Hi, how is that?
If you use ffmpeg with copy parameters only, recording to a .ts file, this shouldn't be a problem at all IMHO. Way more straightforward than making a multi-source capture utility for Premiere.
I am debugging our FFMPEG processes continually. Or at least attempting to. I am not the coder and don't know FFMPEG all too well. And from wha I understand there is a lot in FFNPEG. So I am hoping it can still resolve my problems.
We run 4 RTSP sources into server. Our FFMPEG services also do HLS segmentation. This preps for distribution of Http URL to a number of iPads. At the end of the session we have recorded h.264 .ts file.
However ... and the major issue ... for some reason under certain instances we see heavy scarring and tearing intraframe. We don't see frame loss but video can at times have swathes edge to edge across it as it if lost information in the frame.
I have been told it is a bus writing to disk issue. But these systems are now full i7 mac mini's with SSD's installed.
Also note the problem is not consistent. Our system tends to be attached to FoxSports and SkySports subcontracted trucks. The problem generally comes back fully with one truck in particular on one 1 angle during the game.
Any ideas on how FFMPEG could resolve?
The idea of a dedicated capture window in PP to ensure no other system resources are robbing the recording process. It also allows me to be post event immediately into PP timelines. This also opens possibilities about how metadata can be deposited post event into those timelines from my App in iOS.
ffmpeg-related questions are best asked on their mailing list.
(Please provide command line and full uncut console output, and test with the most recent build or git head, otherwise they'll tell you what I've just written...)
There are different reasons for which these frame errors can occur, if it is really bad with a special one of your OB trucks only, chances are something's wrong with its infrastructure. First thing is to make sure the physical layer is okay, then move on to configuration and so forth...
I don't think a capture window inside of Premiere will give you more resources. It just blocks user input for the rest of the machine. Different story. Also I am not exactly sure if you can do edit while capture this way. From a logical point of view this should be possible when writing transport streams (because that's exactly the reason why), but I have no indication of how Premiere will go about this.
Even several years later, handling network based streams is a different story from handling baseband streams for many applications...
!Thanks for this.
I will move to FFMPEG forum and mailing list.
For you .. mediastreamsegmenter does the segmentation. FFMPEG acquires data from camera and another ffmpeg instance records the mp4 to disk. I cant see it being FFMPEG issue but if another tool allows me to smooth out the signaling from truck then I need to investigate .. i.e. a dedicated capture window. However, as you point out that ma not be the solution either.
This sounds rather complicated and prone to error. Why not split the ffmpeg output so you can get the job done in one instance?
Keep in mind that a capture window is yet another process on top of another process, and what you see is not necessarily what you get.