Hi Mike
Perhaps the P-frame is really that big when sent from the source? I've used Wireshark (http://www.wireshark.org/) to capture the raw network packets, then go to "Statistics", "RTP", "Analyse Streams..." to get an insight what the source is *really* sending... E.g. it turns out that my source was sending flawless audio, but the video stream had corrupted packets every now and then... Roland PS: One way to rule out the "mash-together" theory could be to have a look at the big P-frame with a hex-editor. Look for sequences like 00 00 01 xx. These are start markers for MPEG-2. Make a list of these markers and look them up in the MPEG-2 spec in regards to what they mean (e.g. xx==b3 is sequence_header, xx==00 is picture_start, xx==01-af is slice_start). If you see multiple picture starts or the same slice code repeat, then indeed the frame is mashed together. From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of [EMAIL PROTECTED] Sent: Wednesday, July 30, 2008 7:58 AM To: [EMAIL PROTECTED] Subject: [Live-devel] Very large P-frames in a recording All We're working with an embedded system that receives an MPEG-2 transport stream from a TI DSP. We've built an application that uses the live555 libraries to write this stream out to disk. Every once in a while a very large P-frame, on the order of four times the size of a typical P-frame in the stream, is written out to the file. It's very intermittent -- it might happen once out of 6,000 frames. Has anyone else seen this happen? We're beginning to dig into this. We're concerned that our application might be mashing together several P-frames' worth of data, or that we are not using the live555 library correctly. I've reviewed the last several months of the mailing list, but didn't see anything that seemed relevant. Thanks for your help. -=- Mike Miller Rockwell Collins, Inc. Cedar Rapids, IA
_______________________________________________ live-devel mailing list live-devel@lists.live555.com http://lists.live555.com/mailman/listinfo/live-devel
