ffdshow tryouts project: Discussion & Development

[tal.aloni]

I noticed that deinterlacing is broken in ffdshow r2253 and r2265 :(

When I enable deinterlace then the deinterlacing filter seem to switch on and off randomly!
Happens at least with the KernelBob and the KernelDeint. You can clearly see it with the "Map" option checked.
Looks like sometimes the frames simply by-pass the deinterlace filter.

The older build r2202 works good. This is with live MPEG-2 broadcast in DVBViewer...

Dear lord,
latest ffdshow revisions can deinterlace only interlaced content.
it's possible that some of the frames in the broadcast are not flagged as interlaced. in that case, you can check "process frames flagged as progressive" in the deinterlacing filter tab.

Tal

[albain]

Actually, TrueHD 7.1 is broken in 32bit builds also. With all 7.1 (regardless samplerate) samples i get no sound, while 5.1 play fine.

I found some differences with the version I have from the branch, that should explain the broken 7.1 on TrueHD

[nautilus7]

So we have a saboteur? :D

[STaRGaZeR]

The problem is that is seems to happen randomly and with live broadcast. I will try to capture a clip that exhibits the problem...

EDIT: I got a video capture that can be used to reproduce the problem with MPC + ffdshow r2265. Uploading it now.

Like tal.aloni says now ffdshow only deinterlaces those frames flagged as interlaced. If your stream has random frames flagged as progressive (I have DVDs like that, but reversed: film with random interlaced frames) they will not get deinterlaced. Well they will get the flag, but the "change" from interlace to progressive and vice versa seems to be so fast that there is no time for the deinterlacer filter to properly do its job. You can try the "process frames flagged as progressive" option.

[haruhiko_yamagata]

With the recent changes it seems that all MBAFF samples I have play with the correct field order, great work http://www.xtremesystems.org/forums/images/smilies/thumbsup.gif

Regarding the sample you posted here (http://forum.doom9.org/showpost.php?p=1206065&postcount=4715), it seems ffdshow flags different parts of the video as progressive and others as interlaced. CoreAVC and Cyberlink both flag the whole video as interlaced.Thanks for testing. Yes, I know. I'll look into it.

[TheShadowRunner]

Hey all, I was just wondering if multithreading for SVQ3 was planned at some point?
Those Nasa HD videos are encoded in SVQ3 and I wonder what kind of crazy processor one needs to watch them with ffdshow:
http://anon.nasa-global.edgesuite.net/qt.nasa-global/ksc/ksc_061108_glast_launch_1080i.mov
See you,

TSR

[haruhiko_yamagata]

I use ffdshow-mT (2250) (it's the only version where I don't have any dropped image with mpc homecinema and mt2s file (blu-ray)) but this version doesn't support HD audio codec (E-AC3,MLP,DTS-HD) because it's not the same build.
Do you expect to make a version with ffdshow-mt and HD audio codec support?Sorry, but it's not easy.
ffmpeg-mt is based on old (around 200805) libavcodec and it is hard to update to the latest.
Even if we can, it will make maintenance difficult.
Please wait until ffmpeg-mt is updated by Alexander Strange.

[Octo-puss]

Try to set DefaultDirectSoundDevice on Options>>Playback>>Output>>Directshow audio

Oh! It works! I used to have it set to Realtek HD Audio just like I do everywhere. Weird, but works. Thank you!

Also when we are the filters and stuff, shall I disable all internal ones in MPC/MPC-HC? I mean, I assume FFD is better?

[STaRGaZeR]

Hey all, I was just wondering if multithreading for SVQ3 was planned at some point?
Those Nasa HD videos are encoded in SVQ3 and I wonder what kind of crazy processor one needs to watch them with ffdshow:
http://anon.nasa-global.edgesuite.net/qt.nasa-global/ksc/ksc_061108_glast_launch_1080i.mov
See you,

TSR

I can perfectly play it. E8400@4GHz, deinterlaced through hardware. CPU usage stays between 40-50%.

[TheShadowRunner]

Thanks for testing STaR, can you confirm it only uses 1 core? My dual core @ 2.6ghz just can't cope with it :/
Time to grab a new CPU I guess ^^;

[LoRd_MuldeR]

Okay, here is the sample:
http://www.mediafire.com/file/jznjmmzmz3i/Sample.7z

Please watch it with "KernelBob" deinterlacer enabled and with the "Map" option checked.
Starting at 00:59 ffdshow r2265 will occasionally stop deinterlacing!


