Login | Register For Free | Help
Search for: (Advanced)

Mailing List Archive: MythTV: Users

Asus Pundit and Interlaced Output, part III :)

 

 

MythTV users RSS feed   Index | Next | Previous | View Threaded


wdormann at yahoo

Jun 23, 2003, 5:09 PM

Post #1 of 7 (1271 views)
Permalink
Asus Pundit and Interlaced Output, part III :)

OK, It's good to see that I'm not the only one!

Anyway, the SIS driver also has an option for reducing flicker. I have
turned this off to see how it would affect my interlaced output. It
didn't seem to help the video much, though I could definately see an
increase of flicker in horizontal lines in X before mythfrontend starts up.

A PVR250 system does indeed allow you to deinterlace the TV output.
True, the capture is done via hardware, but once that's done you've got
an interlaced MPEG2 file. Playback is done totally by software, just
as it is with RTJPEG and MPEG-4. (well, maybe playback isn't done in
the exact same method, but it is software playback). So if the
Deinterlace playback option is enabled in MythTV, then it does
deinterlace the video.

As I mentioned in a previous post, I did a capture via: cat
/dev/video0 > /tmp/test.mpg
and I would play it back with mplayer
I figure, this way I would eliminate as many variables as possible.
I've tried different resolutions, flicker options, overscan, and all
that but have yet to find a solution that looks *good*. The most
noticeable problem areas were panning scenes. (try pausing) Since the
above test uses MythTV in no way, then I guess it's not at fault.
:) But I think the topic is still relevant to this discussion list.

I currently have deinterlacing enabled in MythTV and it has good
results. The question is whether or not I am losing motion smoothness
as a result of this, compared to "true" interlaced output.

-WD


craigl at begeek

Jun 23, 2003, 6:27 PM

Post #2 of 7 (1207 views)
Permalink
Re: Asus Pundit and Interlaced Output, part III :) [In reply to]

Will Dormann wrote:

> OK, It's good to see that I'm not the only one!
> Anyway, the SIS driver also has an option for reducing flicker. I
> have turned this off to see how it would affect my interlaced
> output. It didn't seem to help the video much, though I could
> definately see an increase of flicker in horizontal lines in X before
> mythfrontend starts up.
>
> A PVR250 system does indeed allow you to deinterlace the TV output.
> True, the capture is done via hardware, but once that's done you've
> got an interlaced MPEG2 file. Playback is done totally by software,
> just as it is with RTJPEG and MPEG-4. (well, maybe playback isn't
> done in the exact same method, but it is software playback). So if
> the Deinterlace playback option is enabled in MythTV, then it does
> deinterlace the video.

the stream may be interlaced deep inside libavcodec, but if you look at
the mythtv code, the only places deinterlacing is referred to is in the
recording side of things, as far as i can tell. and if you look at the
mpegrecorder.cpp, you will find the ChangeDeinterlacer() stubbed out.
it makes sense to deinterlace it while you're recording if you can, i
think. then it only needs to get done once. unfortunately though, this
excludes the hardware encoders.

> As I mentioned in a previous post, I did a capture via: cat
> /dev/video0 > /tmp/test.mpg
> and I would play it back with mplayer
> I figure, this way I would eliminate as many variables as possible.
> I've tried different resolutions, flicker options, overscan, and all
> that but have yet to find a solution that looks *good*. The most
> noticeable problem areas were panning scenes. (try pausing) Since
> the above test uses MythTV in no way, then I guess it's not at fault.
> :) But I think the topic is still relevant to this discussion list.

i don't think its mythtvs 'fault'. it is just a problem with dealing
with interlaced video. i have read other reports though, that when a
pvr250 file is deinterlaced (and in this particular case at least, also
converted to mpeg-4) it looked great. so, i gotta think there is some
merit to deinterlacing it on playback. i think i'm going to try, it
looks reasonably straightforward to incorporate the avcodec deinterlacer
at least.

i think this topic is definitely relevant to mythtv though, although it
isn't actual mythtv code we're dealing with here, its a couple users
talking about their mythtv experiences as a whole.

