Page 25 of 43 FirstFirst ... 15232425262735 ... LastLast
Results 361 to 375 of 633

Thread: Dtivo Video Extraction

  1. #361
    Join Date
    Jan 2002
    Posts
    4,809
    Originally posted by edpuffmonster
    gives me about 73MB of output for 400MB of input.

    If you want it, I'll try to get this stream over to you tomorrow.
    Yeah. Cut the first oh 100 meg of the file for me. It expect that the problem will be right around the 90 meg point.

    --jdiner

  2. #362
    wolkowh Guest

    what codec do you use?

    Originally posted by jdiner


    BTW, there will be really about 12 minutes of video too. The codec everyone seems to be using has problems and report everything no matter how long as almost exactly 2:11 seconds every time.

    --jdiner
    What codec do you use and can you post it or a link?
    Howard

  3. #363
    Join Date
    Jan 2002
    Posts
    4,809
    Well it is official. Juppers wins. He sent me the worst clip I have ever seen. 20 of the first 25 chunks in it are completely corrupt. Some of them in brand new ways... How on earth does the DTivo ever actually play this stuff...

    --jdiner

  4. #364
    Join Date
    Mar 2002
    Location
    Asylum
    Posts
    85
    http://www.coolstf.com/mpeg/

    In particular, I noticed this:

    DSS uses QPSK modulation for the satellite downlink, just like DVB, however, it's packets are 127 bytes in length and not 188 bytes. The MPEG-2 specification chose 188 byte packets for compatibility with ATM networks - ATM "frames" are 47 bytes in length, so an MPEG-2 transport stream packet can fit into 4 ATM frames, very useful for moving a transport stream over ATM fibre prior to uplink.

    And this:

    http://www.tek.com/Measurement/cgi-b...Set=television
    Last edited by edpuffmonster; 03-26-2002 at 06:45 PM.

  5. #365
    Join Date
    Jan 2002
    Posts
    4,809
    Originally posted by edpuffmonster
    http://www.coolstf.com/mpeg/
    Very cool. The more docs the merrier and some of that looks like what I will need next...

    NEW NOTE:

    Oh BTW, don't remember if I ever mentioned it here or not. But it has become quite clear that what the DTivo is storing is the pieces and part of a Transport Stream. Most of the MUX'ers out there turn it into a Program stream. My current dev code turns it back into a Transport Stream. (Less overall work etc...)

    But I have no idea... Anyone know if the standard MPEG-2 tools all work with Transport Streams? They "should" as it is simple and common to all transmissions of mpeg-2.

    I suppose I could give an option to mux either way but it would be nice if we could just go the more simple route.

    Oh and the other reason I went that way is that it made the output file 1/5 or so the size of a "badly" mux'ed file. Not a tmpgenc output, but some of the others...

    --jdiner
    Last edited by jdiner; 03-26-2002 at 06:12 PM.

  6. #366
    Join Date
    Jan 2002
    Posts
    4,809
    I asked a couple of people via PMs but I thought that I would ask here. You all know that I am having trouble getting enough TyStreams to be able to really fix everything. I am slowly getting there but... I have asked for donations and will sometime soon have my own DTivo.

    Until then, and perhaps until we all conclude that VSplit is done... does anyone have accesses/control/ownsership of a machine on the net that has a pretty fast link some decent HD space free and a willingness to give me an account. I can and have compiled VSplit to run under vanilla Unix via GCC. So rather than the slow downloads to me I would like to have a staging spot out there somewhere where I can get the clips, run the tests, and keep things moving.

    I BY NO MEANS need root. Just a plain access level shell account with access to a c compiler. If anyone can help that is great. If not we will just keep muddling through as have been.

    MANY THANKS...

    Thanks to all that offered my space and and account on their machine. I have chosen 2, the first 2 that got back to me, and I have begun working on the longer "bad" clips that they provided to me. I have already fixed one bug and am hot on the trail of the next.

    And let me just say that all of you people with DVD-r drives make me very jealous. And no I am NOT asking that you all kick in for one. I'm just stating the obvious, seems like everyone here has one...

    --jdiner
    Last edited by jdiner; 03-26-2002 at 08:00 PM.

  7. #367
    Join Date
    Feb 2002
    Posts
    109
    Joshua,
    I could set you up via ssh on a linux box that has access to the dTiVo. I have plenty of space, lots of stuff on my dTiVo, and a cable modem link. But 2 problems, I don't have my turbonet card yet (it's on order) and my TiVo airnet (802.11b) driver is to unstable for reliable use. You could access and get stuff via 115k serial if that is ok until one of the other problems are solved. The second issue is the reclist.tcl script dies halfway through the NowShowing list, see:
    http://www.dealdatabase.com/forum/sh...threadid=11416

    PM or email me if your interested.

  8. #368
    Join Date
    Jan 2002
    Posts
    4,809
    For those that have bad clips where things get really messed up please PM and I will give you the staging site information. Please keep in mind that at a gig and episode disk space is still limited and that a smaller clip with a problem is most likely just as good...

    --jdiner

  9. #369
    Join Date
    Jan 2002
    Posts
    4,809
    Ok. The next major bug is hammered out. For the first time, as mentioned, I got a truly bad stream. One that was not a decoding error but really bad. (From Juppers)

    In talking with him I discovered that he has major rain fade going on where he is. And it stands to reason with what I can see in the data. According to the Audio/Video timestamps the TyStream lost 4 packets. There are 4 completely junked packets in that "hole". I did a full data dump in there and there is nothing to salvage. I.e. no headers or partial packets. It is just junk.