[EDIT]

Doah! :mad: :scared: :o

Sorry, I just noticed tal.aloni's and STaRGaZeR's posts. How could I be so stupid and miss the new option?!
With that option checked everything is fine again, of course ...

[fastplayer]

Hey all, I was just wondering if multithreading for SVQ3 was planned at some point?
1. Wrong forum to ask. Ask here instead:
https://lists.mplayerhq.hu/mailman/listinfo/ffmpeg-user/
2. The video plays fine without one single frame drop on my single-core Athlon64 2.2GHz.

[BatKnight]

Hi,

I am using ffdshow SVN 2265 (x86) over Windows Vista with SPDIF enabled in DTS and AC3. I am using an external decoder that can only process up to DTS ES and DD EX.
My question is: Why can I hear sound via SPDIF when playing a TrueHD 7.1 96Khz m2ts file?

Does this mean that ffdshow strips the AC3 part of the TrueHD (if there is one in the first place) and SPDIFs only the AC3? Is it SPDIFing the full TrueHD and then the external decoder only plays the AC3 part?

This also happens on DTS Master Audio files.

Can someon please explain the details of the process? :helpful:

Thanks
Bat

[TheShadowRunner]

Ok, actually nope, contrary to what i just wrote, it's not a matter of hw deinterlacing being enabled or not, it's a matter of splitter!?
With Gabest MP4 Splitter, it plays just fine.
With Haali Media Splitter, it lags like crazy!
Odd..

[STaRGaZeR]

Thanks for testing STaR, can you confirm it only uses 1 core? My dual core @ 2.6ghz just can't cope with it :/
Time to grab a new CPU I guess ^^;

Yes, the decoder only uses 1 core. And I can confirm Haali's needs a lot more CPU power. Also with Haali's the field order is wrong ...

@Lord

I've been watching your sample and I've found 0 problems with it, every single frame is flagged as interlaced and it gets deinterlaced properly. Tested with kernelbob, hardware deinterlacing and various renderers. BTW the Map option produces white garbage on screen? I can't see anything with it enabled :p

[LoRd_MuldeR]

@Lord

I've been watching your sample and I've found 0 problems with it, every single frame is flagged as interlaced and it gets deinterlaced properly. Tested with kernelbob, hardware deinterlacing and various renderers. BTW the Map option produces white garbage on screen? I can't see anything with it enabled :p

The "Map" shows you where KernelBob detects motion. The "white" pixels indicate the pixels that will be interpolated by KernelBob.
Of course you wouldn't activate the "Map" option for playback. You simply use it to adjust the "Threshold" parameter and then turn it off.

With "Map" on you can clearly notice that KernelBob suddenly stops to interpolate. Hence the deinterlacer is toggled off for certain frames.
It starts at 0:59 and happens occasionally. The ranges where it happens are identical for each run. It's 100% reproducible for me!

However if I check the "process frames flagged as progressive" option all works fine again :)

[STaRGaZeR]

Oh I see now, thanks. However I can't notice any interpolate stop in the entire video.

And I was wrong! Open the sample with DGIndex. It will report progressive frames all over the place! :confused:...

[LoRd_MuldeR]

And I was wrong! Open the sample with DGIndex. It will report progressive frames all over the place! :confused:...

I can confirm this, so it's not ffdshow's fault for sure:

http://img183.imageshack.us/img183/3671/progressivehd1.th.png (http://img183.imageshack.us/my.php?image=progressivehd1.png)http://img183.imageshack.us/images/thpix.gif (http://g.imageshack.us/thpix.php)


I don't know why that is. That's how the stream came from the broadcast station.
Maybe their MPEG-2 encoder is borked or they don't know how to configure it properly.

But as long as I know how to fix it, there is no problem for me :)

[ikarad]

Sorry, but it's not easy.
ffmpeg-mt is based on old (around 200805) libavcodec and it is hard to update to the latest.
Even if we can, it will make maintenance difficult.
Please wait until ffmpeg-mt is updated by Alexander Strange.

thanks.

Do you know if alexander strange works on a new version of multithreading libavcodec?

With ffdshow-mt it lacks only the support of HD-audio codec and the support of .PG subtitles (HD subtitles) and with these supports ffdshow-mt would be perfect for myself.

[albain]

Hi,

I am using ffdshow SVN 2265 (x86) over Windows Vista with SPDIF enabled in DTS and AC3. I am using an external decoder that can only process up to DTS ES and DD EX.
My question is: Why can I hear sound via SPDIF when playing a TrueHD 7.1 96Khz m2ts file?

