ReClock DirectShow filter : try it ! - Page 11

ookzDVD · 8th January 2003, 05:51

@ogo,

your latest filter 0.99b is wroking fine on my system,
I notice faster loading, but sometimes I have a problem which the movie is played faster and then normal again.

ogo · 8th January 2003, 22:21

Version 0.99c is out on my website

It mainly fix problems with sound, add a tray icon, and has some tweaks for AC3

Cheers
OGO

Sumster · 9th January 2003, 02:17

Ogo... excellent work as usual. I have been doing quite a bit of testing and logging with the AC3 output on the latest 0.99c. I have set the MaxLatency to 9 so that I can look at overall trends only, and not microadjustments.

1) First, I have noticed that AC3 repeats (ie. slowing down audio by adding time) sound way worse than AC3 dropouts on my external receiver (dropouts are hardly noticable)

2) I used a 20 minute DIVX and logged the output at various settings. The findings were quite interesting...

a) If I played the DIVX at original speed (23.976 fps), then over time there was an average of 6.43 AC3 REPEATS per 100 seconds of video. This means that the AC3 stream had to slow down 205.76 msec per 100 seconds to stay in sync with the video (0.21%). These repeats were very noticable and there were NO DROPOUTS.

b) If I sped up the video to 24 fps, then the number of repeats decreased to 4.43 AC3 repeats per 100 seconds. Here the AC3 stream only had to slow down by 141.76 msec to stay in sync with the video (since the video has been sped up)... 0.14% difference. Again, these repeats were noticable and there were NO DROPOUTS.

c) If I sped up the video to 25 fps, then there were no repeats, but only DROPOUTS (ie. the AC3 stream was trying to speed up). There were 122.7 DROPOUTS per 100 seconds of video or the AC3 has to speed up by 3925 msec or 3.9%. Ironically, despite the high number of dropouts, they were not as noticable as the low number of repeats!

3) If the framerate of the DIVX were to be calculated to match the unaltered AC3 stream (and create no repeats/dropouts), it would have to be in case a) 24.025 fps, b) 24.034 fps, and c) 24.055 fps.

Ogo, when Reclock isn't used, then I am assuming that Windows tries to match the video to the audio by default, but it does not change the overall framerate of the video, but just adds/drops frames creating jerkiness. Would it be possible to have a setting of 24.05 fps so that there are only a few dropouts and no repeats over time in the AC3 stream? Even better, could the framerate of the video change dynamically to reduce latency rather than changing the audio (leaning towards the faster video speed so as to reduce repeats)?

I am assuming, of course, that changing the global framerate of the video will not introduce significant jerkiness (using Reclock my video has remained very smooth at all fps settings, even if they are not a multiple of my refresh rate).

Thanks again...

Sumster

ogo · 9th January 2003, 11:27

Sumster, you understood very well how it works

It's a pity i don't have an AC3 decoder to listen the results, but it's interesting to know that repeats are not nice to hear whereas drops nearly can't be listened.

Quote:

Ogo, when Reclock isn't used, then I am assuming that Windows tries to match the video to the audio by default

Yes, the default audio renderer slave the reference clock to the audio clock.

Quote:

using Reclock my video has remained very smooth at all fps settings, even if they are not a multiple of my refresh rate).

Interesting. It seems that the default audio renderer use some rough algorithm to slave the clock. I believe this is why the sound interleaving settings in AVI files seems so important. Also, the resolution of the default reference clock is not very good (in mS), whereas mine is under 1uS (it's not based on windows clock).

Quote:

Would it be possible to have a setting of 24.05 fps so that there are only a few dropouts and no repeats over time in the AC3 stream? Even better, could the framerate of the video change dynamically to reduce latency rather than changing the audio (leaning towards the faster video speed so as to reduce repeats)?

What you want indeed, is a good audio-slaved clock. It is already implemented in ReClock when you choose "Disable clocks correction", but only for normal PCM sound. So i will try to implement this for AC3 SPDIF in next version

Sumster · 9th January 2003, 14:51

Ogo,

You are right! I have this particular clip i use... a long panning scene at the beginning of Godfather part III that is absolutely horrible using the default Windows clock. I switched to PCM sound, and "disabled clocks correction" within Reclock --> voila! it still played smoothly!

