucket, LocalFilePath=upload_path, Key= "/ new_mp4/" + key.split ('/') [- 1])

The main operation here is to copy the ffmpeg to the executable directory when the container is established, or when the function starts cold, and set its permissions to 755.

When you are finished, you can write serverless.yaml:

MyVideo: component: "@ serverless/tencent-scf" inputs: name: MyVideo codeUri:. / handler: index.main_handler runtime: Python3.6 region: ap-guangzhou memorySize: 128timeout: 200environment: variables: secret_id: user key id secret_key: user key key region: ap-guangzhou events:-cos: name: video-1256773370. Cos.ap-guangzhou.myqcloud.com parameters: bucket: video-1256773370.cos.ap-guangzhou.myqcloud.com filter: prefix: source/ events: cos:ObjectCreated:* enable: true

After the deployment is complete, we upload a test MP4 file to the corresponding storage source/ folder:

Wait a moment, we can see the corresponding video appears in the target folder:

You can see the gap between the two video files.

Of course, here we only generate video for video compression through / tmp/ffmpeg-I original video-r 10-bcanmeng a 32k. In addition, we can also use ffmpeg for additional operations (the following content is from canmeng's blog):

Ffmpeg-ss 00:00:00-t 00:00:30-I test.mp4-vcodec copy-acodec copy output.mp4

-ss specifies when to start

-t specifies how long to intercept

-I specify the input file

This command is to cut from 00 seconds to 00: 30: 30 seconds, a total of 30 seconds of video. This command is executed quickly because it is only a copy of the original data, and there is no encoding and decoding process. After executing this command, you can get the output file output.mp4. You can use video playback software to play this video.

There may be some video clipping results, such as expected consistent, 00 seconds to start, 30 seconds to end, a total of 30 seconds of video, but some videos after clipping you will find that the beginning and end may not be very accurate, it may start in 00 seconds and end in 33 seconds. Why is that?

Because the 30 seconds in these videos don't happen to be keyframes, ffmpeg will circle to the nearest Keyframe near the point in time you enter, and then do the next thing. If you don't know what a Keyframe is, it doesn't matter, it doesn't affect your use of this command.

Merge video

/ / intercept 30sffmpeg-ss 00:00:00-t 00:00:30-I keyoutput.mp4-vcodec copy-acodec copy split.mp4// from scratch and intercept 30sffmpeg-ss 00:00:30-t 00:00:30-I keyoutput.mp4-vcodec copy-acodec copy split1.mp4// from 30s to merge ffmpeg-f concat-I list.txt-c copy concat.mp4.

In the list.txt file, the video clips to be merged are described. The contents are as follows

File. / split.mp4file. / split1.mp4

More common commands:

