LAV Filters - DirectShow Media Splitter and Decoders

[dead_screem]

I was unable to reproduce any issues with MPEG2 with that version, let me know if problems still occur.

Well in asasadad_1's sample and latest test, after a second after opening, the video stops for a split second. Happens with software decoding. Seems to be framerate related? Happens with LAV Splitter plus LAV Video. MPC-HC splitter plus LAV Video, And LAV Splitter plus MPC-HC decoder are fine.

As for the corruption mentioned, It happens on my 9800GTX which is VP2/VDPAU A, which afaik only has a Full H.264/AVC bitstream decoder, MPEG-2 and VC-1 are partial. But with LAV CUVID 0.13 it happily enables for MPEG-2 and VC-1 (not for ASP) and worked perfect with full adaptive deinterlacing... Although the MPEG-2 CPU usage seemed similar to software... don't know about VC-1 since LAV Video never worked with interlaced VC-1.

With the latest LAV Video CUVID, MPEG-2 still enables, but there is blocking on the bottom of the video, and like before has similar CPU usage to software. VC-1 no longer enables.

Any ideas why this is happening? LAV CUVID 0.13 didn't have a built in software fallback... and afaik the CUVID API itself doesn't have one right? does CUVID actually support partial acceleration modes? What is going on?

[nevcairiel]

CUVID has indeed support for partial acceleration. You can decode MPEG2 and VC1 with any device since Feature Set A/VP2, except that it'll do alot of things in software, of course. It does however not have a full software fallback mode. LAV Video adds that and will fallback to software ffmpeg decoding on unsupported content.
VC-1 is completely disabled in LAV Video by default, you'll have to enable it on the Formats tab on the property page, or it'll never work. Its independent from the setting in the hardware accel box.

I cannot reproduce any blocking (but i have a VP4), and the decoder code is really 99% the same from LAV CUVID, so that would be kinda odd.
All i can see in that file is that audio starts 320ms before the video, which means the first frame is frozen for that amount of time, but thats really unavoidable.
It looks like MPC-HCs splitter ignores that offset and starts the frame timestamps with "0", which would probably result in a slight desync eventually.

If you can easily reproduce the blocking problem, please compare to this version of LAV CUVID, it uses the same APIs as LAV Video does now:
http://files.1f0.de/cuvid/LAVCUVID-0.13-LegacyCUDA.zip

If that also shows the problem, i may need to add some code to manually switch to the "newer" APIs when a recent driver is installed.

[dead_screem]

VC-1 is completely disabled in LAV Video by default, you'll have to enable it on the Formats tab on the property page, or it'll never work. Its independent from the setting in the hardware accel box.alright, that fixed that. VC-1 is enabled and seems fine.

I cannot reproduce any blocking (but i have a VP4), and the decoder code is really 99% the same from LAV CUVID, so that would be kinda odd.
All i can see in that file is that audio starts half a second before the video, which means the first frame is frozen for that amount of time, but thats really unavoidable.No. that's not what I ment. That's what is supposed to happen. It gets stuck on the first frame. But with the latest test and that file the video starts playing imediately than stops (to resync) after about 1 second, then continues. enabling CUVID and it correctly stays stopped until the audio catches up. It's just software that is doing this. LAV Splitter plus MPC-HC decoder has correct behavior. LAV Splitter plus LAV Video. is the problem.

If you can easily reproduce the blocking problem, please compare to this version of LAV CUVID, it uses the same APIs as LAV Video does now:
http://files.1f0.de/cuvid/LAVCUVID-0.13-LegacyCUDA.zip
if that is the same build as on the LAV CUVID thread than that works perfect. I have always used Legacy LAV CUVID and it's fine.

here are some shots of asasadad_1's sample. I get similar results with nearly all 1080i content, but is just the minor blocking at the bottom, or appearing similar to 480i DVD content as a black screen with minor blocking up top. I have at least 1 1080i clip that exhibits no problems whatsoever.
http://imageshack.us/g/194/cctvhdsampletssnapshot0.png/

with 480i DVD content the entire screen is black except for some blocking at the top.

[nevcairiel]

I may have an idea why its happening.

Can you try this version?
http://files.1f0.de/lavf/LAVVideo-cuvid-ctxlocking.zip

Just drop it onto the previous cuvid2 version.

[dead_screem]

I may have an idea why its happening.

