DirectVobSub and Overlay mixer =stutter?

[cca]

I ‘m experiencing video stutter every few seconds when I have DirectVobSub(version 2.23) loaded with Overlay Mixer. This is something very frustrating as I have ReClock installed exactly to eliminate such problems but it seems unable to cope with frames dropped because of DirectVobSub. In a quick test it seems that the problem appears only when the Overlay Mixer is used, VMR7 and VMR9 seem unaffected. I hope that there is some workaround for this, I know it is easier to modify the video displayed with the VMR renderers, but VMR7 does not offer brightness and color controls directly, and VMR9’s output is just horrible. When DVSub is not loaded the video is smooth with all renderers. Also, the type of the subtitle (text or subpicture) does not matter. Gabest, I hope you can think of something :(.

[cca]

OK, it seems I was wrong here. After extensive testing I found out that the problem appears only when I play OGMs, while playback is perfectly smooth when I play AVIs. My appologies to Gabest.

[gabest]

No problem :P

So, is it not playing smooth with and also without dvobsub while viewing an ogm?

[cca]

Well, the funny thing is that does play smooth without DirectVobSub loaded. But then why my AVIs (either with MP3 or AC3 sound but always with XviD for video) are all playing smooth with DVobSub? It must be something with the OGG DS filters.

[Blankman]

Originally posted by cca
Well, the funny thing is that does play smooth without DirectVobSub loaded. But then why my AVIs (either with MP3 or AC3 sound but always with XviD for video) are all playing smooth with DVobSub? It must be something with the OGG DS filters.

Not surprising at all. I have encoded the same material using both .AVI and .OGM as the container. When I play the files, the average CPU Utilization for processing .OGM is consirably higher than for .AVI. Almost doubled on an 800 MHz P3.

[cca]

@Blankman:

Maybe you do have CPU problems, but my Athlon is barely at 50% utilization at most movies. There's some bad interaction between DVSub and OGG ds filters. Also, it seems that when .SRT external subtitles are used, the jerky effect is not that big, but it is still there. By the way, I always use external subtitles and not muxed in the OGM or AVI. If use FFDSHOW to display the subtitles everything is fine, but this is not a solution as 99% of my movies have Vobsub subtitles (subpicture) and not SRT.

[cca]

Here's another interesting observation I made: If I set my player (I use ZoomPlayer 3.0) to open at a Realtime priority (I tried High, no go) the stutter is gone! Also, if you look at ReClock's OnScreen Vsunc monitor, one can observe significant less variations than when the player is running with normal priority. I hope Gabest can make out something from this, my knowledge about directshow, multitasking,and all about modern programming is practicaly none.

[Blankman]

I don't think the root cause for studder is CPU. It more likely to be ACPI. With my 800 MHz desktop which suffers from studder, just about every PCI card has been bridged to IRQ 11. With my 850 Mhz laptop which is studder-free, each major device (audio, video, storage, USB,etc.) has its own IRQ.

Getting a faster CPU to support the additional load created by ACPI is only a work-around, not a solution. So I shall attempt to disable both ACPI and "Plug and Pray" on my Desktop and manully configure all the IRQs. (The heck with ACPI, give me back the good old DIP switches and jumper blocks!)

[Blankman]

After an intense reconfiguration, I've managed to reduce CPU utilization for OGM playback with subtitles down to 12-15% on my 800 MHz P3 desktop. Finally the stutter problem has been squashed!

[cca]

Interesting. What did you do? Just removed ACPI? Please give some details, maybe you can help more people like you and me.

[Blankman]

I never had problems with stutter until I added four 200 GB hard drives to my system. So I suspected it was a hardware or memory issue. (In case you didn't know, to use a hard drive larger than 137 GB in Windows, you need to replace the default 28 bit disk driver with a 48 bit driver.)

I attacked the problem from hardware by disabling ACPI on my motherboard's (Abit BH6) soft config menu, then manullay assigning unique IRQs to the video, audio, network, and mass storage cards. This worked as I expected.

In addition to this I found an unexpected alternative solution. This alternative avoids disabling ACPI and manually setting IRQs.

In Windows Media Player 6.4, I disabled hardware acceleration. Now I would think this would make the stutter worst, and increase the CPU load, but not so. Instead CPU load dropped from 48% to 11%, and there was absolutely no stutter. Other players (e.g. BSPlayer and Media Player Classic) do not have this setting, so there was no way for me to correct the problem for other players.