Hah. Pick your poison eh?
When i first started the project, I was capturing video and audio separately, and then merging them together with FFMPEG. There was a noticeable drift. And it was computationally expensive on the PI to do so where it was going to create a backlog of work unless I only did it when motion actually occurred. . So when I saw I could record to a MP4 with Audio, that was Awesome! I just wish there was a way to setup a 30 second circular buffer and then just say give me video AND audio from timestamp x to y, and dump it to a MP4 file. It could adjust the timestamp requests perhaps and tell me what it could give me around any key frames, and that way I would make sure my follow up request wouldn't have any overlap,under lap. That would be the ticket! No missing frames / audio (hopefully). Right now, starting/stopping encoders and recording causes my video to look slightly pulsating to jerky or like a bad film splice job of the 1970's. I'm sure it's going to rival the profanity overdubs to when I need to extract audio for text to speech processing. "Shut the FRONT DOOR".
So anyhow, I'll will do some experimenting and see what kind of solution I can make with this. I'm just surprised that there isn't a real solid solution for delivering high quality forever video/audio in digestible chunks.
When i first started the project, I was capturing video and audio separately, and then merging them together with FFMPEG. There was a noticeable drift. And it was computationally expensive on the PI to do so where it was going to create a backlog of work unless I only did it when motion actually occurred. . So when I saw I could record to a MP4 with Audio, that was Awesome! I just wish there was a way to setup a 30 second circular buffer and then just say give me video AND audio from timestamp x to y, and dump it to a MP4 file. It could adjust the timestamp requests perhaps and tell me what it could give me around any key frames, and that way I would make sure my follow up request wouldn't have any overlap,under lap. That would be the ticket! No missing frames / audio (hopefully). Right now, starting/stopping encoders and recording causes my video to look slightly pulsating to jerky or like a bad film splice job of the 1970's. I'm sure it's going to rival the profanity overdubs to when I need to extract audio for text to speech processing. "Shut the FRONT DOOR".
So anyhow, I'll will do some experimenting and see what kind of solution I can make with this. I'm just surprised that there isn't a real solid solution for delivering high quality forever video/audio in digestible chunks.
Statistics: Posted by UltimateCodeWarrior — Thu Dec 19, 2024 5:44 pm