Can you try this version?
http://files.1f0.de/lavf/LAVVideo-cuvid-ctxlocking.zip

Just drop it onto the previous cuvid2 version.

no change.

update: with cuvid2 or the new build, no corruption when haali splitter is used. and significantly less corruption (but still some) when paired with mpc-hc splitter. But LAV CUVID 0.13 works with LAV Splitter just fine...

[nevcairiel]

This is so odd. Ill have access to a VP2 system tomorrow, maybe i can reproduce it there. The problem is limited to interlaced mpeg2, right?

[nevcairiel]

I may have a new idea, at least it would make sense

Last try for today: http://files.1f0.de/lavf/LAVVideo-mpeg2cuvid.zip

If this doesn't work, i'll have to try tomorrow. :)

[dead_screem]

This is so odd. Ill have access to a VP2 system tomorrow, maybe i can reproduce it there. The problem is limited to interlaced mpeg2, right?

I thought so but just verified it happens with 1080p as well.
This is basically all that is displayed throughout the entire video.
http://imageshack.us/photo/my-images/824/pv20090811gacktthenextd.png/

Also, to quote myself
and significantly less corruption (but still some) when paired with mpc-hc splitter.specifically with mpc-hc splitter, with 1080i, the corruption happens for a few seconds at the start of the file then all is fine...

with that same 1080p clip i posted a pic of above, together with mpc-hc splitter it looks like this now, corruption only at bottom
http://imageshack.us/photo/my-images/689/pv20090811gacktthenextd.png/

[dead_screem]

I may have a new idea, at least it would make sense

Last try for today: http://files.1f0.de/lavf/LAVVideo-mpeg2cuvid.zip

If this doesn't work, i'll have to try tomorrow. :)

Fixed!

[nevcairiel]

Great!

Thanks for your help debugging this. ;)

[dead_screem]

Great!

Thanks for your help debugging this. ;)

glad i could help.

one other thing i just noticed, with lav video when "Send Interlace Flags In Output Media Type" is set it will set dwInterlaceFlags: 0x00000081 even when CUVID is active. LAV CUVID didn't do this, will it screw up anything de-interlace wise?

[nevcairiel]

I don't think it hurts anything. Its always set on progressive content as well, and then all progressive frames flagged properly so that they dont get double-deinterlaced.

[dead_screem]

I don't think it hurts anything. so CUVID and direct show won't interfere with each other as long as the frames are flagged progressive after CUVID is finished decoding?

Its always set on progressive content as well,

and then all progressive frames flagged properly so that they dont get double-deinterlaced.

actually in lav video it's set to 0 when progressive is used. and 81 only when interlaced.

which flags do 0x81 correspond to anyway?

I do know there is a bob or weave mode, which when paired with flags on each sample allows for a mix of Bob TFF, Bob BFF, and weave/progressive. But I have yet to see a DVD decoder do this right for mixed TFF/BFF Interlaced DVDs as well as handling 2:3 pulldown progressive content correctly as progressive when dshow output interlace flags are set.

[nevcairiel]

0x81 is AMINTERLACE_IsInterlaced | AMINTERLACE_DisplayModeBobOrWeave, which are the proper flags to indicate mixed interlaced/progressive content, and instructs DirectShow to pay attention to the flags on the samples. Works perfectly for deinterlacing with EVR.

LAV Video always sets 0x81, as long as the option to report flags is ticked. Its currently not possible to directly find out if a video is pure progressive, or may contain interlaced parts, so i always indicate mixed content, and then flag all frames properly.
LAV CUVID never set any flags, because obviously it didn't need it. ;) I could probably disable the flags completely if the CUVID decoder is in use (and deinterlacing is not set to weave)

See Remarks on this page for details: http://msdn.microsoft.com/en-us/library/windows/desktop/dd407326(v=vs.85).aspx

[dead_screem]

LAV Video always sets 0x81, as long as the option to report flags is ticked. Its currently not possible to directly find out if a video is pure progressive, or may contain interlaced parts, so i always indicate mixed content, and then flag all frames properly.
LAV CUVID never set any flags, because obviously it didn't need it. ;)I coulda swore I checked and it set 0 for progressive. I must've confused it with CoreAVC which does infact set 0 for progressive and something else, probably bob only for interlace...

I could probably disable the flags completely if the CUVID decoder is in use (and deinterlacing is not set to weave)seems logical.

[jmone]

I may have a new idea, at least it would make sense

Last try for today: http://files.1f0.de/lavf/LAVVideo-mpeg2cuvid.zip

If this doesn't work, i'll have to try tomorrow. :)

Thanks - This version fixed the DTV in JTV corruption but the Aspect Ratio is now wrong with everything being presented in 4:3. Checked MPG and TS (MEPG2) files and they play correctly.

