If you have dynamically generated h264 frames on one end (server), how would you go about making a real time (low latency) video stream viewable in browser (client) ?
In other words, what I'm trying to achieve is streaming graphically intensive c++ application from gpu server to browser.
As a simple solution, I'm thinking of something like this:
c++ app | ffmpeg | server | html5 client
h264 frames created by application are sent to standard output, where ffmpeg is used to remux this stream into mp4 that is passed to server, which upon request can pass it to client.
Is this a good approach ? Is that even possible to create a low latency 30fps video stream using this approach ?