[mythtv-users] Lossless Transcode 0.25

[Scott and Nicole Harris]

>>>>> Is it just me, or is the lossless transcode in 0.25 RIDICULOUSLY fast?
>>>>> I queued up 3 transcodes and by the time I finished editing another
>>>>> cutlist for transcode (and I’m pretty quick at blowing through a
>>>>> cutlist), all 3 were done.
>>>> Of course it's ridiculously fast! It doesn't do much of anything. Aside
>>>> from re-encoding up to a few dozen frames for each cutpoint, all you've
>>>> got is some very minor container reprocessing, and a file copy. Out of
>>>> an hour recording, you're looking at a few seconds worth of video that
>>>> needs processing.
>>> I don't recall it being that fast previously. In fact, it's why I 
>>> decided
>>> to use the frontend in our gym as a SBE to offload transcode jobs to 
>>> because
>>> they took so long. They took quite some time on the SBE too.

>>There haven't been any changes that should make it significantly faster
>>or slower. My lossless transcodes have always been disk IO limited
>>rather than CPU limited. Offloading to a remote machine that would have
>>to pull content off a network share, and push the result back to the
>>same network share, would have been all that much slower.

>Actually, I have discovered something a little disconcerting. After 
>watching some shows that I transcoded with 0.25 last night and tonight (some 
>of them recorded with 0.25 and some on 0.24), there are flashes of vertical 
>purple lines on the screen approximately every minute and at every 
>commercial transition that was removed. The same issues are not present 
>on the recordings before they have been transcoded. I tested playback on 5 
>different frontends, different recordings, etc and it's pretty consistent; 
>anything transcoded with 0.25 in the last two days has vertical purple line 
>flashes every minute or so; everything else is fine.

>It sounds a lot like this http://code.mythtv.org/trac/ticket/10110 except 
>these are lossless transcodes. Also, the same issue here 
>http://www.gossamer-threads.com/lists/mythtv/users/495911?do=post_view_threaded#495911 
>indicates that an ffmpeg update fixed it for him to the point that the 
>affected recording were no longer affected, indicating it was a playback 
>issue. 4 of my frontends are Mythbuntu 11.10 fresh installs and one is a 
>pretty old 10.04 install that has never displayed this behavior before these 
>0.25 transcodes. 
When reading through ticket #10110, it concludes with "Closing at reporter request. As an aside, this was likely due to the Debian external libmpeg2 insanity.".  I am not entirely sure what the Debian external libmpeg2 insanity is, or what I can try to do to resolve this?  It's pretty bad, it is affecting every single transcode I do to the point that I probably won't be able to do any transcodes other than to try to test the result of possible resolutions.