[dead_screem]

Bug, when MPEG-2 29.97 2:3 pulldown flagged progressive content is played with LAV Splitter + LAV Video the playback is really choppy, but with no dropped fames in the renderer. switch to MPC-HC splitter + LAV Video and is fine, a smooth 23.97 playback.

Also is DVD support planned ever? It would be nice to use LAV Video as a DVD Decoder...

[06_taro]

use dxva by divx h.264 Decoder?:eek:

Just software decoding.

[nautilus7]

Feature Request for LAV video

Add an option to show frame number and frame type (I/B/P) and maybe file name, like ffdshow.

http://i52.tinypic.com/2q39gtt.png

I need this to be able to compare frames (source vs encode) when backing up my sources.

[nevcairiel]

Bug, when MPEG-2 29.97 2:3 pulldown flagged progressive content is played with LAV Splitter + LAV Video the playback is really choppy, but with no dropped fames in the renderer. switch to MPC-HC splitter + LAV Video and is fine, a smooth 23.97 playback.

Can you get me a sample that shows the problem clearly?

Also is DVD support planned ever? It would be nice to use LAV Video as a DVD Decoder...

Its planned, but its sadly not as trivial as DVD audio decoding..

Feature Request for LAV video

Add an option to show frame number and frame type (I/B/P) and maybe file name, like ffdshow.

http://i52.tinypic.com/2q39gtt.png

I need this to be able to compare frames (source vs encode) when backing up my sources.

I don't really have that much of an interest in an OSD like that.
Feel free to open a feature request ticket on the google code tracker so it doesn't get lost, maybe i'll one day be bored enough.

[azaze1]

nevcairiel,

I'm shooting in the dark here at what may be the culprit I'm seeing, but I figured the audio decoder is the best place to start.

The issue I'm having is Bitstreaming the new Star Wars Episode 1 (phantom menace) Bluray. Specifically on chapter 3 at the very beginning after the opening credits. If I chapter SKIP to this chapter, moments before the ship flys by... and my audio is configured to bitstream... I get a horrifically LOUD sound artifact coming primarily from my rear speakers. If done with the volume at a modest level, it absolutely bombards my speakers with the loudest junk ever. I'm affraid my speakers have perm damage. This does not happen when I decode the audio with the Arcsoft dll and send 6.1 PCM to receiver. It also does not happen when simply playing the movie through normally. Only when bitstreaming and chapter skipping.

Here are some facts:

7.1 setup
Latest LAV Filters 0.35
Bitstreaming the Master Audio 6.1 track
ATI 11.8 drivers
Windows 7 32bit, latest patches etc
Latest nightlies of MPC-HC (updated here and there no difference)
Movie is in an MKV container... I purchased the bluray and ran eac3to to extract the streams in preparations for compression, but decided to start by simply repackaging as MKV.

Other facts:
If I play the Bluray in the PS3 slim and bitstream... no issue.

EVERY SINGLE OTHER MOVIE in my library is fine. I have hundreds, this is the first time I've had this issue happen, and I really hope it's got something to do with the rare 6.1 Master Audio source. Though it's not happening with the episodes 2 or 3. It does happen to other chapters in Episode 1... so it would appear to be JUST this audio track.

I can reproduce it every time. The only thing I've yet to try is loading the Bluray source and playing the movie from PC within m2ts containers. I will try that next, but I'm not sure what could be the culprit other than:

mkvmerge issue ? (can rule that out after m2ts playback test)
ATI Drivers ? (wouldn't be the first time, but why just this track?)
MPC-HC ? (I could try another player... but not sure what else will bitstream)
LAV Splitter (could the splitter have any influence?)
LAV Audio Decoder

I've tried going to back to FFDSHOW for DTS-HD bitstreaming which is what I did for more than a year but I can't seem to get MPC-HC to send any audio whatsoever after installing LAV 0.35. It's like ffdshow is broken for audio decoding after LAV install. I get some other AVI -> DTS filter in the mix rather than FFDSHOW Audio Decoder in the filters list while playing... no audio.

[dead_screem]

Can you get me a sample that shows the problem clearly?sure can.

http://www.megaupload.com/?d=YZZW0VD7
also, no problem with CUVID and this btw, just software...