> I currently have deinterlacing enabled in MythTV and it has good
> results. The question is whether or not I am losing motion smoothness
> as a result of this, compared to "true" interlaced output.

well, again, i'm just not sure. from what i see, i would say you're not
deinterlacing, but i haven't stepped through all the code. i wish
someone who knew would provide a definitive answer though.

cheers

CraigL->Thx();


ijr at po

Jun 23, 2003, 6:52 PM

Post #3 of 7 (1229 views)
Permalink
Re: Asus Pundit and Interlaced Output, part III :) [In reply to]

On Monday 23 June 2003 09:27 pm, Craig Longman wrote:
> > Will Dormann wrote:
> > A PVR250 system does indeed allow you to deinterlace the TV output.
> > True, the capture is done via hardware, but once that's done you've
> > got an interlaced MPEG2 file. Playback is done totally by software,
> > just as it is with RTJPEG and MPEG-4. (well, maybe playback isn't
> > done in the exact same method, but it is software playback). So if
> > the Deinterlace playback option is enabled in MythTV, then it does
> > deinterlace the video.
>
> the stream may be interlaced deep inside libavcodec, but if you look at
> the mythtv code, the only places deinterlacing is referred to is in the
> recording side of things, as far as i can tell. and if you look at the
> mpegrecorder.cpp, you will find the ChangeDeinterlacer() stubbed out.
> it makes sense to deinterlace it while you're recording if you can, i
> think. then it only needs to get done once. unfortunately though, this
> excludes the hardware encoders.

The only time it deinterlaces anything is on playback if the 'Deinterlace
playback' setting is selected.

Isaac


wd at pobox

Jun 23, 2003, 8:55 PM

Post #4 of 7 (1210 views)
Permalink
Re: Asus Pundit and Interlaced Output, part III :) [In reply to]

At 09:52 PM 6/23/2003 -0400, you wrote:
>On Monday 23 June 2003 09:27 pm, Craig Longman wrote:
>> > Will Dormann wrote:
>> > A PVR250 system does indeed allow you to deinterlace the TV output.
>> > True, the capture is done via hardware, but once that's done you've
>> > got an interlaced MPEG2 file. Playback is done totally by software,
>> > just as it is with RTJPEG and MPEG-4. (well, maybe playback isn't
>> > done in the exact same method, but it is software playback). So if
>> > the Deinterlace playback option is enabled in MythTV, then it does
>> > deinterlace the video.
>>
>> the stream may be interlaced deep inside libavcodec, but if you look at
>> the mythtv code, the only places deinterlacing is referred to is in the
>> recording side of things, as far as i can tell. and if you look at the
>> mpegrecorder.cpp, you will find the ChangeDeinterlacer() stubbed out.
>> it makes sense to deinterlace it while you're recording if you can, i
>> think. then it only needs to get done once. unfortunately though, this
>> excludes the hardware encoders.
>
>The only time it deinterlaces anything is on playback if the 'Deinterlace
>playback' setting is selected.


yes, this is the option that I currently have enabled. And there is a
significant difference with it, so it *is* doing something. When enabled,
there are no interlacing effects visible. This is when I'm playing my
PVR-250 encoded MPEG2 video.

-WD


craigl at begeek

Jun 23, 2003, 9:04 PM

Post #5 of 7 (1210 views)
Permalink
Re: Asus Pundit and Interlaced Output, part III :) [In reply to]

Isaac Richards wrote:

