Dear Forum
I work on a proprietary app which runs on a GPU-equipped server and renders images which are currently sent to web (local or remote) clients as on-demand still frames. We are looking to enhance the rendering to be more interactive and continuous, and I am interested to know if OBS has components that could be used for this.
I have played with it in its regular consumer mode to capture a desktop and stream it via Twitch, and it seems to work very well, at least in terms of image quality (the latency via Twitch is huge, but I guess that doesn't matter for that application).
Architecturally, I am (perhaps naively) envisaging the minimal amount of OBS being compiled into the app itself, capturing the off-screen GL render buffer, using NVENC to compress it, and sending RTMP directly, perhaps via our existing separate web server process (if it can be adapted to handle RTMP) or a separate one (perhaps nginx with the RTMP plug-in).
I would appreciate a reality check on this idea from any of you with high-level knowledge of the OBS architecture, and what sort of minimum latency could be expected from such a set-up (web client on same fast local network).
Thanks!
I work on a proprietary app which runs on a GPU-equipped server and renders images which are currently sent to web (local or remote) clients as on-demand still frames. We are looking to enhance the rendering to be more interactive and continuous, and I am interested to know if OBS has components that could be used for this.
I have played with it in its regular consumer mode to capture a desktop and stream it via Twitch, and it seems to work very well, at least in terms of image quality (the latency via Twitch is huge, but I guess that doesn't matter for that application).
Architecturally, I am (perhaps naively) envisaging the minimal amount of OBS being compiled into the app itself, capturing the off-screen GL render buffer, using NVENC to compress it, and sending RTMP directly, perhaps via our existing separate web server process (if it can be adapted to handle RTMP) or a separate one (perhaps nginx with the RTMP plug-in).
I would appreciate a reality check on this idea from any of you with high-level knowledge of the OBS architecture, and what sort of minimum latency could be expected from such a set-up (web client on same fast local network).
Thanks!