Live TV Audio/Video Sync (1 Viewer)

[Owlsroost]

Re: AW: Live TV Audio/Video Sync

@owlroost
could you post patch files for your tsreader versions?

Attached - for SVN trunk revision 27941

Note that this is really a work-in-progress to get some feedback on the changes - I'm still thinking about how to work around the FFDShow problems but keep the slow-motion A/V sync method working for the other decoders.

Tony

 

[tourettes]

Re: AW: Live TV Audio/Video Sync

Note that this is really a work-in-progress to get some feedback on the changes - I'm still thinking about how to work around the FFDShow problems but keep the slow-motion A/V sync method working for the other decoders

It is always possible to write video decoder detection code and then use workaround for only the badly behaving filter(s). Just don't use ffdshow's name to detect since it can change

 

[Owlsroost]

Re: AW: Live TV Audio/Video Sync

Note that this is really a work-in-progress to get some feedback on the changes - I'm still thinking about how to work around the FFDShow problems but keep the slow-motion A/V sync method working for the other decoders

It is always possible to write video decoder detection code and then use workaround for only the badly behaving filter(s). Just don't use ffdshow's name to detect since it can change

That's more-or-less what I'd thought about doing, but how do you know it's the FFDShow video decoder without using at least part of it's name ?

(I guess I could look at the older MP player graph building code, where FFDShow video decoder was detected so that it could be blocked for live TV)

Tony

 

[tourettes]

Re: AW: Live TV Audio/Video Sync

(I guess I could look at the older MP player graph building code, where FFDShow video decoder was detected so that it could be blocked for live TV)

Older code was using full filter name, and that failed at least when ffdshow added something to the filter's name in some cases. Partial match of the name might be good enough, or then GUID (not sure if it is available).

 

[disaster123]

AW: Live TV Audio/Video Sync

I think it is possible to query for CLSID / GUID. The way is: IPin::QueryPinInfo() -> PIN_INFO:Filter -> IBaseFilter::GetClassID()

But why putting anouther workaround into MP? Wouldn't it be possible to fix ffdshow?

 

[Owlsroost]

Re: AW: Live TV Audio/Video Sync

I think it is possible to query for CLSID / GUID. The way is: IPin::QueryPinInfo() -> PIN_INFO:Filter -> IBaseFilter::GetClassID()

I had a quick look at how FFDShow does it last night - it gets the CLSID using *pReceivePin passed into one of the pin connect functions, which looks quite simple, so I'll try that soon.

Wouldn't it be possible to fix ffdshow?

A nice project for you then (the impression I get from looking at the doom9 dev threads is that even setting up a working build environment for FFDShow isn't easy...)

Tony

 

[disaster123]

AW: Re: AW: Live TV Audio/Video Sync

Wouldn't it be possible to fix ffdshow?

A nice project for you then (the impression I get from looking at the doom9 dev threads is that even setting up a working build environment for FFDShow isn't easy...)

Sorry i think i can't ;-) i just wanted to ask if you have thought about that.

 

[Owlsroost]

I thought about it for 10 seconds (or less).....

Got the CLSID GetClassID() etc code working now, so at least TsReader now knows what it's connected to

Tony

 

[disaster123]

AW: Live TV Audio/Video Sync

OK fine ;-) i think this is generally good ;-)

 

[mm1352000]

Hi everyone

I have to say that I don't generally use FFDShow because I find the picture quality of the Cyberlink 10 decoder to be *a lot* better (maybe due to the VA deinterlacing of the video card in comparison to the software deinterlacing done by FFDShow). Anyhow, I've done a quick test *of live TV without the patch* using FFDShow tryouts revision 3556 (September 2 2010), and I don't experience any sync issues whatsoever unless I enable the surface overlay post processing with the DxVA version. I have about 0.5 seconds of slow playback after channel changes, but I experience the same thing with the Cyberlink codec. Are we certain that the problems mentioned in this thread are not stream specific? I can provide a sample for testing if necessary...

OS: XP SP3
Video: ATI HD 4670
Driver: Catalyst 10.4
Streams: h.264 1080i, 576i (Freeview|HD New Zealand DVB-T)
MP: 1.2.0 SVN 27898
Codecs: FFDShow 3556, Cyberlink 10, AC3Filter 1.63b, Monogram 0.9.6.0

mm