>On Monday 23 June 2003 09:27 pm, Craig Longman wrote:
>
>
>>>Will Dormann wrote:
>>>A PVR250 system does indeed allow you to deinterlace the TV output.
>>>True, the capture is done via hardware, but once that's done you've
>>>got an interlaced MPEG2 file. Playback is done totally by software,
>>>just as it is with RTJPEG and MPEG-4. (well, maybe playback isn't
>>>done in the exact same method, but it is software playback). So if
>>>the Deinterlace playback option is enabled in MythTV, then it does
>>>deinterlace the video.
>>>
>>>
>>the stream may be interlaced deep inside libavcodec, but if you look at
>>the mythtv code, the only places deinterlacing is referred to is in the
>>recording side of things, as far as i can tell. and if you look at the
>>mpegrecorder.cpp, you will find the ChangeDeinterlacer() stubbed out.
>> it makes sense to deinterlace it while you're recording if you can, i
>>think. then it only needs to get done once. unfortunately though, this
>>excludes the hardware encoders.
>>
>>
>The only time it deinterlaces anything is on playback if the 'Deinterlace
>playback' setting is selected.
>
thanks for the input isaac.

i think i managed to track down where the deinterlacing is actually
setup, one little innoccuous call to SetVideoFilters(), right?

all the other references to deinterlace where all happening in the
recording side of things, so naturally i assumed it was all happening in
there. i don't have a bt based card to have tested that out.

i'm curious when you say 'the only time it deinterlaces anything', did
you mean for pvr based cards, or is all the other deinterlacing code not
used anymore?

cheers,

CraigL->Thx();


craigl at begeek

Jun 25, 2003, 4:40 PM

Post #6 of 7 (1222 views)
Permalink
Re: Asus Pundit and Interlaced Output, part III :) [In reply to]

Craig Longman wrote:

> Isaac Richards wrote:
>
>> The only time it deinterlaces anything is on playback if the
>> 'Deinterlace playback' setting is selected.
>
> thanks for the input isaac.
>
> i think i managed to track down where the deinterlacing is actually
> setup, one little innoccuous call to SetVideoFilters(), right?
>
> all the other references to deinterlace where all happening in the
> recording side of things, so naturally i assumed it was all happening
> in there. i don't have a bt based card to have tested that out.
>
> i'm curious when you say 'the only time it deinterlaces anything', did
> you mean for pvr based cards, or is all the other deinterlacing code
> not used anymore?

ok, well i guess i'll go buy a bt based card, plug it in and figure this
out myself if i ever need to know.

cheers,

CraigL->Thx();


_______________________________________________
mythtv-users mailing list
mythtv-users [at] snowman
http://lists.snowman.net/cgi-bin/mailman/listinfo/mythtv-users


ijr at po

Jun 25, 2003, 4:54 PM

Post #7 of 7 (1232 views)
Permalink
Re: Asus Pundit and Interlaced Output, part III :) [In reply to]

On Wednesday 25 June 2003 07:40 pm, Craig Longman wrote:
> Craig Longman wrote:
> > Isaac Richards wrote:
> >> The only time it deinterlaces anything is on playback if the
> >> 'Deinterlace playback' setting is selected.
> >
> > thanks for the input isaac.
> >
> > i think i managed to track down where the deinterlacing is actually
> > setup, one little innoccuous call to SetVideoFilters(), right?
> >
> > all the other references to deinterlace where all happening in the
> > recording side of things, so naturally i assumed it was all happening
> > in there. i don't have a bt based card to have tested that out.
> >
> > i'm curious when you say 'the only time it deinterlaces anything', did
> > you mean for pvr based cards, or is all the other deinterlacing code
> > not used anymore?
>
> ok, well i guess i'll go buy a bt based card, plug it in and figure this
> out myself if i ever need to know.

The code in the recording stuff has never been used. Someone was working on
it and disappeared -- I've been meaning to yank it out sometime.

Isaac
_______________________________________________
mythtv-users mailing list
mythtv-users [at] snowman
http://lists.snowman.net/cgi-bin/mailman/listinfo/mythtv-users

MythTV users RSS feed   Index | Next | Previous | View Threaded
 
 


Interested in having your list archived? Contact Gossamer Threads
 
  Web Applications & Managed Hosting Powered by Gossamer Threads Inc.