Book Descriptions: Alors Ca Mord is good choice for you that looking for nice reading experience. We hope you glad to visit our website. Please read our description and our privacy and policy page. How it works: • 1. Register a free 1 month Trial Account. Download as many books as you like (Personal use) • 3.

Cancel the membership at any time if not satisfied. Note: We cannot guarantee that every book is in the library. But if You are still not sure with the service, you can choose FREE Trial service. You can cancel anytime. You can also find customer support email or phone in the next page and ask more details about availability of this book. Click 'CONTINUE' button to proceed.

With playing (DXVA with hardware deinterlacing, not sure about software decoding), it takes two frame steps to actually change frames. Also, as soon as you take exit full screen by way of Alt+Tab, for example, the status bar text changes to 'Shader:' even with shaders off followed by Playing a couple of seconds later. Eh, more things.

When in full screen mode, if you show/hide some of the view options (Ctrl+5 or Ctrl+6 come to mind), the controls, seek bar, and everything else you have enabled stays visible until you move your mouse. Assuming you have the auto-hide setting enabled. I don't think that DirectVobSub is used for DVD subtitles.

Eduardo Peixoto Download HTML Embed. 225: Believing is no longer believing in another world, in a transformed world. It is only believing in the. He took the risk of hiring an academic mestiza, sparked my history of technology flame and last but not least believed in and supported me against all odds. The of and to a in for is on s that by this with i you it not or be are from at as your all have new more an was we will home can us about if page my has no search free but our one other do information time they site he up may what which their news out use any there see c so only his e when contact here business who web also.

I think that is handled internally. Nevertheless, this delay issue should be fixed. With what video renderer(s) does it happen? Does it happen with ALL your DVDs? Is the delay always the same value? With MPC-HC it is impossible to load separate subtitles when I see DVD subtitles. MPC-HC only allow to load subtitles included in the DVD but not subtitles which are not included in the DVD (it is the same thing with ffdshow).

Consequently vobsub is the only choice. For example, I buy japan DVD anime from japan and there isn't subtitles in the DVD. I extract subtitles from.mkv (available in sites like tokyotosho.com) and I retime (if necessary) to see japan dvd with english subtitles (using MPC-HC and vobsub). But between XP and vista there is a delay of neraly 1610 milliseconds (If I compare the time of the start of the subtiles (available in.srt or.txt file), I have noticed that on vista, there is a delay of 1610 milliseconds (subtitles are always display at the same time but video start with neraly 1600 milliseconds of delay). It is the same delay for all dvd I tested. My video renderer is libmpeg2 (because i use ffdshow). I tried to use integrated mpeg2-video codec of MPC-HC but there is the same problem of delay of subtitles display with vista.

Yes it happen with all dvd with separated subtitles under vista. I give you 20 seconds of DVD japan anime (NADESICO) You must decompressed in a folder (video_TS) and use launch dvd function in MPC and open folder (video_TS) (not the.vob or.ifo, just the folder) I give you also the subtitles files to see subtitles (you must use vobsub to display these subtitles) You can test under xp and under vista and compare. Thanks, all subtitles are now displayed, however some subtitles are not displayed with the right timing.

In the video I gave you, there is one case: - 5 subtitles are displayed with a noticeable delay (ie. They start later than expected, however they end exactly as they should), ** In the 1min video I gave you, here are the subtitles which start with a delay: “8 00:00:12,679 -->00:00:15,148 Just relax and sleep.” “13 00:00:41,641 -->00:00:44,644 My name's Eluluu.” “15 00:00:47,714 -->00:00:50,984 And this is my house.” “17 00:00:52,986 -->00:00:57,323 Well, I found you collapsed in the forest.” “21 00:01:03,630 -->00:01:07,133 You're through the worst of it, that's what Grandma says.” EDIT: I made a mistake, I have said that one subtitle ends sooner than expected but it is wrong. In fact the original subtitle ends really slightly before the character stops speaking, this has leading me to make this mistake. Casimir666, have you suceed to eliminate the delay between voice and blu-ray subtitles display? VC-1 stutters with internal splitter? I'm so close to getting complete playback of BD straight from the disc, with DXVA disabled, so I can get a Gamma control, which my projector desperately needs. The biggest problem is VC-1 is choppy with the MPC-HD MPEG PS/TS/PVA splitter - which 'sees' TrueHD and names the streams.