Does this mean that ffdshow strips the AC3 part of the TrueHD (if there is one in the first place) and SPDIFs only the AC3? Is it SPDIFing the full TrueHD and then the external decoder only plays the AC3 part?

This also happens on DTS Master Audio files.

Can someon please explain the details of the process? :helpful:

Thanks
Bat

That's right, I made an internal parser that strips off AC3 frames from TrueHD streams (if there are), but only if AC3 codec is set to SPDIF. In that case the truehd part is not decoded at all

Concerning DTS-HD, it is slightly different : ffmpeg does not handle DTS-HD decoding, so for now FFDShow parser strips off the DTS core and sends it to the DTS decoder (or SPDIF if selected)

[tal.aloni]

Albain,
did you figure why we have 2 places to select spdif-passthrough instead of one?

Tal

[STaRGaZeR]

I can confirm this, so it's not ffdshow's fault for sure:

http://img183.imageshack.us/img183/3671/progressivehd1.th.png (http://img183.imageshack.us/my.php?image=progressivehd1.png)http://img183.imageshack.us/images/thpix.gif (http://g.imageshack.us/thpix.php)


I don't know why that is. That's how the stream came from the broadcast station.
Maybe their MPEG-2 encoder is borked or they don't know how to configure it properly.

But as long as I know how to fix it, there is no problem for me :)

DVB streams are almost always bad flagged, dunno why. But again, I don't have any problems with ffdshow and that sample, it deinterlaces all frames using xvidvideo.ru's rev2265 x86. So who's right?

[albain]

Albain,
did you figure why we have 2 places to select spdif-passthrough instead of one?

Tal

Well, if I understand you correctly, there are 2 places (and maybe more tomorrow with "HDMI passtrough") because if your receiver knows how to handle Dolby Digital decoding and not DTS, it is better to have the choice to select SPDIF for AC3 and libdts or libavcodec for DTS.

[tal.aloni]

Well, if I understand you correctly, there are 2 places (and maybe more tomorrow with "HDMI passtrough") because if your receiver knows how to handle Dolby Digital decoding and not DTS, it is better to have the choice to select SPDIF for AC3 and libdts or libavcodec for DTS.

Edit:
I was referring to both the checkbox and the drop down each format have.
why do I need to "Use SPDIF when AC3 output set" if I can just select "SPDIF" for this audio format?
any good reason to keep it?

Edit:
I checked the code, the checkbox does exactly what it says.

1. it seems that albain didn't implement that "functionality" for libavcodec.
2. this "functionality" is totally unnecessary in my opinion (and confusing).

TffdshowDecAudio.cpp:

// use SPDIF pass through when AC3 output and 'Use SPDIF when AC3 output set' are checked.
if ( codecId == CODEC_ID_LIBA52
&& inpin && inpin->audio
&& (inpin->audio->codecId == CODEC_ID_LIBA52 || inpin->audio->codecId == CODEC_ID_SPDIF_AC3)
&& globalSettings->ac3SPDIF
&& presetSettings && presetSettings->output
&& (presetSettings->output->outsfs & TsampleFormat::SF_AC3)
)
inpin->audio->codecId = codecId = CODEC_ID_SPDIF_AC3;
if ( codecId == CODEC_ID_LIBDTS
&& inpin && inpin->audio
&& (inpin->audio->codecId == CODEC_ID_LIBDTS || inpin->audio->codecId == CODEC_ID_SPDIF_DTS)
&& globalSettings->ac3SPDIF
&& presetSettings && presetSettings->output
&& (presetSettings->output->outsfs & TsampleFormat::SF_AC3)
)
inpin->audio->codecId = codecId = CODEC_ID_SPDIF_DTS;

[albain]

I implemented this checkbox for the following purpose : if you output sound to 2 rooms, one over SPDIF (living room) and one over analog output (bedroom for ex), if you set AC3 to SPDIF, you will never get sound to your bedroom with AC3 streams.

So the solution is
1/ To define 2 audio presets :
1a/one with AC3 output checked for the living room preset (and resampling to 5.1 48khz or 96khz),
1b/one with AC3 output unchecked for the bedroom preset (and resampling to stereo 44khz).
2/ Set AC3 codec to libavcodec or liba52

As a result, SPDIF will be used for living room preset and not for bedroom preset

This is not ideal I agree : SPDIF should not be a global setting but a preset setting, one for DTS, one for AC3, and maybe others for HD audio formats

However, I indeed forgot to update this code for the new audio codecs with libavcodec

[haruhiko_yamagata]

Yes, that's what the spec says. And that's how it's supposed to be. If all DVDs were authored correctly, we could rely on "progressive_flag". No DVD player would then need an expensive video processing chip for deinterlacing because it could simply read out the flags and everything would be fine. The problem is that not all DVDs are authored correctly. You can not fully trust these flags because some DVD encodes are plain broken. That's why every good standalone DVD player converts all DVDs to 480i60 first and only afterwards uses expensive video processing chips (e.g. HQV Reon) to perform IVTC.

