MP1 EVR Presenter/dshowhelper community development (58 Viewers)

[robyf]

Re: AW: MP1 EVR Presenter/dshowhelper community development

Tourettes, with the text above you give me the impression audio is going to be sped up/slowed down to get in in pace with the video. The 'worry' I got is if this present a problem for people with digital audio output? My knowledge about dts, etc, is limited, but I thought it contains a digital datastream which is fixed in time and can not be manipulated to go faster/slower. Decoding>Re-encoding is an option ofcourse, but would this go with a quallity loss? If so, how much loss is this?
Can you share your thoughts/knowledge about this (in short)? I could be totally wrong with my perceptions ofcourse, since im a noob in this.

This is true, the only thing you can do with digital audio is decoding and reencoding it in real time. Reclock does it (but it only reencodes it in AC3, no DTS). This is not a good solution for quality, though. I think reclock is good for small adjustments but you have to use the matching refresh rate-fps to have good results. For me speeding up audio from 25->30 fps is no way to go. Better to buy a good video card/tv combination that can support at least 24, 50 and 60hz.

 

[mrmojo666]

small update:


finally i got my first bluray, (24p at 24hz)

and i have stutter with all dlls, the best is again the dll #4.

this makes me think that the problem is also dependent to the stream complexity and the time to compute it.

As aspected with powerdvd10 player that bluray goes perfectly.

 

[gibman]

For months on end I've had problems with micro stuttering on 25 FPS avi/mkv clips on 50hz.

using this "dshowhelper_owlsroost_200410_v004_vsync_always_dll.zip" helped solve the issue for me.

Thanks.

 

[tourettes]

Re: AW: MP1 EVR Presenter/dshowhelper community development

Tourettes, with the text above you give me the impression audio is going to be sped up/slowed down to get in in pace with the video. The 'worry' I got is if this present a problem for people with digital audio output? My knowledge about dts, etc, is limited, but I thought it contains a digital datastream which is fixed in time and can not be manipulated to go faster/slower. Decoding>Re-encoding is an option ofcourse, but would this go with a quallity loss? If so, how much loss is this?
Can you share your thoughts/knowledge about this (in short)? I could be totally wrong with my perceptions ofcourse, since im a noob in this.

This is true, the only thing you can do with digital audio is decoding and reencoding it in real time. Reclock does it (but it only reencodes it in AC3, no DTS). This is not a good solution for quality, though. I think reclock is good for small adjustments but you have to use the matching refresh rate-fps to have good results. For me speeding up audio from 25->30 fps is no way to go. Better to buy a good video card/tv combination that can support at least 24, 50 and 60hz.

In Short:

audio needs to be resampled there is no magic workaround if you want to have perfect playback (no stuttering on 2 hour length movie on today's GPU / audio chip setup)

The Whole Explanation:

There are two different cases that we are talking about with the audio renderer experiments I'm doing. They are both requiring the audio to be decoded and resampled (time stretching, no pitch alteration) and Reclock is as well doing both of those:

1) minor reference clock rate correction - This is required since GPU and audio chip don't share any common crystal for their clock signal sources. Not even modern GPUs that have audio support (don't ask me why... most likely to cut some costs as they buy the audio chips from different supplier?). When you have two different clock sources there will be always drifting in the A/V. And after enough drifting has happened one video frame needs to be dropped or presented twice (you cannot resample video )

2) non 1:1 (1:2, 1:3) matching playback - in short, playing 23.976 material on 25 display or using a PAL speed down to slow down the DVD etc approx 5% to get to the real playback speed (us europeans have faster brains and can process movies in shorter time than US fellows ).


Both of those cases require audio to be decoded and resampled. I see no benefit of trying to re-encode the audio again. It will be enough to decode the audio, resample it and then pass it to the HDMI (8-channel available) as PCM. Encoding will lower the quality (unless using TrueHD or that other lossless format which I forget... we don't want to pay lisencing for the encoding as well) and cause more CPU to be used.

Is it bitstream compatible? No it isn't. It wont be unless GPUs start to have only single clock chip to be shared with audio and video. And in that case the 1st case is not valid anymore and the 2nd should be fixed by a buying a proper tv/beamer.

 

[jet70]

I have stuttering both 25hz (badly) as well as 23.97 hz (to a lesser extent) xvid files.

Can someone confirm - Do I need to upgrade to RC2 to test these dlls or can I try with RC1.

Thank you.

 

[Jay_UK]

Hi there,

These DLL's are RC2 only.

On a separate note, trying to look at the "sweet spot", my SOP appears to be "relatively stable" at 200, but my EOP is either 0 or jumps to 300 - Although its hard to tell as the values are changing so quick. So what value would you use for EOP?

With DLL4 and Reclock (Vsync enabled), picture seems stable, no dropped frames reported, but VS Corr Frames is going through the roof (virtually matches drawn frames) - I have tried using the sweetspot formula (based on 200 and 300 values)

Cheers,

J.

 

[edterbak]

Re: AW: MP1 EVR Presenter/dshowhelper community development

.... PAL speed down to slow down the DVD etc approx 5% to get to the real playback speed (us europeans have faster brains and can process movies in shorter time than US fellows )....

Ghehehehe...

Thank you for the explaining story. It is very clear to me what you mean now. THANKS!

I wonder if the decoding > resampling route will result in an audio quality loss which is noticable anyway. I myself doubt that.

 

[Owlsroost]

OK - it's time for new version - v0010 - to be unleashed on an unsuspecting public - see first post in the thread.

Have fun....

Tony

(and thanks to everyone who's provided feedback on the earlier versions)

 

[Jay_UK]

Tony,

Many thanks for further progressing the great work you're already done

I'll install and test later

Cheers,

J.

 

[Sebastiii]

Hi Tony,

I have to say like Jay_UK
Cheers,
Seb.