Thanks for the explanation.
Yes, on the Pi4 I handled the raw frame data from /dev/video0 myself, but now switching to the Pi5 I found out that rpicam-vid does everything I need, so for me presently no need to dive that deep into the image processig pipeline, as educational as it would be![;)]( "Wink")
My application (wildlife surveillance, 850nm illumination, several days of runtime, potentially off-grid) requires to save ~90fps monochrome video to disk without frame loss, embedding some information into each video frame.
Rpicam-vid seems to do all of that out-of-the-box, including OpenCV post-processing to embed frame number, shutter speed, gain and date/time into the frames. For further image manipulation/text annotation I am exploring the possibilities to re-compile rpicam-vid myself, in order to mess with the image data before passing them to the encoder.
Cheers!
Yes, on the Pi4 I handled the raw frame data from /dev/video0 myself, but now switching to the Pi5 I found out that rpicam-vid does everything I need, so for me presently no need to dive that deep into the image processig pipeline, as educational as it would be
My application (wildlife surveillance, 850nm illumination, several days of runtime, potentially off-grid) requires to save ~90fps monochrome video to disk without frame loss, embedding some information into each video frame.
Rpicam-vid seems to do all of that out-of-the-box, including OpenCV post-processing to embed frame number, shutter speed, gain and date/time into the frames. For further image manipulation/text annotation I am exploring the possibilities to re-compile rpicam-vid myself, in order to mess with the image data before passing them to the encoder.
Cheers!
Statistics: Posted by 2blmaster — Sat Oct 11, 2025 11:19 am