Therefore, the problem on my system must be the inaccurate Windows audio-slaved clock as you said. Also, most of my older DIVX collection has AC3 muxed using Nandub (ugggh!).... but, somehow, Reclock seems to overcome this terrible muxing problem.

Waiting anxiously to the next release

Sumster

Chibi Jasmin · 9th January 2003, 18:00

@ogo:

Now ReClock is becoming better and better, it's time to start nit-picking:

1. Docs: Since you have the sourcecode in your distribution archive, it is no longer a secret how 'ReClock will correct the system clock with information gathered in real time from your video card.'

2. I still don't think (reasons explained above), your instructions on how to use ReClock with 'DualHead' (more on this in 3) cards make any sense for Matrox cards, so I vote for making a note about that. We should finally investigate, if there are two independent clocks on Matrox cards (might also be interesting for the other cards, I am not sure about the instructions for them either, but don't know too much about other cards) or not...does having 2 RAMDACs automatically mean having two clocks? If yes, they still might be dependent on each other, am I right?
3. The term 'DualHead' is a trademark of Matrox and refers to their very own two-display technology. So 'dual head video card (a card capable to display different images on your monitor and on your TV at the same time like a Radeon for example)' isn't correct. They all have their own names for that (NVidia=TwinView, ATI=?). Maybe you should just remove the term 'dual head' and describe it differently...I vote for that as a proud Matrox owner not wanting to see the term 'DualHead' abused!

Okay, enough of that...I am still happily using ReClock for movie viewing since it came out...thanx for a great one!

kallekill · 9th January 2003, 18:28

What kind of audiocards are you guys using? On my old computer I had a soundblaster live and the playback of movies was really smooth. The only problem I had was that often when I was downloading stuff at the same time as watching a movie, the movie sometimes stuttered for less than half a second.

Now I have build a new computer and I have this problem that the movie sometimes skips a second or pauses, than fast forward a second. It usually happen less than once/movie but when it does it is really irritating. I have tried Win 2000, XP, removed every component in the computer without any luck. The only thing that seems to help is running Win 98. I haven't had the time to see if reclock solves the problem, but I will report if it does.

I have a audigy card in the new computer (I have also tried the sb live but still got skips) and the playback is very smooth otherwise. I haven't found any problems with using the "do resampling with hardware" option in reclock so I guess the audigy can handle this option without any sync problems.

(When watching movies with TV-out the framerate isn't that smooth on both my computers. Using reclock seems to make it much smoother)

Chibi Jasmin · 9th January 2003, 18:37

I am using an SBLive 5.1!

Sumster · 9th January 2003, 18:59

Kallekill....

I'm using an old SB Live with digital out.... I had the exact same problems as yourself with 98% of the video being smooth, but just certain scenes having stuttering or a sudden speed up in playback speed. Long panning scenes are notorious bringing this phenomenon out.

Reclock has fixed all the above mentioned problems --> skipped/repeated video frames, sudden framerate increases, and AC3 muxing problems.

Sumster

P.S. I found that without Reclock, using VMR was much smoother than using the overlay mixer (with the old renderer) in these long panning scenes. With Reclock, however, it does not matter which video renderer I use, or how many extra filters I throw into the video, the playback remains smooth (I am using an 1800XP, so the processor never becomes the bottleneck).

kallekill · 9th January 2003, 19:14

Sumster...

I don't think we have the exact problem, because if I play the scene that just skipped a second time, there is no skip this time, so my problem doesn't seem to be related to specific scenes. It just happens randomly. I was thinking it could be my motherboard causing it because I had a lot of problems with it before and finally found out that my motherboard was incompatible with IBM's latest harddrives. The IDE controller seems a little weird.

A friend has a motherboard with the same brand, alhough he has a compleatly different chipset. He also has the same problem. It is a little weird though that I only get skips in Windows 2000 and and XP and not with Windows 98 so it might be software-related after all. I hope that reclock solves it because i'm really tired of this.

kallekill · 9th January 2003, 23:26

I might add that both the picture and the sound skips. Quite irritating when you miss part of the dialoge because it skips a second. I'm going to watch a movie with Resync now and see if it helps

Sumster · 10th January 2003, 05:25

Ogo,

I just had a theoretical question.... remember how you mentioned that Windows, by default, uses a audio-based clock and that this clock is inaccurate by millisecondss. Reclock however is accurate to the microsecond. Isn't Reclock going to base it's clock on the audio?? Therefore, if the audio clock is inaccurate (or has a lot of variation), won't Reclock by default have this variation as well?

