Page 18 of 70 FirstFirst ... 816171819202868 ... LastLast
Results 256 to 270 of 1041

Thread: MUX'ing, VSplit, and MPG2 files.

  1. #256
    Join Date
    Jun 2002
    Location
    Was Frozen North now Sunny South
    Posts
    351
    Originally posted by jdiner
    ...I have had one full 1 hour clip make it start to finish...
    Whew, I was looking for this & am glad to see it. It seems that sync is lost over multiple FSIDs.

    For my 1/2 hour Twilight Zone episodes (low-res B&W and less than one FSID long) once I get the start right (in sync) they appear to stay that way. With multi-FSID programs they RARELY stay that way!

    Oh, and I should add that this sync loss on longer programs (which I believe occurs at FSID changes) happens without my doing any editing at all. I can get the sync right at the start, but it's lost, and appears to get worse, at each step of an FSID location change.
    Last edited by laserfan; 10-02-2002 at 11:09 AM.
    Philips Standalone v3.01 w/2-80G drives and Tivonet.

  2. #257
    Join Date
    May 2002
    Posts
    234
    in the little bit of muxing I have attempted I notice that if I split/mux the entire recording I get much worse sync then if I skip the first little bit (where the program break happens)

  3. #258
    Join Date
    Jan 2002
    Posts
    4,809
    Originally posted by laserfan
    Whew, I was looking for this & am glad to see it. It seems that sync is lost over multiple FSIDs.

    For my 1/2 hour Twilight Zone episodes (low-res B&W and less than one FSID long) once I get the start right (in sync) they appear to stay that way. With multi-FSID programs they RARELY stay that way!

    Oh, and I should add that this sync loss on longer programs (which I believe occurs at FSID changes) happens without my doing any editing at all. I can get the sync right at the start, but it's lost, and appears to get worse, at each step of an FSID location change.
    Hummm. Now that is interresting. At an FSID change there is often garbage. But not always. That problem is what VSplit does it's best to fix. But it sounds like some is still there... Although I using the oft posted VDub check I don't see that with test streams. So.... I have no idea at the moment.

    --jdiner

  4. #259
    Join Date
    Jan 2002
    Posts
    4,809
    Originally posted by Pr.Sinister
    i say we all chip in 5$ so you can buy that newer manual
    from the ISO group.

    I mean this is so frustrating for some of us that we don't
    care what it costs anymore.
    It is very frustrating... I can do almost anything with the data at this point. I know what is where. How it is packaged etc... No problems, no worries. But then I have to take what we ahve and force it into the packaging needed for an as-yet-undocumented standard... Without know what that standard is I have little hope of just pulling it off.

    I have spent more than a hundred hours searching the web, pouring over everything I can find. But there is little to no information there. A few sources for programs that don't seem to work right anyway. So what's next?

    The docs seem the best bet to me, but I have heard some disturbing things from people who have newer versions that I do that there is still nothing about VBR in them. It would be a shame to have people spend that kind of money to get no closer.

    --jdiner

  5. #260
    Join Date
    Jun 2002
    Location
    Was Frozen North now Sunny South
    Posts
    351
    I hope I haven't misled you...this opinion is really just a guess based on observation, and no disciplined examination of the program material. I probably should, but I haven't tried to determine exactly where the sync loss occurs & match this up with the FSID breaks.

    It occurs to me now to ask, perhaps naively, what tool might I use to extract FSIDs individually? I should really try to extract and process them separately, and then concatenate them later, to see if this fixes my supposed problem. Does TyTool have an option to do this kind of extraction (i.e. a tystream mode option)?
    Philips Standalone v3.01 w/2-80G drives and Tivonet.

  6. #261
    Join Date
    May 2002
    Location
    West Hartford CT
    Posts
    620
    Originally posted by jdiner
    Hummm. Now that is interresting. At an FSID change there is often garbage. But not always. That problem is what VSplit does it's best to fix. But it sounds like some is still there... Although I using the oft posted VDub check I don't see that with test streams. So.... I have no idea at the moment.

    --jdiner

    Perhaps I'm WAY off base here... as I'm out of my league in this area. But I have been extracting, and edtting out commericals - without losing audio sync.

    here is how I do it.

    Tytool(5r2)

    Vsplit12 to video/audio streams

    use BeSweet to Upsample audio to 48khz

    mplex to mux -> .mpg file

    import into MPEG2VCR using the mpeg GOP fixer utility -

    run to scan and fix errors on the .mpg

    then load the file into MPEG2VCR's editor & cut/trim out commericals.

