[mythtv-users] Strange time counter issue with recent NBC OTA recordings

[Tom Dexter]

On Mon, Dec 9, 2013 at 9:57 AM, Jim Stichnoth <stichnot at gmail.com> wrote:
> On Sun, Dec 8, 2013 at 2:30 PM, Tom Dexter <digitalaudiorock at gmail.com>
> wrote:
>>
>> On Sun, Dec 8, 2013 at 2:11 PM, Jim Stichnoth <stichnot at gmail.com> wrote:
>> > On Sun, Dec 8, 2013 at 9:58 AM, Tom Dexter <digitalaudiorock at gmail.com>
>> > wrote:
>> >>
>> >> On Sat, Dec 7, 2013 at 12:48 PM, Jim Stichnoth <stichnot at gmail.com>
>> >> wrote:
>> >>
>> >> > It would be enlightening if you could dump the recording markup with
>> >> > "mythutil --getmarkup markup.xml --chanid <chanid> --starttime
>> >> > <starttime>"
>> >> > and post/attach the resulting markup.xml file.
>> >> >
>> >> > Jim
>> >>
>> >> I sent that as an attachment.  Apparently it's awaiting list moderator
>> >> approval because of the size of the attachment.
>> >
>> >
>> > Pretty sure there is no moderation. :)  Try a pastebin.com link, or feel
>> > free to email me directly.
>> >
>> > Jim
>>
>> Ah....ok.  Here it is.
>>
>> http://pastebin.com/x6CiDUhH
>
>
> Offhand I don't see any problems with the markup/seektable.  Here are some
> interesting bits:
>
>       <mark frame="0" type="33" data="3539156"/>
> The recording is 3539.156 seconds long, i.e. 59 minutes.
>
>       <mark frame="0" type="34" data="89908"/>
> The recording contains 89908 frames.
>
>       <mark frame="1" type="12"/>
> The recording is identified as 16:9 aspect ratio.
>
>       <mark frame="1" type="30" data="1920"/>
>       <mark frame="1" type="31" data="1080"/>
>       <mark frame="1" type="32" data="29970"/>
> The recording is identified as 1920x1080 at 29.97, i.e. 1080i.
>
> The interesting bit is that frames/framerate = 89908/29.97 = 3000s = 50
> minute length, which is what 0.25 incorrectly reports.
>
> The bulk of the file is seektable data, e.g.:
>       <seek frame="15" type="9" data="1050920"/>
>       <seek frame="15" type="33" data="501"/>
> This says that frame 15 is a keyframe (only keyframes are listed here) and
> that it is 1050920 bytes and 501ms from the start of the file.
>
> Looking at the differences from one seektable entry to the next, it seems
> that the framerate is quickly and mostly randomly bouncing between 24fps and
> 30fps.  This is most likely due to insertion of the repeat_pict flag which
> tells the player to show the frame for an extra half frame interval.
>
> All in all, it looks in order and is exactly what the 0.27 recorder and
> player (including mythcommflag --rebuild) changes were designed to handle.
>
> In rereading the thread, it seems that at this point you're saying that with
> this rebuilt seektable, during playback it always shows a length of 52
> minutes, right?  I have no explanation for that.  It would be very helpful
> to get a copy of the 6GB recording file to try to reproduce the issue.
> (Given your gmail account, you could probably upload it using Google Drive
> and share it as you like, if you have the time/capacity/patience.)
>
> Jim
>

Yes, that program currently shows a length of 52 minutes in the
frontend.  I'll see if I can get that thing up on Google Driver (or
somewhere) and reply after that.  Thanks!

Tom