I know that it works (even with clock correction disabled, just using Reclock smoothens out the video).... just curious as to the actual theory behind it.

Thanks...

Airw0lf · 10th January 2003, 11:11

I might have found a bug with the colour indication on the tray icon. I assume it should be changing colour dynamically, right? For example, when I play a PAL DVD, the tray icon starts off red, then turns green as the framerate and refresh rate matching and detection kicks in. This is obviously the desired scenario. On the other hand, when I play an NTSC DVD off my hard disk (using Zoom Player), the icon is initially red (since NTSC framerate detection is not supported). But when I set the framerate manually to 29.97, shouldn't the icon turn green afterwards?

Minor thing, doesn't worry me, but thought you (ogo) would like to know...

telexxingou · 10th January 2003, 12:39

hello....

does anybody have Cmedia 8738 based sound card and working with reclock ?
i've an Asus P4b266-E with this audio chip and i can't get any AC3 spdif sound if reclock is loaded

i've tested many Cmedia drivers and nothing

somebody can help me ?

thx

My config :
P4 1,8Ghz
Asus P4B266-E with Cmedia 8738 chip sound
640Mo
GEFORCE 4 TI4200
...
codecs :
Video : Divx 5.02, Xvid...
Sound : MP3, Moonlight Audiodecod 1.23 and/or Intervideo
Divx Player : Bsplayer, Zoom Player, Media player 6.4,...
nothing work on AC3...with reclock

under Bsplayer, i must select Cmedia Wave Device out else no AC3

under Zoomplayer, reclock selected but no 5.1 (Ac3) out on Spdif, only surround....

kallekill · 10th January 2003, 14:51

I noticed that Reclock doesn't seem to load with the latest version of BS Player. Just so that you know

ogo · 10th January 2003, 15:11

Sumster> Do you know which version of BSPlayer does not work with reclock ? There was a problem on some PC when you quit bsplayer with version 0.99c but it's solved in next version (not out yet)

telexxingou> You're not alone ! I'm working with someone that have the same card than you, and the same problems. For now i don't have the solution (if only i had this card in my hands ...). I have however a question for you : when you activate AC3 using the default wave renderer (which should be working, what do you see in the properties page of the waveout renderer (format tag, etc ...) ?

Airw0lf> You're right, i changed a little bit tray icons colors for next version.

Sumster> The audio clock speed difference (what i call inacuracy in my bad english) has no importance, and has no consequence on reclock. However variation in speed of audio clock would have bad consequence, but i doubt the audio clock has such variations, since it's an hardware clock based on some crystal.

Next version of reclock will be out maybe saturday or sunday if i find time to finish my stuff with AC3

Chibi Jasmin · 10th January 2003, 16:20

Is it possible, you make ReClock not load (and use DefaultDirectSound instead), if I play audio-only files through DirectShow? I wouldn't have to unregister ReClock every time I want to play an mp3/ac3/dts with directshow.

As rendering these files with ReClock (with clock correction) doesn't work here, because it constantly corrects the clock going higher and higher since I finally have some crackling (and disabling clock correction is an additional step just as unregistering)....

So as clock correction is not needed for audio only, why not just make reclock load, if there is also a video to be played?

ogo · 10th January 2003, 16:27

Don't worry next version automatically disable clock correction when playing files with no video

Maybe i should rename the setting "disable clocks correction" to "slave reference clock to audio" don't you think ?

Audio slaved clocks in new version are completely new for both PCM and AC3/SPDIF. Hope it will work for everyone

Chibi Jasmin · 10th January 2003, 17:11

Sounds great...

And the renaming you propose also make sense IMHO...

Your tool is getting better and better...and closer and closer to V1.0

kallekill · 10th January 2003, 17:21

I didn't get Reclock to load in BS player 0.86 in windows 2000. Seems to work fine in Windows 98 though.