Having a deinterlacing mode which strictly makes use of the MPEG2 bitstream flags is not a bad idea. It will probably work well with many DVDs - but not with all. So it must still be possible to output any MPEG2 stream (except those with progressive_sequence set to 1) to 50i/60i.
I've just updated H.264 interlacing detection and now, I think those flags are used to define decoding processes.
Forgive me randomly guessing, in my opinion, after decoding processes such as deinterlacing and color space conversion are not defined by those flags.

PAFF is of course picture adaptive, some pictures are encoded using fields, some pictures frames. And the frame pictures are not necessarily still pictures, they have severe combing.
If it is MPEG-2 such pictures are judged progressive. And H.264 decoded by ffdshow before rev 2268 as well.

Fortunately H.264 has supplemental information for after decoding processes. In MPEG-2 we have no choice but to derive after decoding processes information from decoding information.

I mean interlacing/progressive mode of after decoding processes is not always same as decoding processes. I guess that's one of the reasons why flags are not reliable in MPEG-2.

Of course MPEG-2 is not H.264. Just a random guess.

[haruhiko_yamagata]

Okay, here is the sample:
http://www.mediafire.com/file/jznjmmzmz3i/Sample.7z

Please watch it with "KernelBob" deinterlacer enabled and with the "Map" option checked.
Starting at 00:59 ffdshow r2265 will occasionally stop deinterlacing!


[EDIT]

Doah! :mad: :scared: :o

Sorry, I just noticed tal.aloni's and STaRGaZeR's posts. How could I be so stupid and miss the new option?!
With that option checked everything is fine again, of course ...
I think "Process frame flagged as progressive" should be checked by default. Any opinions?

[tal.aloni]

This is not ideal I agree : SPDIF should not be a global setting but a preset setting, one for DTS, one for AC3, and maybe others for HD audio formats


I understand.
there are two solutions I can think of, both are better and easier to understand:
1. we can make the entire codec selection based on presets.
2. we can add another two checkboxes to the "output" tab, so it would look like that:
[x] AC3
__ [x] Encode mode
_____ [x] encode only 5.1ch streams ...
_____ Bit Rate: [640]
[x] DTS (passthrough)

I think 2 will be much easier to implement, what do you say?

[cc979]

i just tried to compile svn-2266 and i get this error

windres ffdshow.rc -o ffdshow.res.o -D_WIN32_IE=0x0500 -DIDCLOSE=8
dllwrap -mno-cygwin --target=i686-pc-mingw32 \
-o ../bin/ffdshow.ax -def ffdshow.def.gcc \
-Wl,--strip-all,--enable-stdcall-fixup \
ffdshow_all.o TffDecoder_reg.o Tconfig.o ffdebug.o cpuid.o memcpy.o subtitles/fontRenderer.o convert/colorspace_rgb_mmx.o acm/Tacm.o audioFilters/ffdshow_audiofilters_all.o codecs/ffdshow_codecs_all.o codecs/TaudioCodecTremor.o codecs/TvideoCodecTheora.o codecs/TaudioCodecLibavcodec.o codecs/TaudioCodec.o codecs/TvideoCodecLibavcodec.o convert/ffdshow_convert_all.o dialog/ffdshow_dialog_all.o dialog/TSpecifyPropertyPagesVE.o filters/ffdshow_filters_all.o imgFilters/ffdshow_imgfilters_all.o muxers/ffdshow_muxers_all.o xiph/ogg/bitwise.o xiph/ogg/framing.o settings/ffdshow_settings_all.o settings/lzo/minilzo.o subtitles/ffdshow_subtitles_all.o baseclasses/baseclasses_all.o ffmpeg/libavcodec/golomb.o audioFilters/TaudioFilterConvert.o audioFilters/dotproduct.o imgFilters/simple_idct_mmx.o imgFilters/Skal/skl_dct_sse.o imgFilters/Skal/skl_fdct_mmx.o ffdshow.res.o \
-lmsvcrt -lstdc++ -lsupc++ \
-lgdi32 -lcomdlg32 -lwinmm -lcomctl32 -luuid -lole32 -loleaut32 \
-ldinput -ldxguid -lshlwapi
ffdshow_all.o:ffdshow_all.cpp:(.text+0x442dd): undefined reference to `TaudioParser::TaudioParser(IffdshowBaseA*, IdecAudioSink*)'
d:\mingw\bin\dllwrap.exe: d:\mingw\bin\gcc exited with status 1
make: *** [../bin/ffdshow.ax] Error 1


anyone else got problems ?