Ffmpeg-I in.mp4-filter:v "crop=in_w:in_h-40"-copy out.mp4// remove audio from video ffmpeg-I input.mp4-vcodec copy-an output.mp4//-an: remove audio;-vcodec: video option, usually followed by copy to copy / / extract audio ffmpeg-I input.mp4-acodec copy-vn output.mp3//-vn: remove video -acodec: audio option, usually followed by copy to indicate copy / / audio / video synthesis ffmpeg-y-I input.mp4-I input.mp3-vcodec copy-acodec copy output.mp4//-y overwrite output file / / cut video ffmpeg-ss 0:1:30-t 0:0:20-I input.mp4-vcodec copy-acodec copy output.mp4//-ss start time -t duration / / Video screenshot ffmpeg-I test.mp4-f image2-t 0.001-s 320x240 image-%3d.jpg//-s setting resolution;-f forcing the format fmt / / decompose the video into pictures ffmpeg-I test.mp4-r 1-f image2 image-%3d.jpg//-r specify the screenshot frequency / / synthesize the picture ffmpeg-f image2-I image%d.jpg output.mp4 / / ffmpeg-f concat-I filelist.txt-c copy output.mp4 / / convert the video to gifffmpeg-I input.mp4-ss 0:0:30-t 10-s 320x240-pix_fmt rgb24 output. Gif//-pix_fmt specifies encoding / / converting the first 30 frames of video to gifffmpeg-I input.mp4-vframes 30-f gif output.gif / / rotating video ffmpeg-I input.mp4-vf rotate=PI/2 output.mp4// scaling video ffmpeg-I input.mp4-vf scale=iw/2:-1 output.mp4// iw is the width of the input Iw/2 is half. -1 to maintain aspect ratio / / Video variable speed ffmpeg-I input.mp4-filter:v setpts=0.5*PTS output.mp4 / / Audio variable speed ffmpeg-I input.mp3-filter:an atempo=2.0 output.mp3 / / Audio and video change at the same time, but audio and video are reciprocal ffmpeg-I input.mp4-filter_complex "[0filter:v setpts=0.5*PTS output.mp4 v] setpts=0.5*PTS [v] [0output.mp4//] atempo=2.0 [a] "- map" [v] "- map" [a] "output.mp4// video add watermark ffmpeg-I input.mp4-I logo.jpg-filter_complex [0map v] [1map v] overlay=main_w-overlay_w-10:main_h-overlay_h-10 [out]-map [out]-map 0vig a-codec:a copy output.mp4// main_w-overlay_w-10 video Width of-width of the watermark-watermark margin / / capture part of the video ffmpeg-I in.mp4-filter:v "crop=out_w:out_h:x:y" out.mp4// to capture part of the video, starting from the position of [80Magne60], with a width of 20000. Video ffmpeg-I in.mp4-filter:v "crop=80:60:200:100"-CMV a copy out.mp4// truncates 1/4 ffmpeg-I in.mp4-filter:v "crop=in_w/2:in_h/2:in_w/2:in_h/2"-CMV a copy out.mp4// in the lower right corner truncates the bottom 40 pixel height ffmpeg-I in.mp4-filter:v "crop=in_w:in_h-40"-CMV a copy out.mp4

Parameter description:

-vcodec xvid uses xvid compression

-s 320 × 240 specify resolution

-r fps sets the frame rate by default 25

-b specify compression bit

-acodec aac sets sound coding

-ac sets the number of channels. 1 is mono and 2 is stereo.

-ar sets the sound sampling rate. PSP only recognizes 24000.

-ab sets the sound bit rate

-vol sets the volume

-y overwrite the output file

-t duration setting recording time hh:mm:ss [.xxx] format recording time is also supported

-ss position search to the specified time [-] hh:mm:ss [.xxx] format is also supported

-title string sets the title

-author string setting author

-copyright string setting copyright

-hq activates high quality settings

-aspect aspect sets the aspect ratio 4:3 16:9 or 1.3333 1.7777

-croptop size sets the pixel size of the top excision band

-cropbottom size-cropleft size-cropright size

-padtop size sets the size pixel unit of the top complement

-padbottom size-padleft size-padright size-padcolor color sets the color of the complement bar (hex,6 hexadecimal numbers, red: green: blue arrangement, for example, 000000 represents black)

-bt tolerance sets video bitrate tolerance kbit/s

-maxrate bitrate sets the maximum video bitrate tolerance

-minrate bitreate sets the minimum video bitrate tolerance

-bufsize size sets the bit rate to control the buffer size

-vcodec codec enforces the use of codec codec. If copy is used to indicate that the original codec data must be copied

-sameq uses the same video quality as the source (VBR)

-pass n selects the number of times to process (1 or 2). Coding twice is very useful. The first time generates statistics, and the second time generates the exact bit rate of the request

-the record file selected by passlogfile file twice is named file.

-map file:stream sets input stream mapping

-debug prints specific debugging information

This is how to achieve video compression and format conversion in the Serverless shared by the editor. If you happen to have similar doubts, you might as well refer to the above analysis to understand. If you want to know more about it, you are welcome to follow the industry information channel.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.