9th January 2003, 02:17	#203 \| Link
Sumster Registered User Join Date: Sep 2002 Posts: 53	Excellent Ogo... excellent work as usual. I have been doing quite a bit of testing and logging with the AC3 output on the latest 0.99c. I have set the MaxLatency to 9 so that I can look at overall trends only, and not microadjustments. 1) First, I have noticed that AC3 repeats (ie. slowing down audio by adding time) sound way worse than AC3 dropouts on my external receiver (dropouts are hardly noticable) 2) I used a 20 minute DIVX and logged the output at various settings. The findings were quite interesting... a) If I played the DIVX at original speed (23.976 fps), then over time there was an average of 6.43 AC3 REPEATS per 100 seconds of video. This means that the AC3 stream had to slow down 205.76 msec per 100 seconds to stay in sync with the video (0.21%). These repeats were very noticable and there were NO DROPOUTS. b) If I sped up the video to 24 fps, then the number of repeats decreased to 4.43 AC3 repeats per 100 seconds. Here the AC3 stream only had to slow down by 141.76 msec to stay in sync with the video (since the video has been sped up)... 0.14% difference. Again, these repeats were noticable and there were NO DROPOUTS. c) If I sped up the video to 25 fps, then there were no repeats, but only DROPOUTS (ie. the AC3 stream was trying to speed up). There were 122.7 DROPOUTS per 100 seconds of video or the AC3 has to speed up by 3925 msec or 3.9%. Ironically, despite the high number of dropouts, they were not as noticable as the low number of repeats! 3) If the framerate of the DIVX were to be calculated to match the unaltered AC3 stream (and create no repeats/dropouts), it would have to be in case a) 24.025 fps, b) 24.034 fps, and c) 24.055 fps. Ogo, when Reclock isn't used, then I am assuming that Windows tries to match the video to the audio by default, but it does not change the overall framerate of the video, but just adds/drops frames creating jerkiness. Would it be possible to have a setting of 24.05 fps so that there are only a few dropouts and no repeats over time in the AC3 stream? Even better, could the framerate of the video change dynamically to reduce latency rather than changing the audio (leaning towards the faster video speed so as to reduce repeats)? I am assuming, of course, that changing the global framerate of the video will not introduce significant jerkiness (using Reclock my video has remained very smooth at all fps settings, even if they are not a multiple of my refresh rate). Thanks again... Sumster

9th January 2003, 14:51	#205 \| Link
Sumster Registered User Join Date: Sep 2002 Posts: 53	You're right! Ogo, You are right! I have this particular clip i use... a long panning scene at the beginning of Godfather part III that is absolutely horrible using the default Windows clock. I switched to PCM sound, and "disabled clocks correction" within Reclock --> voila! it still played smoothly! Therefore, the problem on my system must be the inaccurate Windows audio-slaved clock as you said. Also, most of my older DIVX collection has AC3 muxed using Nandub (ugggh!).... but, somehow, Reclock seems to overcome this terrible muxing problem. Waiting anxiously to the next release Sumster

9th January 2003, 18:28	#207 \| Link
kallekill Registered User Join Date: Nov 2001 Posts: 174	What kind of audiocards are you guys using? On my old computer I had a soundblaster live and the playback of movies was really smooth. The only problem I had was that often when I was downloading stuff at the same time as watching a movie, the movie sometimes stuttered for less than half a second. Now I have build a new computer and I have this problem that the movie sometimes skips a second or pauses, than fast forward a second. It usually happen less than once/movie but when it does it is really irritating. I have tried Win 2000, XP, removed every component in the computer without any luck. The only thing that seems to help is running Win 98. I haven't had the time to see if reclock solves the problem, but I will report if it does. I have a audigy card in the new computer (I have also tried the sb live but still got skips) and the playback is very smooth otherwise. I haven't found any problems with using the "do resampling with hardware" option in reclock so I guess the audigy can handle this option without any sync problems. (When watching movies with TV-out the framerate isn't that smooth on both my computers. Using reclock seems to make it much smoother) Last edited by kallekill; 9th January 2003 at 18:30.

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

8th January 2003, 05:51	#201 \| Link
ookzDVD DVD Rebuilder! Join Date: Oct 2001 Posts: 1,147	@ogo, your latest filter 0.99b is wroking fine on my system, I notice faster loading, but sometimes I have a problem which the movie is played faster and then normal again.

