On Monday 07 Jun 2004 11:35, Luca Abeni wrote: > Hi Samuel, > > [...] > > > Well, finally I've written that code. Using the decoding time to > > calculate the bitrate of the PES stream, the audio bitrate is perfectly > > constant, but the video's does not look constant at all. Plotting the > > result with xgraph you can see a saw with sharp teeth. > > I did some other similar measurements, and my results are similar to > yours... > > I was encoding CBR video at 4000kbps, MPEG1 layer 2 audio at 256kbps, > and I set the TS bitrate at 6000kbps. > > After a lot of experiments, I tried to encode CBR video at 4000kbps, > audio at 256kbps, and to set the TS bitrate at 4500, and... It did the > trick! Now, the video bitrate is almost constant - but the ts analyzer > is complaining about buffer overruns and underruns... :(
I found the card was very sensitive to the bitrate parameters used.. if I chose the wrong ones it would start to lock up after a few minutes, and/or have data loss. I spoke to a guy in Philips at one point - he said that there is actually an SDK available that gives sample code etc, and probably explains the contraints on the bitrate settings. I never got round to investigating this further however.. knowing Philips, you'd probably have to pay quite a lot for it.
