evilclive
9th September 2007, 14:56
I am aware of a consensus that if you are backing up a DVD-quality film (MPEG-2, 720x576) to DivX or XVid, you should not try to squeeze it onto one CDR if it is more than about 90-100 minutes long.
If we take this as given, is there a similarly strong opinion about the results if a three hour recording (MPEG-2, 720x576) is compressed onto a single-layer DVDR (at roughly 3Mbit/s ABR video plus 256kbps AC3 audio)?
I'll tell you why I'm asking. My parents have just bought a MPEG-2 DVD recorder with UK Freeview tuner, and have asked me to recommend a one-size-fits-all recording mode. The sensible options are:
"SP": 2 hours per DVD-RW (720x576 pixels @ 4.8 Mbit/s ABR)
"LSP": 2.5 hours per DVD-RW (720x576 pixels @ 3.9 Mbit/s ABR)
"ESP": 3 hours per DVD-RW (720x576 pixels @ 3.1 Mbit/s ABR)
"LP": 4 hours per DVD-RW (352x576 pixels @ 2.3 Mbit/s ABR)
The native bit-rate of UK Freeview streams works out at about:
BBC1: 130 minutes per 4.7GB (720x576 pixels)
BBC2: 160-180 minutes per 4.7GB (720x576 pixels)
Other BBC channels: 180-200 minutes per 4.7GB (720x576 pixels)
Non-BBC channels: 240-300 minutes per 4.7GB (varies between 720x576, 544x576 or 352x576 pixels)
The limiting factor of the source bit-rate means there isn't that much point using SP or an even higher setting, except perhaps when recording BBC1.
I've made a few test recordings, and my subjective opinions of the results are:
"LSP": Indistinguishable from the original
"ESP": Nearly transparent, but not quite
"LP": Very blocky
However, I don't trust my results. They were taken from different clips of different TV programmes, and I suspect that the lower resolution of the LP recording should compensate for the lower bit rate, and make its quality similar to that of the ESP one.
My question: If I take a random minute-long video stream (720x576 MPEG-2 @ 9 Mbit/s, extracted from a commercial DVD), and recode it in TMPGEnc Plus 2.5 at CQ=65, is it a sensible result if the output file at 720x576 pixels is about 33% larger than the output file at 352x576 pixels? Is that roughly how a MPEG-2 video elemental stream at fixed CQ should scale with increasing number of pixels?
If so, then my result from the "ESP" test recording was made better by the random source data being easier to encode than the source data from the "LP" test; and I'll recommend the "LSP" setting.
If we take this as given, is there a similarly strong opinion about the results if a three hour recording (MPEG-2, 720x576) is compressed onto a single-layer DVDR (at roughly 3Mbit/s ABR video plus 256kbps AC3 audio)?
I'll tell you why I'm asking. My parents have just bought a MPEG-2 DVD recorder with UK Freeview tuner, and have asked me to recommend a one-size-fits-all recording mode. The sensible options are:
"SP": 2 hours per DVD-RW (720x576 pixels @ 4.8 Mbit/s ABR)
"LSP": 2.5 hours per DVD-RW (720x576 pixels @ 3.9 Mbit/s ABR)
"ESP": 3 hours per DVD-RW (720x576 pixels @ 3.1 Mbit/s ABR)
"LP": 4 hours per DVD-RW (352x576 pixels @ 2.3 Mbit/s ABR)
The native bit-rate of UK Freeview streams works out at about:
BBC1: 130 minutes per 4.7GB (720x576 pixels)
BBC2: 160-180 minutes per 4.7GB (720x576 pixels)
Other BBC channels: 180-200 minutes per 4.7GB (720x576 pixels)
Non-BBC channels: 240-300 minutes per 4.7GB (varies between 720x576, 544x576 or 352x576 pixels)
The limiting factor of the source bit-rate means there isn't that much point using SP or an even higher setting, except perhaps when recording BBC1.
I've made a few test recordings, and my subjective opinions of the results are:
"LSP": Indistinguishable from the original
"ESP": Nearly transparent, but not quite
"LP": Very blocky
However, I don't trust my results. They were taken from different clips of different TV programmes, and I suspect that the lower resolution of the LP recording should compensate for the lower bit rate, and make its quality similar to that of the ESP one.
My question: If I take a random minute-long video stream (720x576 MPEG-2 @ 9 Mbit/s, extracted from a commercial DVD), and recode it in TMPGEnc Plus 2.5 at CQ=65, is it a sensible result if the output file at 720x576 pixels is about 33% larger than the output file at 352x576 pixels? Is that roughly how a MPEG-2 video elemental stream at fixed CQ should scale with increasing number of pixels?
If so, then my result from the "ESP" test recording was made better by the random source data being easier to encode than the source data from the "LP" test; and I'll recommend the "LSP" setting.