We have a situation where we have a large number of incoming HD video streams which we would like to monitor in a way similar to this wall of monitors each showing 16 channels.

This can probably be done by having the channels downscaled and put next to each other by e.g. VLC on a suitably powerful PC but at the expense of a lot of CPU-power. I would like to see if a Raspberry can help for example by being able to display 2x2 or similar or just render the bitstream into a HDMI output. We may look into an off-the-shelf solution for additional combination of multiple HDMI inputs into a single HDMI output. For this example this could be four raspberries each rendering a 2x2 corner, which is then combined by a single 4-input device to a full 4x4 image.

Can the GPU in the Raspberry handle that given suitable omxplayer options? (I solved the "capture and buffer http stream" problem earlier)

wall of monitors each showing 16 channels


Im no expert but I know the Pis network bandwidth isn't super duper, so I'm assuming you'd get away fine with one HD to start from each Pi, then the more videos you cram for a single Pi to render, you'd have to do the downscale, but the downscaling would need to come from the 'processing' server, that the cameras will need to be fed to, downscaled, and piped to the RPi's.

I personally had 4 cameras hooked up to pi to serve out, but the bandwidth easily bottlenecked and got choppy, and sometimes cut out, it was the slow bus line which the most of the components on the pi etc all share. I'd venture out to say that unless newer models of pi address this slower bus issue, then you might need to find solutions in other similar SoC type devices, that might be able to move that much bandwidth.

  • These are around 1,3 Megabytes pr second so four streams is well within what the network interface allows. – Thorbjørn Ravn Andersen Feb 6 '18 at 21:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.