avdmb
21st May 2002, 11:47
In order to solve various issues with SVCD playing on a standalone Pioneer DV-444 DVD player, the miracle solution would be to use a combination of :
. I-author muxer, to solve video blocking issue at low bitrate (see thread http://forum.doom9.org/showthread.php?s=&threadid=21104 and http://forum.doom9.org/showthread.php?s=&threadid=23887).
. bbMPEG muxer with "forced muxrate=6972") to solve A/V sync issue when switching between chapters (see thread http://forum.doom9.org/showthread.php?s=&threadid=25156).
I-Author muxer is not parameterizable. Therefore, the only solution to the dilemma seems to mimic I-Author with the right bbMPEG settings. Hereafter the results of my (not successfull) tests to do this. I first made a test movie with a few chapters (3), with forced low bitrate. Encoding with CCE, VBR 4 passes, bitrate avg 600, min 300, max 2500. I then played with different muxer and authoring settings, and always burned with fullproof DDogg method, before testing on Pioneer DV-444 DVD player. The goal was to find bbMPEG muxer settings to avoid video blockiness.
1) Complete bypass of bbMPEG muxer. I directly muxed with I-Author the outputs of CCE and BeSweet, and authored with VCDXbuild. When playing the 3 chapters in a row, there was no single blockiness issue (of course, it is visually as good as avg 600 bitrate gives !). However, seems there will never be a definitive solution : when swithing between chapters, video blockiness appeared at the beginning of chapters.
I used MPEGanalizzatore to compare movies muxed either with bbMPEG or I-Author. I then tried to set bbMPEG settings to emulate I-Author.
2) I tried various "forced muxrate" values. Notice that this value is equal to bitrate divide by (8*50). MPEGanalizzatore gives for I-Author the following settings :
. Rate bound = 2814400 (corresponding to muxrate=7036)
. Packet muxrate = 2788000 (corresponding to muxrate=6972).
When setting "forced muxrate" to either values in bbMPEG, both "Rate bound" and "Packet muxrate" are set to the same "forced muxrate" value. In both cases, blockiness was present. Maybe the fact that values are different in I-Author is why it works so well, but it seems not possible to set different values in bbMPEG. Notice that when leaving "forced muxrate" to the default value 0, the computed value is 7599 corresponding to a bitrate of 3039600. As outlined by Gerti67, this is probably a bitrate that the DV-444 can not handle.
3) With both I-Author and bbMPEG, the VBV buffer size=112 (224 kB). In bbMPEG options, it is recommended to set Video buffer size to a value higher (default is 230, which is OK). However, MPEGanalizzatore shows that I-Author is using a video buffer size = 200. This is maybe the reason why there is a problem when switching between chapters. Unfortunately, it is not possible to modify this value in I-Author. I tried bbMPEG with Video buffer size = 200 or 230, but in both cases, video blockiness issue was not solved.
4) I-Author sets different audio/video delays than bbMPEG. I played with bbMPEG setting to mimic the ones of I-Author. I tried several combinations, with longs delays (> 400), video delay >, = or < to audio delay. No improvement of video blockiness.
5) I noticed that I-Author is using minimum extra info, thus probably not using SVCD scan offsets ... I tried all variants of bbMPEG settings with/without "Add Scan Offsets" and/or "bbMPEG Align Sequence Headers" and/or "Use Private Stream 2" and/or SVCD/~DVD. No improvement of video blockiness.
6) I noticed that I-Author is using only timestamps on each GOP beginning. By default with bbMPEG, timestamps are on all frames. I set bbMPEG parameter to "I-frames only", but no improvement of video blockiness.
7) The last difference I noticed was that with I-Author, CSPS flag = 1 (Constrained params) while with bbMPEG it is 0 (unconstrained). I have not found a way to change it with bbMPEG, and have no idea if it can have an effect.
That's it for the time being. I am running out of ideas now ...
. I-author muxer, to solve video blocking issue at low bitrate (see thread http://forum.doom9.org/showthread.php?s=&threadid=21104 and http://forum.doom9.org/showthread.php?s=&threadid=23887).
. bbMPEG muxer with "forced muxrate=6972") to solve A/V sync issue when switching between chapters (see thread http://forum.doom9.org/showthread.php?s=&threadid=25156).
I-Author muxer is not parameterizable. Therefore, the only solution to the dilemma seems to mimic I-Author with the right bbMPEG settings. Hereafter the results of my (not successfull) tests to do this. I first made a test movie with a few chapters (3), with forced low bitrate. Encoding with CCE, VBR 4 passes, bitrate avg 600, min 300, max 2500. I then played with different muxer and authoring settings, and always burned with fullproof DDogg method, before testing on Pioneer DV-444 DVD player. The goal was to find bbMPEG muxer settings to avoid video blockiness.
1) Complete bypass of bbMPEG muxer. I directly muxed with I-Author the outputs of CCE and BeSweet, and authored with VCDXbuild. When playing the 3 chapters in a row, there was no single blockiness issue (of course, it is visually as good as avg 600 bitrate gives !). However, seems there will never be a definitive solution : when swithing between chapters, video blockiness appeared at the beginning of chapters.
I used MPEGanalizzatore to compare movies muxed either with bbMPEG or I-Author. I then tried to set bbMPEG settings to emulate I-Author.
2) I tried various "forced muxrate" values. Notice that this value is equal to bitrate divide by (8*50). MPEGanalizzatore gives for I-Author the following settings :
. Rate bound = 2814400 (corresponding to muxrate=7036)
. Packet muxrate = 2788000 (corresponding to muxrate=6972).
When setting "forced muxrate" to either values in bbMPEG, both "Rate bound" and "Packet muxrate" are set to the same "forced muxrate" value. In both cases, blockiness was present. Maybe the fact that values are different in I-Author is why it works so well, but it seems not possible to set different values in bbMPEG. Notice that when leaving "forced muxrate" to the default value 0, the computed value is 7599 corresponding to a bitrate of 3039600. As outlined by Gerti67, this is probably a bitrate that the DV-444 can not handle.
3) With both I-Author and bbMPEG, the VBV buffer size=112 (224 kB). In bbMPEG options, it is recommended to set Video buffer size to a value higher (default is 230, which is OK). However, MPEGanalizzatore shows that I-Author is using a video buffer size = 200. This is maybe the reason why there is a problem when switching between chapters. Unfortunately, it is not possible to modify this value in I-Author. I tried bbMPEG with Video buffer size = 200 or 230, but in both cases, video blockiness issue was not solved.
4) I-Author sets different audio/video delays than bbMPEG. I played with bbMPEG setting to mimic the ones of I-Author. I tried several combinations, with longs delays (> 400), video delay >, = or < to audio delay. No improvement of video blockiness.
5) I noticed that I-Author is using minimum extra info, thus probably not using SVCD scan offsets ... I tried all variants of bbMPEG settings with/without "Add Scan Offsets" and/or "bbMPEG Align Sequence Headers" and/or "Use Private Stream 2" and/or SVCD/~DVD. No improvement of video blockiness.
6) I noticed that I-Author is using only timestamps on each GOP beginning. By default with bbMPEG, timestamps are on all frames. I set bbMPEG parameter to "I-frames only", but no improvement of video blockiness.
7) The last difference I noticed was that with I-Author, CSPS flag = 1 (Constrained params) while with bbMPEG it is 0 (unconstrained). I have not found a way to change it with bbMPEG, and have no idea if it can have an effect.
That's it for the time being. I am running out of ideas now ...