Its planned, but its sadly not as trivial as DVD audio decoding..
Is it the additional pin requirements? DVD Subpicture in (which gets mixed with the main video by the decoder) and Line 21 output? or is it something else?

[nevcairiel]

The issue I'm having is Bitstreaming the new Star Wars Episode 1 (phantom menace) Bluray.

I believe its a issue with the disc, alot of people keep reporting similar issues.

http://www.avsforum.com/avs-vb/showthread.php?t=1360886

Sadly, the German edition of the collection is muxed differently, and only has a DTS-ES 6.1 track (no HD) for English (only German is DTS-HD, go figure, screw them).
So without the actual track in question i cannot check it out myself.

[STaRGaZeR]

Sadly, the German edition of the collection is muxed differently, and only has a DTS-ES 6.1 track (no HD) for English (only German is DTS-HD, go figure, screw them)

The same edition is sold here, shame on them.

[nevcairiel]

sure can.

http://www.megaupload.com/?d=YZZW0VD7
also, no problem with CUVID and this btw, just software...


Thanks, fixed in LAV Splitter (it was reporting the wrong fps for such mpeg2 files, resulting in LAV Video to use the wrong value to interpolate missing timestamps), and fixed in LAV Video to not trust the splitter values. ;)

[azaze1]

Understood. I went to Target yesterday to purchase the 3 disc set for Episodes 1, 2 and 3 so if it's an issue with the 1st movie disc then it's mass produced. It's not 1 bad disc source that spread like a virus.

Still, the issue is only on HTPC's bitstreaming. I'd also like to try either TMT or PowerDVD to see if bitstreaming in those players produces the same issue. If not.. it's got to be the open source realm.

I believe its a issue with the disc, alot of people keep reporting similar issues.

http://www.avsforum.com/avs-vb/showthread.php?t=1360886

Sadly, the German edition of the collection is muxed differently, and only has a DTS-ES 6.1 track (no HD) for English (only German is DTS-HD, go figure, screw them).
So without the actual track in question i cannot check it out myself.

[Thunderbolt8]

does anyone know VC-1 files which cause problems with haali splitter? seems like they all run comparably fine, seeking works instantly etc. maybe it would be useful to implement some of the ways haali works with VC-1 into LAV splitter.

[nevcairiel]

Whats wrong with VC-1?

Also, as if Haali is such a great tool to learn from, being closed source with an author that doesn't show around these parts anymore. :p

[Thunderbolt8]

VC-1 still seems to be quite slow with seeking on some BBC sources, at least with 0.35. maybe its going to be better in the next release.

[dead_screem]

Here is a sample that I found that has some problems.
http://www.megaupload.com/?d=0M5YGWJI

The DTS-HD MA track when played through AC3Filter (as just the DTS core) the frame rate goes absolutely nuts. with the audio playing way faster than the video, like over 30000fps fast. . oddly enough If CoreAVC is used instead of LAV Video it tries to keep the video in sync with the out of sync audio...
Use MPC-HC splitter and it's fine.

Also it has a DTS Express track that gets detected as MP1.

[Midzuki]

Its planned, but its sadly not as trivial as DVD audio decoding..

Do you mean LAV Source Filter does/will support .AOB demuxing and AUDIO_TS.IFO navigation? :confused: :)

[dead_screem]

Here is another DVD 29.97 progressive 2:3 pulldown file, except it seems to have been authored wrong.
According to mediainfo, both Progressive and 2:3 pulldown are not indicated...

http://www.megaupload.com/?d=JV41VWB8

might want to make sure this plays right (if possible of course)
once again mpc-hc gets 23.976 playback.

[nevcairiel]

Here is a sample that I found that has some problems.
http://www.megaupload.com/?d=0M5YGWJI

The DTS-HD MA track when played through AC3Filter (as just the DTS core) the frame rate goes absolutely nuts. with the audio playing way faster than the video, like over 30000fps fast. . oddly enough If CoreAVC is used instead of LAV Video it tries to keep the video in sync with the out of sync audio...
Use MPC-HC splitter and it's fine.

Also it has a DTS Express track that gets detected as MP1.

AC3Filter just does not support DTS-HD and goes mad with it. Nothing i can fix. Broken decoders will be broken.
Regarding the sync: I didn't implement options to drop frames on desync (yet). Not sure its really worth it. Playback will be broken no matter what.

I can look at the DTS Express detection, i think i know why it stopped working, i didn't pay close enough attention when merging ffmpeg changes...