8th January 2003, 22:21	#202 \| Link
ogo Registered User Join Date: Oct 2001 Posts: 156	Version 0.99c is out on my website It mainly fix problems with sound, add a tray icon, and has some tweaks for AC3 Cheers OGO

9th January 2003, 18:00	#206 \| Link
Chibi Jasmin Registered User Join Date: Oct 2001 Posts: 800	@ogo: Now ReClock is becoming better and better, it's time to start nit-picking: 1. Docs: Since you have the sourcecode in your distribution archive, it is no longer a secret how 'ReClock will correct the system clock with information gathered in real time from your video card.' 2. I still don't think (reasons explained above), your instructions on how to use ReClock with 'DualHead' (more on this in 3) cards make any sense for Matrox cards, so I vote for making a note about that. We should finally investigate, if there are two independent clocks on Matrox cards (might also be interesting for the other cards, I am not sure about the instructions for them either, but don't know too much about other cards) or not...does having 2 RAMDACs automatically mean having two clocks? If yes, they still might be dependent on each other, am I right? 3. The term 'DualHead' is a trademark of Matrox and refers to their very own two-display technology. So 'dual head video card (a card capable to display different images on your monitor and on your TV at the same time like a Radeon for example)' isn't correct. They all have their own names for that (NVidia=TwinView, ATI=?). Maybe you should just remove the term 'dual head' and describe it differently...I vote for that as a proud Matrox owner not wanting to see the term 'DualHead' abused! Okay, enough of that...I am still happily using ReClock for movie viewing since it came out...thanx for a great one!

9th January 2003, 18:37	#208 \| Link
Chibi Jasmin Registered User Join Date: Oct 2001 Posts: 800	I am using an SBLive 5.1!

9th January 2003, 18:59	#209 \| Link
Sumster Registered User Join Date: Sep 2002 Posts: 53	Kallekill.... I'm using an old SB Live with digital out.... I had the exact same problems as yourself with 98% of the video being smooth, but just certain scenes having stuttering or a sudden speed up in playback speed. Long panning scenes are notorious bringing this phenomenon out. Reclock has fixed all the above mentioned problems --> skipped/repeated video frames, sudden framerate increases, and AC3 muxing problems. Sumster P.S. I found that without Reclock, using VMR was much smoother than using the overlay mixer (with the old renderer) in these long panning scenes. With Reclock, however, it does not matter which video renderer I use, or how many extra filters I throw into the video, the playback remains smooth (I am using an 1800XP, so the processor never becomes the bottleneck).

9th January 2003, 19:14	#210 \| Link
kallekill Registered User Join Date: Nov 2001 Posts: 174	Sumster... I don't think we have the exact problem, because if I play the scene that just skipped a second time, there is no skip this time, so my problem doesn't seem to be related to specific scenes. It just happens randomly. I was thinking it could be my motherboard causing it because I had a lot of problems with it before and finally found out that my motherboard was incompatible with IBM's latest harddrives. The IDE controller seems a little weird. A friend has a motherboard with the same brand, alhough he has a compleatly different chipset. He also has the same problem. It is a little weird though that I only get skips in Windows 2000 and and XP and not with Windows 98 so it might be software-related after all. I hope that reclock solves it because i'm really tired of this.

9th January 2003, 23:26	#211 \| Link
kallekill Registered User Join Date: Nov 2001 Posts: 174	I might add that both the picture and the sound skips. Quite irritating when you miss part of the dialoge because it skips a second. I'm going to watch a movie with Resync now and see if it helps

10th January 2003, 05:25	#212 \| Link
Sumster Registered User Join Date: Sep 2002 Posts: 53	Ogo, I just had a theoretical question.... remember how you mentioned that Windows, by default, uses a audio-based clock and that this clock is inaccurate by millisecondss. Reclock however is accurate to the microsecond. Isn't Reclock going to base it's clock on the audio?? Therefore, if the audio clock is inaccurate (or has a lot of variation), won't Reclock by default have this variation as well? I know that it works (even with clock correction disabled, just using Reclock smoothens out the video).... just curious as to the actual theory behind it. Thanks...