    "record" or save to a new file and Viola!

    what appears to be a perfect in-sync stream, without commercials. I can then use Spruce-Up /DVD Patcher to burn a disc.

    Keep in mind I haven't done allot of these yet, just started playing with MPEG2VCR a few days ago. But never had much success until I used the GOP fixer. Seems critical in the process to keep things in sync.

    What I noticed was that if I played the mpg file with media player prior to the MPEG2VCR step, it appeared fine. Yet without the GOP fixer step, as soon as I started to edit the file in MPG2VCR it was obvious that the autio was out of sync.

    Not sure if any else has used this process, but it seems to be working for me. The editting is not SMOOTH -- frame by frame stepping is tough, but for choping commericals out it's ok.

    perhaps this is what is needed? the GOP fixer? it found & fixed over 3400 errors (PTS i think) in one 1/2 hour file.

    it will also fix the GOP size problem for DVD compliance if needed. But I haven't used that 3rd pass repair process yet. Not completely sure what or why it's needed. ???
    Last edited by Hi8; 10-02-2002 at 02:44 PM.
    ~Hi8

    (4) Hughes SD DVR40
    (1) Hughes HR10-250
    (2) xbmc XBOX X2 & xbit

  7. #262
    Join Date
    Jan 2002
    Posts
    4,809
    Originally posted by Hi8
    Keep in mind I haven't done allot of these yet, just started playing with MPEG2VCR a few days ago. But never had much success until I used the GOP fixer. Seems critical in the process to keep things in sync.
    Hummm. I will have to take a look at this and see what it does. It could be a usefull technique to include...

    --jdiner

  8. #263
    Join Date
    Jan 2002
    Posts
    4,809
    Originally posted by laserfan
    I hope I haven't misled you...this opinion is really just a guess based on observation, and no disciplined examination of the program material. I probably should, but I haven't tried to determine exactly where the sync loss occurs & match this up with the FSID breaks.

    It occurs to me now to ask, perhaps naively, what tool might I use to extract FSIDs individually? I should really try to extract and process them separately, and then concatenate them later, to see if this fixes my supposed problem. Does TyTool have an option to do this kind of extraction (i.e. a tystream mode option)?
    Use TyTool and and the "Get Parts" button. It will open a small window with each FSID listed seperately. Pick just 1 and extract. Then repeat and pick the next one, and so one. You will get 1 file for each FSID. Doing this is the same as doing mfs_export, or mfs_stream for the matter, with the rest of the strange and sometimes hard to use mechanisms to extract it...

    --jdiner

  9. #264
    Join Date
    Jan 2002
    Posts
    4,809
    Hummm. You know... I took a break from the computer to think about things and I just just a kind of a thought...

    With CBR the data rate is supposed to be darn near constant and for the most part it is. The fluctations tend to center around the right point.

    With VBR this is not the case. Things fluctuate wildly and this is problemtic. But... it is still based on "bytes per second" moving through the playback system. Who says it has to be constant? And who says the SCR values have to be linear. From observation they usually are but this is not a rule as far as I can see.

    So I started thinking. For the DTivo and a VBR source Change the process... Gather up 1 seconds worth of audio and video data. then calculate the total number of bytes. Then for the PACKs that make up that 1 second segment fill the SCR values in so that whether it is smaller or larger than "average" at the end of the second the SCR represents 1 second has "past" for that number of bytes.

    Given the a seconds worth of both changes from roughly 150k to 800k it would be pretty simple to buffer that much up in memory. And in so doing the problem of "getting off" either too far ahead or too far behind is gone.

    What do you all think? Sound reasonable? Am I missing something?

    --jdiner

  10. #265
    Join Date
    Jul 2001
    Posts
    657
    Originally posted by jdiner
    ...
    What do you all think? Sound reasonable? Am I missing something?

    --jdiner
    Sounds reasonable to me!

  11. #266
    Join Date
    Jun 2002
    Location
    Was Frozen North now Sunny South
    Posts
    351
    Originally posted by Hi8
    ...without losing audio sync...here is how I do it....never had much success until I used the GOP fixer...it found & fixed over 3400 errors (PTS i think) in one 1/2 hour file...
    I have used the GOP fixer in MPEG-VCR, but it has not resolved sync issues in long programs, though I had been very hopeful . You mention a 1/2 hour file--have you been successful w/any programs >1.5 hours? I have done a similar process to yours, but use MPEG-VCR for muxing (haven't tried mplex). I guess I have to try mplex, tho this is all so tiring (been thru bbmpeg and tmpgenc and another that escapes me at this moment).
    Philips Standalone v3.01 w/2-80G drives and Tivonet.

