[mythtv-users] mythcutprojectx.sh and trunk (was: mythtranscode cutlist argument?)

[Christopher Meredith]

On Sat, Jul 10, 2010 at 3:33 AM, John Pilkington <J.Pilk at tesco.net> wrote:
> On 10/07/10 08:37, John Pilkington wrote:
>>
>> On 09/07/10 23:43, Christopher Meredith wrote:
>>>
>>> On Fri, Jul 9, 2010 at 4:36 PM, John Pilkington<J.Pilk at tesco.net> wrote:
>>>>
>>>> On 09/07/10 22:00, Christopher Meredith wrote:
>>>>>
>>>>> According to the help output, mythtranscode can be used with the
>>>>> --infile and --honorcutlist flags and the latter "Optionally takes a
>>>>> cutlist as an argument when used with --infile." Can anyone tell me
>>>>> what format this argument should be? I've tried creating a plaintext
>>>>> file with this:
>>>>>
>>>>> 1836,13700-18207,29554-36778,50077-54307,55801-
>>>>>
>>>>> and passing the text file as an argument, and I've tried passing that
>>>>> whole string as an argument. In both cases, the file is transcoded but
>>>>> no cuts are honored. How should this argument be formatted?
>>>>>
>>>>> Thanks!
>>>>
>>>> I've no hard evidence on this, but I suspect you need a
>>>> one-value-to-the-line text file, possibly starting with a zero to
>>>> define the
>>>> polarity. But why not use the default as created by the MythTV editor?
>>>>
>>>> And the usual advice for people having problems with lossless
>>>> transcoding of
>>>> a perhaps imperfect mpeg2 recording is to use Project-X.
>>>> You might like to look at http://www.mythtv.org/wiki/Mythcutprojectx.
>>>> I have a report that the effect of this (and mythtranscode) is at
>>>> present
>>>> inverted in trunk; it works in 0.23-fixes for me.
>>>>
>>>> Don't overlook the searchable archive at
>>>> http://www.gossamer-threads.com/lists/mythtv/
>>>
>>> I tried modifying the file thus:
>>> 0-1836
>>> 13700-18207
>>> 29554-36778
>>> 50077-54307
>>> 55801-
>>>
>> Again, no guarantees, but this looks like two values to the line to me
>>
>> 0
>> 1836
>> 13700
>> 18207
>> ...etc
>>
>>> to no avail. I am trying to compensate for a busted mpeg2 lossless
>>> transcode as you suspect. I checked out mythcutprojectx and am having
>>> some issues getting it to work (ionice: execvp failed: No such file or
>>> directory) and am hesitant to spend too much time with it because as I
>>> understand, it requires user intervention which means I can't make it
>>> part of a userjob.
>>
>> True: but I get my cutlist interactively too. The ionice -c3 stuff isn't
>> essential to the operation but it stops resource hogging. Maybe someone
>> could build the core algorithms, perhaps without PID selection, into the
>> transcode wrapper stub.
>>
> Sorry to add to my own post, but the interactive calling of the script is
> needed only if the required PIDs are not known in advance.  For me they are
> usually linked directly to the channel number and they could be passed as
> arguments to a user job.

I played around with mythcutprojectx.sh some more and there are some
additional problems I'm finding. First of all, it seems "hardcoded"
for only mpeg2 audio even though many DVB and firewire recordings will
have ac3 audio. Line 143 is the remuxing line and it assumes the audio
track will be mpeg2. If you use it on a file with ac3 audio, it dies
at this stage.

Perhaps the more important problem is that it seems to be working in
reverse. In both tests I ran, it cut out the program, leaving ONLY the
commercials. I've seen other threads on trouble with lossless
commercial cutting in trunk (I'm running trunk, by the way) and I
suspect this is related.

I think I can probably modify the script to account for ac3 audio as
well as hacking together a way to autodetect the video and audio
stream IDs. But the inverse cutting is really fatal.