cheers

[tal.aloni]

I think "Process frame flagged as progressive" should be checked by default. Any opinions?

I thought about it thoroughly before I submitted the patch,
I know the upgrade path is imperfect, but I'm convinced I made the right call:
only DVB viewers would want that checked, and those are a small group of savvy users that would notice that something is "different" after the upgrade. LoRd_MuldeR just proved that the advanced users will know how to handle.

I know this is far from perfect, but it allows us to to good with a much larger crowd:
for a short period of time, I assumed that "deinterlacing" is only applied to "interlaced content", like done in many other decoders (like coreavc). I soon found out that something is wrong, but perhaps many others, who casually watch interlaced content, haven't had such luck.

in addition, let's not forget that this function is only useful if you happen to have a badly authored stream. I don't want to "suggest" using that function to the average user who wish to deinterlace when coming across interlaced stream.

Tal

[clsid]

I understand.
there are two solutions I can think of, both are better and easier to understand:
1. we can make the entire codec selection based on presets.
2. we can add another two checkboxes to the "output" tab, so it would look like that:
[x] AC3
__ [x] Encode mode
_____ [x] encode only 5.1ch streams ...
_____ Bit Rate: [640]
[x] DTS (passthrough)

I think 2 will be much easier to implement, what do you say?
I would suggest these options. Makes it a bit more clear that encoding applies to other audio formats.
==== SPDIF ====
[ ] Pass through AC3
[ ] Pass through DTS
[ ] Encode other formats to AC3
[x] 5.1ch only
Bitrate: [640]

[tal.aloni]

I would suggest these options. Makes it a bit more clear that encoding applies to other audio formats.
==== SPDIF ====
[ ] Pass through AC3
[ ] Pass through DTS
[ ] Encode other formats to AC3
[x] 5.1ch only
Bitrate: [640]

even better, I assume "Pass through AC3" has the priority.

p.s. it should be "Encode to AC3", since some would want to apply filters for AC3 stream and re-encode.

[clsid]

Yes, unchecking AC3 passthrough should allow encoding the decoded/processed audio to AC3 again.

[STaRGaZeR]

I think "Process frame flagged as progressive" should be checked by default. Any opinions?

Yes and no. There is a problem with streams with mixed content, and specially those with only a few frames in between the dominant flag like in my last sample posted here (the Gundam one): the change from one to another produces 1 or 2 frames in where interlaced video is showed as progressive and progressive is still deinterlaced. I don't know if that is a fault of the deinterlacer, that has some latency, or because the flags themselves have some latency (this seems unlikely). For the average user that might be an issue, so here I vote for yes. On the other hand, deinterlace a pure progressive stream can cause artifacts, with also might be an issue. Here I vote for no. For those special, bad flagged streams the option is there, no need to deinterlace all videos played by ffdshow, another no here.

If those first frames after interlaced->progressive or progressive->interlaced changes can be fixed by modifying the code or something, another no.

[tal.aloni]

In that case, we don't even need the "SPDIF" option in the AC3/DTS combo box, right?

[clsid]

In that case, we don't even need the "SPDIF" option in the AC3/DTS combo box, right?
If SPDIF passthrough is enabled in the Output section, then the decoder setting in the Codecs section is effectively ignored. No need to have a spdif option in the dropdown there anymore.

[SeeMoreDigital]

I would suggest these options. Makes it a bit more clear that encoding applies to other audio formats.
==== SPDIF ====
[ ] Pass through AC3
[ ] Pass through DTS
[ ] Encode other formats to AC3
[x] 5.1ch only
Bitrate: [640]Yep... I like that :)

[clsid]

How about having these two options for the deinterlacing issue:

[ ] Deinterlace frames flagged as progressive (in mixed streams)
[ ] Deinterlace frames flagged as progressive (always)

A stream classifies as 'mixed' if an interlaced frame has been encountered at least once.

If no streams exist that have interlaced parts, but have all frames flagged as progressive, then option #2 is not needed.