Here is another DVD 29.97 progressive 2:3 pulldown file, except it seems to have been authored wrong.
According to mediainfo, both Progressive and 2:3 pulldown are not indicated...

http://www.megaupload.com/?d=JV41VWB8

might want to make sure this plays right (if possible of course)
once again mpc-hc gets 23.976 playback.

You played that with the last compile i provided, or with one after i applied the fix?
Anyhow, i'll just look at it when i get home.

Do you mean LAV Source Filter does/will support .AOB demuxing and AUDIO_TS.IFO navigation? :confused: :)

No, with "DVD audio" i just meant "audio on Video DVDs". :p

[disto]

Hi Nev are you planning to integrate deinterlacing option for software decoding not only for CUDA.

[ForceX]

Hi Nev are you planning to integrate deinterlacing option for software decoding not only for CUDA.
Yes, he is. It has been answered many times in the thread.
Deinterlacing is one post-processing task that i will probably allow sooner or later in LAV Video. ffmpeg already has a neat yadif filter.
And yes, LAV could have YADIF, and i do have plans to offer it eventually, but it'll be some time.

[nevcairiel]

Software deinterlacing might happen for 0.37, if everything pans out.
0.36 is basically finished, and will be released soon. Just some minor tweaks and alot of testing to do.

[dead_screem]

AC3Filter just does not support DTS-HD and goes mad with it. Nothing i can fix. Broken decoders will be broken.
Regarding the sync: I didn't implement options to drop frames on desync (yet). Not sure its really worth it. Playback will be broken no matter what.
Well, I know AC3filter doesn't support DTS-HD, that's not the point. Both DTS-HD HRA and DTS-HD MA both contain a DTS core that any decoder that doesn't support HRA or MA should still be able to play back, that's the whole point. when HRA is split out with LAV, AC3filter plays back the DTS core with no problems at all.
it's just MA that this happens with. And like I mentioned, use MPC-HC splitter and AC3filter can playback the DTS core in DTS-HD MA just fine...


You played that with the last compile i provided, or with one after i applied the fix?
Anyhow, i'll just look at it when i get home.
I checked with the latest build, it plays back correct, except it seems to have it's frames marked interlaced. And deinterlacing gets engaged causing a weird 36fps playback (why 36 thou I wonder?) (with de-interlacing off it plays back at 23.976).
Like I mentioned, It appears to be authored wrong, but it is from an official R2 DVD. According to mediainfo, neither Progressive nor Interlaced is indicated and 2:3 pulldown is not either. LAV must be assuming it's interlaced?

[nevcairiel]

Both DTS-HD HRA and DTS-HD MA both contain a DTS core that any decoder that doesn't support HRA or MA should still be able to play back, that's the whole point. when HRA is split out with LAV, AC3filter plays back the DTS core with no problems at all. it's just MA that this happens with.

I know how DTS-HD works. The point is that AC3Filter is just bugged and fails at playing the DTS Core.


And like I mentioned, use MPC-HC splitter and AC3filter can playback the DTS core in DTS-HD MA just fine...
Thats because the MPC-HC splitter removes the HD parts if it sees AC3Filter. I have no intention in supporting broken decoders through hacks.
Especiall AC3Filter. It hasnt been updated for quite a while, and thats not the only bug in it.



I checked with the latest build, it plays back correct, except it seems to have it's frames marked interlaced. And deinterlacing gets engaged causing a weird 36fps playback (why 36 thou I wonder?) (with de-interlacing off it plays back at 23.976).
Like I mentioned, It appears to be authored wrong, but it is from an official R2 DVD. According to mediainfo, neither Progressive nor Interlaced is indicated and 2:3 pulldown is not either. LAV must be assuming it's interlaced?

You can't "assume" interlaced. Either a frame is progressive, or its interlaced. Its a difference in the encoding of the movie, you can't forget to flag it.
What they probably forgot is to flag the pulldown, which is why it treats it as a mixed progressive/interlaced thing. 23.976 with a mixed sequence of maybe 1 progressive, 1 interlaced, etc, would end up as 36 fps. 12 progressive frames per second, and 12 interlaced "frames" (24 fields), being deinterlaced to 24 frames, making 36.

Double-rate/Video mode deinterlacing will usually always result in such weirdness. I could add some heuristic, and if it looks like mixed progressive/interlaced, turn off double rate or something....

[dead_screem]

Thats because the MPC-HC splitter removes the HD parts if it sees AC3Filter. I have no intention in supporting broken decoders through hacks.
Especiall AC3Filter. It hasnt been updated for quite a while, and thats not the only bug in it.
You should have said that in the first place :)

Any idea why AC3filter can playback the DTS core in HRA but not MA?

You can't "assume" interlaced. Either a frame is progressive, or its interlaced. Its a difference in the encoding of the movie, you can't forget to flag it.
What they probably forgot is to flag the pulldown, which is why it treats it as a mixed progressive/interlaced thing. 23.976 with a mixed sequence of maybe 1 progressive, 1 interlaced, etc, would end up as 36 fps. 12 progressive frames per second, and 12 interlaced "frames" (24 fields), being deinterlaced to 24 frames, making 36.

Double-rate/Video mode deinterlacing will usually always return in such weirdness. I could add some heuristic, and if it looks like mixed progressive/interlaced, turn off double rate or something....so the clip actually is mixed progressive/interlaced? would it have played back without deinterlacing enabled even if the pulldown was flagged?
well, the clip is broke either way, no need to add hacks just for this.

[nevcairiel]

No idea, i haven't read its code. It probably stumbles over the HD header off the lossless part in the MA profile or something.

[nevcairiel]

so the clip actually is mixed progressive/interlaced? would it have played back without deinterlacing enabled even if the pulldown was flagged?
well, the clip is broke either way, no need to add hacks just for this.

Its mixed, but its not really interlaced. Its meant to be pulldown, but without the flags it doesn't know that, so it only sees a frame consisting of two fields, and claims its interlaced. If you had the pulldown flags, it would know that its progressive, and re-assemble the frames properly.

Without deinterlacing/ivtc, you would probably see weave artifacts, because the pulldown is not corrected properly.

[dead_screem]

Its mixed, but its not really interlaced. Its meant to be pulldown, but without the flags it doesn't know that, so it only sees a frame consisting of two fields, and claims its interlaced. If you had the pulldown flags, it would know that its progressive, and re-assemble the frames properly.

Without deinterlacing/ivtc, you would probably see weave artifacts, because the pulldown is not corrected properly.

In that case, then If you can't heuristicly detect that pulldown should have been flagged when it wasn't to handle it "correctly", I'd leave it alone.
trying to detect mixed than force off double framerate just seems like a hack.

[nevcairiel]

I could check if its really a 1 progressive, 1 interlaced sequence. In that case, its quite likely that its meant to be pulldown. Or if someone encodes such a movie on purpose, they deserve to get only half-rate deinterlacing. :p

[dead_screem]

I could check if its really a 1 progressive, 1 interlaced sequence. In that case, its quite likely that its meant to be pulldown. Or if someone encodes such a movie on purpose, they deserve to get only half-rate deinterlacing. :p

So your saying you can detect if pulldown should have been flagged and then have the decoder treat it as if it was flagged properly and ivtc the frames properly?
And yeah, your probly right. There is probably no way someone would encode it like that on purpose and not mean for it to be pulldown.

As long as it's not a force-off double framerate hack, that's cool.

[Qaq]

That pulldown is really a pain in ass for me (because of incorrect flags I guess). I wonder how good eac3to removes pulldown or maybe there is a better tool for it?

[glith]

Nev, Any chance of implementing the audio delay option in LAV Audio before 0.36? Thanks! <3

[Kaotech]

Nev, Any chance of implementing the audio delay option in LAV Audio before 0.36? Thanks! <3

Why only audio delay, LFE Crossover and volume by channel it would be nice to :D

[nand chan]

In the options menu of the Audio Decoder, I've noticed a setting titled “Auto A/V Sync correction”.

What exactly does this setting do? Does it tie into LAV Video to ensure both streams start playing at the exact same time?

Does it do any time shifting / clock adjustments like ReClock does? (Most importantly, is it bit perfect?)

Why only audio delay, LFE Crossover and volume by channel it would be nice to :D

Audio delay would be considerably easier to implement than LFE crossover or other audio adjustment logic.

[dead_screem]

Couple feature requests,

In the splitter stream switch menu all the info is lower case. Is there a reason for this? Could it be changed to proper mixed case?
Also level isn't indicated for MPEG-2 only the profile, but it is for H.264 and VC-1.

And track titles in MP4 arn't displayed. This would be nice to have.

[psymed]

I tried to install LAVSplitter but...

RegSvr32

The module "LAVSplitter.ax" was loaded but the call to DllRegisterServer failed with error code 0x80070005.

edit: nvm... run as administrator worked...