    This was causing the "from something else" junk detector to go off and get lost. I now have it retraining once the junk ends... I figured out how to catch it "coming back" to something real...

    So that problem is fixed. Sort of. You'll get the rest of the file split nicely. The problem... things can and probably will be off. But in this mode there is not much I can do about it. In a flat file like I am producing there is no way to "hop" the gap and maintain A/V sequence. The reason is in some chunks there are 50 video packets and 2 audio. In others it is the other way around. So it is not a 1-1 loss. Nothing I can do for it. At least nothing I can think of right now. I had thought of empty packets being shoved in. But I have no clear way of know what is missing with going even deeper into the monitoring of the MPEG-2 protocol. Not what I want to do right now.

    Now once we start mux'ing that gap will cause GOP corruption as things are displayed/played. But that will be for a few frames. Not that big a deal and sync will be maintained. (This is the other reason I am not going to go insane trying to fix it...)

    Now this is the first clip this has happened in. Not the junk detection, but something truly missing out of the middle. It will be interesting to see how often it occurs. So as you use alpha 7, which is not being released quite yet, If you should see the "missing block" report, please report it to me. I would like to keep a count to see how common it is.

    --jdiner

  10. #370
    Join Date
    Jan 2002
    Location
    Sonoran Desert
    Posts
    2,829
    jdiner: just out of curiosity, did you find any major differences between the mfs_export'ed stream and the extractstream one that we should all know about? If it is necessary, I could also send you another one which is a multi part ty. (mfs_export'ed streams are 512mb max per segment)
    Before PMing me: Iím not your personal tech support. If you have a question, ask in public so I don't have to repeat if somebody else asks. If you want images or slices, use emule. I will ignore all support PMs.

    Sponsor a vegetarian! I have taken the pledge, how about you?

  11. #371
    Join Date
    Aug 2001
    Location
    S. FLorida
    Posts
    11

    Possible Cause of Errors

    Hey Guys,

    Just saw JDiner mention rain fade as a problem with one of the streams. This got me thinking.... I wonder if satellite signal strength could be a factor in all of the variation in stream results.


    - ird2

  12. #372
    Join Date
    Jan 2002
    Posts
    4,809
    Ok. I got to thinking about things today and I realized I had been kind of stupid about the clips. I don't technically need the whole thing. i just need the part where the error occurs. So I made the following changes:

    1- In Alpha #7 the progress tick marks (.) now alternate. Every 100 ticks, 1000 total 128k chunks in the file, print the number value of the current chunk. So you will see something like:

    .........100.........200.........300.........400.........500.........600
    .........700...

    and so on. By itself this is not much use, but we couple it with #2.

    2- I built a TyStream smart split utility. Called of all things TyFileSplit. (You will find it attached to this message...)

    This tool has the following usage:
    TyFileSplit <infile> <outfile> <starting chunk> <num chunks>

    And as is suggested takes a raw TyStream file as input, output a new raw TyStream file. But it starts the copying out at at the starting chunk. This is a chunk count from above, not a Byte position. It will then copy over to the output up to as many chunk as are specified.

    Now this becomes very nice and heres why. Someone provided me with an DA episode clip that was 1.5 gig long. When you run it through VSplit you get to about 12 minutes into it and it stops working right. Now all I need to see is the problem. So you run it through Vsplit7 and get the following output:

    Processing '../../Darktest.ty': (10 chunks per tick)
    .........100.........200.........300.........400.........500.........600
    .........700.........800.........900.........1000.........1100.........1200
    .........1300.........1400.........1500.........1600.........1700.........1800
    .........1900.........2000....Found an OOB packet...
    The Video Diff is: 00:00:00.950
    Found an OOB packet...
    The Audio Diff is: 00:00:01.8
    Found an OOB packet...
    The Video Diff is: 00:00:01.334

    So the problems location should be more clear at shortly after 2040. But I need enough information for VSplit to come up, so back up to the nearest main block, 2000 in this case, and then clip somewhere between 150 and 250. In this case do 150.

    So you use on unix:

    ./TyFileSplit ../../Darktest.ty dark.ty 2000 150

    or on Windows:

    TyFileSplit ..\..\DarkTest.ty dark.ty 2000 150

    And you get a 18.5 meg file that contains all of the error information and enough of the rest of the data for VSplit to start up and run again reporting the same error very quickly for me. So I can put a fix in place.

    Again not much use until you get Alpha #7... It is on it's way I am just trying to put the finishing touches on the retraining code.

    --jdiner

  13. #373
    Join Date
    Jan 2002
    Posts
    4,809

    Re: Possible Cause of Errors

    Originally posted by ird2
    Just saw JDiner mention rain fade as a problem with one of the streams. This got me thinking.... I wonder if satellite signal strength could be a factor in all of the variation in stream results.
    [/B]
    I don't believe it is. It seems to be all or nothing. You either get data or you get garbage. But like I said it will be interesting to see how often it happens and what it looks like when it does.

    --jdiner

  14. #374
    Join Date
    Mar 2002
    Location
    Asylum
    Posts
    85
    How about a linux binary this time too? ( Since it sounds like you'll be compiling one for your own use anyway... )

  15. #375
    Join Date
    Oct 2001
    Posts
    126

    yeah!

    Linux all the way baby :-)

Posting Permissions

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