  12. #267
    Join Date
    Feb 2002
    Posts
    54
    Clarification question:

    Used the fixing functions in MPEG-VCR successfully on a stream from a [D/SA]? tivo.

    I tried it on a Dtivo stream once, after an hour it wasn't even at 1%. I guess it really didn't like that stream.

    I've only ever heard of it being useful in CBR streams. (But I could be wrong, it happens more often than I care to admit)

  13. #268
    Join Date
    Jan 2002
    Posts
    9
    I'm new to this thread, having just finished reading it all, and would like to make a few comments. First, I think we really should be concentrating our efforts on what's needed to produce high-quality, standards-compliant DVDs. In my experience with that, I find that the data rate from DTivo files averages 4 mb/sec., with typical peaks to 6mb/sec. That allows fitting 2+ hours of video at the full original quality on one disc. The cost of burners and media has dropped dramatically in the past year. Maybe I'm missing something, but I don't see how VCDs and SVCDs have much of a future, except for special-purpose applications.

    My experience with burning DVDs is using DVD Studio Pro on Mac. Like some other DVD authoring programs, it requires the m2a and m2v files as input, not a multiplexed file. I use TyTool on a PC to generate those files, FTP them to my Mac, then burn from them. DVD Studio Pro has no problem multiplexing the files, although it does issue some warnings about finding VOBUs longer than 1.1 seconds. At the end it states that the disc complies to version 1.1 of the DVD spec. I haven't had any sync problems, and the discs work fine in my Pioneer player. DVDSP doesn't generate a separate multiplexed file, but I'd be glad to provide a sample DVD if that would help in figuring out multiplexing.

    I'd also like to be able to edit out commercials, and editing at GOP boundaries would be fine. I'm wondering if it would make sense to write the editor to edit the m2a and m2v files in parallel, rather than edit a multiplexed file. That would be ideal for people like me who don't need the multiplexed file, and people who do need the multiplexed file could do the multiplexing after editing. I have no idea what the implications are as far as making the editor harder or easier to write.

  14. #269
    Join Date
    Jun 2002
    Location
    Was Frozen North now Sunny South
    Posts
    351
    Originally posted by Kythorn
    Clarification question:

    Used the fixing functions in MPEG-VCR successfully on a stream from a [D/SA]? tivo.

    I tried it on a Dtivo stream once, after an hour it wasn't even at 1%. I guess it really didn't like that stream.

    I've only ever heard of it being useful in CBR streams. (But I could be wrong, it happens more often than I care to admit)
    Obviously I'm an SA guy, dunno about DTivo, but last time I checked MPEG-VCR's GOP fixer only works on 720x480 files, not even on 352x480 which are otherwise DVD-compliant. But I'd have guessed that it would even be grayed-out for your Dtivo stream, i.e. not even allow an 'output file' to be selected.

    Needless to say this severely limits its useability. I find very few programs worth recording at that res--most of what I do is at 352X480 and I don't have to fool SpruceUp.
    Philips Standalone v3.01 w/2-80G drives and Tivonet.

  15. #270
    Join Date
    Jan 2002
    Posts
    4,809
    Originally posted by daa
    In my experience with that, I find that the data rate from DTivo files averages 4 mb/sec., with typical peaks to 6mb/sec. That allows fitting 2+ hours of video at the full original quality on one disc.
    What are you recording that has that data on a DTivo. That is pretty close for an SATivo stream. But an average 1 hour episode is 920meg - 1250meg. But if you take 4meg/sec * 3660 seconds == 14,640 meg or 14.6 gig. Ummm that is extreme even for an SATivo which is normally 2.5gig for an hour.


    I'd also like to be able to edit out commercials, and editing at GOP boundaries would be fine. I'm wondering if it would make sense to write the editor to edit the m2a and m2v files in parallel, rather than edit a multiplexed file.
    That is how I will be doing it in my tools. It is the best way to do it. However it is very difficult to work on the files post editing without splitting. Go back a page or 2 and read what I wrote about mux'ing and timestamps. More than simple editing is needed for editing of seperate streams where things are not mux'ed at the same time. Lining them back up would be tedious at best.

    --jdiner

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •