This repository contains three streaming demos all focused around video processing. All the demos below will assume you have a video file locally that you would like to process. I used the Big Buck Bunny movie from the Blender Foundation. You may use whatever video you'd like assuming its in a supported format for the APK released ffmpeg.
The bash-streaming template I'm using here is from https://github.com/alexellis/openfaas-streaming-templates
Preview is a function that generates a quick 15s webm preview of the video you provide. It is a great use case for streaming because ffmpeg will finish processing once its reached 15s of video and will terminate the connection before the entire video has even been received.
faas up -f preview.yml --build-arg ADDITIONAL_PKG="ffmpeg"
curl -SLsfv http://$GATEWAY/function/preview --data-binary @$VIDEO_PATH > preview.webmGenerates a contact-sheet from the provided video. This uses ffmpeg to generate a keyframe thumbnail every 1m of video and then they are put together into a collage using imagemagick's montage tool.
The size of the resulting image scales directly with the length of the video you pipe in.
faas up -f contact-sheet.yml --build-arg ADDITIONAL_PKG="ffmpeg imagemagick"
curl -SLsfv http://$GATEWAY/function/contact-sheet --data-binary @$VIDEO_PATH > contact_sheet.png
Re-encodes the video with the provided image as a watermark in the upper left corner of the video. We're
using this technique with ffmpeg In this
example we're using the CGI Environment Variables injected by the of-watchdog to allow arguments to be passed in to the script.
There are two query params that can be passed:
| Param | Type | Description | Default |
|---|---|---|---|
opacity |
Float | Sets the opacity of the watermark | 0.7 |
image |
String | URL to the image asset to use | "https://avatars2.githubusercontent.com/u/27013154?s=50&v=4" |
faas up -f watermark.yml --build-arg ADDITIONAL_PKG="ffmpeg"
curl -SLsfv http://$GATEWAY/function/watermark?opacity=0.5&image=https://emojipedia-us.s3.dualstack.us-west-1.amazonaws.com/thumbs/240/apple/198/thinking-face_1f914.png --data-binary @$VIDEO_PATH > watermarked.mkvWill result in a thinking face overlaid as a watermark on your video.
Because video processing can take a while, you might need to increase some settings on your gateway to allow long running connections. For my demo processing BigBuckBunny I set the following for my gateway:
readTimeout=305s
writeTimeout=305s
upstreamTimeout=300s
Because I am streaming the result video I'm using a container that does not require end of write metadata. I
am using the Matroska container format for the response, however the codec used should be the same as the
input provided for all use cases except Preview