10th January 2003, 11:11	#213 \| Link
Airw0lf Registered User Join Date: Nov 2001 Location: Out There Posts: 89	I might have found a bug with the colour indication on the tray icon. I assume it should be changing colour dynamically, right? For example, when I play a PAL DVD, the tray icon starts off red, then turns green as the framerate and refresh rate matching and detection kicks in. This is obviously the desired scenario. On the other hand, when I play an NTSC DVD off my hard disk (using Zoom Player), the icon is initially red (since NTSC framerate detection is not supported). But when I set the framerate manually to 29.97, shouldn't the icon turn green afterwards? Minor thing, doesn't worry me, but thought you (ogo) would like to know...

10th January 2003, 12:39	#214 \| Link
telexxingou Registered User Join Date: Jan 2003 Posts: 2	hello.... does anybody have Cmedia 8738 based sound card and working with reclock ? i've an Asus P4b266-E with this audio chip and i can't get any AC3 spdif sound if reclock is loaded i've tested many Cmedia drivers and nothing somebody can help me ? thx My config : P4 1,8Ghz Asus P4B266-E with Cmedia 8738 chip sound 640Mo GEFORCE 4 TI4200 ... codecs : Video : Divx 5.02, Xvid... Sound : MP3, Moonlight Audiodecod 1.23 and/or Intervideo Divx Player : Bsplayer, Zoom Player, Media player 6.4,... nothing work on AC3...with reclock under Bsplayer, i must select Cmedia Wave Device out else no AC3 under Zoomplayer, reclock selected but no 5.1 (Ac3) out on Spdif, only surround....

10th January 2003, 14:51	#215 \| Link
kallekill Registered User Join Date: Nov 2001 Posts: 174	I noticed that Reclock doesn't seem to load with the latest version of BS Player. Just so that you know

10th January 2003, 15:11	#216 \| Link
ogo Registered User Join Date: Oct 2001 Posts: 156	Sumster> Do you know which version of BSPlayer does not work with reclock ? There was a problem on some PC when you quit bsplayer with version 0.99c but it's solved in next version (not out yet) telexxingou> You're not alone ! I'm working with someone that have the same card than you, and the same problems. For now i don't have the solution (if only i had this card in my hands ...). I have however a question for you : when you activate AC3 using the default wave renderer (which should be working, what do you see in the properties page of the waveout renderer (format tag, etc ...) ? Airw0lf> You're right, i changed a little bit tray icons colors for next version. Sumster> The audio clock speed difference (what i call inacuracy in my bad english) has no importance, and has no consequence on reclock. However variation in speed of audio clock would have bad consequence, but i doubt the audio clock has such variations, since it's an hardware clock based on some crystal. Next version of reclock will be out maybe saturday or sunday if i find time to finish my stuff with AC3

10th January 2003, 16:20	#217 \| Link
Chibi Jasmin Registered User Join Date: Oct 2001 Posts: 800	Is it possible, you make ReClock not load (and use DefaultDirectSound instead), if I play audio-only files through DirectShow? I wouldn't have to unregister ReClock every time I want to play an mp3/ac3/dts with directshow. As rendering these files with ReClock (with clock correction) doesn't work here, because it constantly corrects the clock going higher and higher since I finally have some crackling (and disabling clock correction is an additional step just as unregistering).... So as clock correction is not needed for audio only, why not just make reclock load, if there is also a video to be played?

10th January 2003, 16:27	#218 \| Link
ogo Registered User Join Date: Oct 2001 Posts: 156	Don't worry next version automatically disable clock correction when playing files with no video Maybe i should rename the setting "disable clocks correction" to "slave reference clock to audio" don't you think ? Audio slaved clocks in new version are completely new for both PCM and AC3/SPDIF. Hope it will work for everyone

10th January 2003, 17:11	#219 \| Link
Chibi Jasmin Registered User Join Date: Oct 2001 Posts: 800	Sounds great... And the renaming you propose also make sense IMHO... Your tool is getting better and better...and closer and closer to V1.0

10th January 2003, 17:21	#220 \| Link
kallekill Registered User Join Date: Nov 2001 Posts: 174	I didn't get Reclock to load in BS player 0.86 in windows 2000. Seems to work fine in Windows 98 though.