[tal.aloni]

If SPDIF passthrough is enabled in the Output section, then the decoder setting in the Codecs section is effectively ignored. No need to have a spdif option in the dropdown there anymore.

Great, the result will be much more friendly.
I have all that I need to implement this. Albain, I would still be glad to hear your opinion.

Tal

[tal.aloni]

How about having these two options for the deinterlacing issue:

[ ] Deinterlace frames flagged as progressive (in mixed streams)
[ ] Deinterlace frames flagged as progressive (always)


1. There could be interlaced streams which have no single interlaced frame, like old HuffYUV streams, which are lossless, but don't contain interlaced information.

2. This should be easy to implement, but I don't like it. end users do not get incorrectly flagged mixed streams, except for a handful of badly authored DVDs, and the DVB crowd.

Tal

[LoRd_MuldeR]

end users do not get incorrectly flagged mixed streams, except for a handful of badly authored DVDs, and the DVB crowd.

...which includes a great part of the ffdshow users. Analog TV is dieing, DVB will become more and more important ;)

[STaRGaZeR]

How about having these two options for the deinterlacing issue:

[ ] Deinterlace frames flagged as progressive (in mixed streams)
[ ] Deinterlace frames flagged as progressive (always)

A stream classifies as 'mixed' if an interlaced frame has been encountered at least once.

If no streams exist that have interlaced parts, but have all frames flagged as progressive, then option #2 is not needed.

Bad workaround for mixed, imagine 10000 frames of progressive content. Only frame 50 is interlaced. Are you going to deinterlace from 50 to 10000?

With the current implementation in that same example you'll get progressive output until frame 49. Then you reach a interlaced frame, 50, but the deinterlacer is slow and can't properly do its job in time. Frame 50 is displayed with combing. Then frame 51 comes as progressive again, and it's showed as progressive. Result: no visible deinterlacing for frame 50, and a little jump. Now imagine a stream where the interlaced frames are 50, 100, 150..., or in groups of 3 frames 50-53, 100-103, 150-153... the rest is progressive (I have lots DVDs like that, even with 50-50% of progressive and interlaced in the main movie). It's a very bad situation, with combing and jumpy video caused every time a change between progressive and interlacing is reached. You can clearly see the effect with this sample (http://www.megaupload.com/?d=MPMULDNV). Check at scene changes, and if possible use MPC-HC's display stats to see the framerate changes and the jumps. The question is: is this fixable?

About the bad flagged streams the only proper solution is to force it always, I agree. ffdshow can't know if a certain stream is bad or not. Is in the hands of the user.

EDIT: it seems the jump only happens when hardware deinterlacing is used. With normal DI there is only combing.

[tal.aloni]

http://iknowu.net/files/public/ffdshow/Output.png

How about that?

[Eragon4ever]

Not good, the encode option has to be in the passthrough border, too. You could place the sample formats next to instead of on top of each other to gain some more room and then move the encode stuff in the correct border. Is LPCM an other format that can be send over SPDIF? If so it should be up there as well.

BTW, what do you plan to display in codecs tab if e.g. AC3 passthrough is selected? You can't display "libavcodec" or "liba52" because that would be wrong, same goes for "disabled".

[tal.aloni]

Not good, the encode option has to be in the passthrough border, too. You could place the sample formats next to instead of on top of each other to gain some more room and then move the encode stuff in the correct border. Is LPCM an other format that can be send over SPDIF? If so it should be up there as well.

Well, it seems that I have to revert to my original plan anyway, since I don't want to add two different output formats ("AC3 Passthrough" and "AC3 Encode") to the sample format enum / registry.


BTW, what do you plan to display in codecs tab if e.g. AC3 passthrough is selected? You can't display "libavcodec" or "liba52" because that would be wrong, same goes for "disabled".
"spdif".

[Eragon4ever]

"spdif"
Ah, ok. I thought you were going to remove that.

[tal.aloni]

OK, after I examined the code, I'm pretty sure that is the right way of doing it:
(the fact that I suggested it that way is just a coincidence, really!)

http://iknowu.net/files/public/ffdshow/Output2.png

[Snowknight26]

Too much blank space on the left.

[tal.aloni]

Too much blank space on the left.
True, I can relocate the LPCM to the left.

[Eragon4ever]

Can't get it to compile right now so here is a screenshot out of VS:

http://s2d1.turboimagehost.com/t/890867_snap0019.png (http://www.turboimagehost.com/p/890867/snap0019.png.html)