It's choppy with any external decoder (I use ffdshow pointing at WMV9 currently). VC-1 is not choppy with the Haali Splitter, but that doesn't see TrueHD and doesn't name the streams.

Using the MPC splitter in another player gives the same result - it's the splitter. With the internal VC-1 decoder (but not DXVA) I get flashing blocks. I know the VC-1 choppiness issue has been mentioned, but it's one of two last stumbling blocks to full BD playback from disc (full DTS-MA in ffdshow audio is the other).so if you guys could fix it. I'LL BE YOUR BEST FRIEND FOREVER!!!

Sorry to pollute the development thread, I didn't want to, but it's driving me nuts. Would it be possible to turn MPC's shader post-processing into a stand-alone filter that could be added into any filter graph? Without a custom renderer (like MPC-HC's VMR9 and EVR custom renderer) it's not possible to use shaders since in that case the video is just some data in your RAM, not a texture on the graphics card. And I don't think you can just rip those renderers out of MPC and have other apps use them without support through the application itself.

Fleischmann - Even Your Glasses Miss Your Eyes (Angst Is Not A Weltanschauung!). Media Player Classic HomeCinema (x86), svn 934 (Media Player Classic HomeCinema (x64), svn 934 (Supported languages: CN, CZ, DE, ES, FR, HU, IT, KR, PL, RU, SK, TR, UA Changes log (FIXED: Bugs on presentation graphics (missing subtitles, better synchronization); FIXED: Audio change on M2TS files reset the selected subtitle; VSFilter resource fix; Updated FFmpeg; CHANGED: LibVorbis updated to v1.2.0.

Media Player Classic HomeCinema (x86), svn 934 (Media Player Classic HomeCinema (x64), svn 934 (Supported languages: CN, CZ, DE, ES, FR, HU, IT, KR, PL, RU, SK, TR, UA Changes log (FIXED: Bugs on presentation graphics (missing subtitles, better synchronization); FIXED: Audio change on M2TS files reset the selected subtitle; VSFilter resource fix; Updated FFmpeg; CHANGED: LibVorbis updated to v1.2.0. No standalone vsfilter build of this revision? Media Player Classic HomeCinema (x86), svn 934 (Media Player Classic HomeCinema (x64), svn 934 (Supported languages: CN, CZ, DE, ES, FR, HU, IT, KR, PL, RU, SK, TR, UA Changes log (FIXED: Bugs on presentation graphics (missing subtitles, better synchronization); FIXED: Audio change on M2TS files reset the selected subtitle; VSFilter resource fix; Updated FFmpeg; CHANGED: LibVorbis updated to v1.2.0. 1) What is 'vsfilter ressource fixed'? 2) I notice a bug with blu-ray subtitles in this version.

Sometimes a subtitles is displayed but doesn't disappeared. Consequently the following subtitles aren't displayed. When I use inverse telecine filter to deinterlace my video (because japan anime are often interlaced), the second subtitle ('burning') of utawarerumono is blocked and don't disappeared. The following subtitles aren't displayed. This problem didn't existed with 1.2.923 version Here is the video 'Here is the link for video (utaweruromo 1 min with subtitles) with subtitles embedded (265 MB) the other link is a txt file with all subtitles (1 kB) and here is the parameters that I use for inverse telecine (is telecide with avisynth) (3)FIXED: Bugs on presentation graphics (missing subtitles, better synchronization); the synchronization isn't perfect because some subtitles are displayed too late but it is maybe normal because casimir666 told me two days ago: I found where the problem is, but not how to fix it. You will have to wait a little.

Hi, i noticed that the new build 934 does disable dxva and falls back to software decoding if set to evr custom presenter under winxp (HD2600Pro). Is there any reason for that? In previous builds (same graphic card driver) it would use DXVA. On a sidenote. Is it possible to enable DXVA decoding for VMR7 renderless? It works in VMR7 windowed. Or are there technical reasons preventing this?

Regards, Casshern Under Windows XP - DXVA can't work with EVR, only VMR9/7. DXVA work with EVR/EVR Custom only in Vista. I don't think that DirectVobSub is used for DVD subtitles.

I think that is handled internally. Nevertheless, this delay issue should be fixed. With what video renderer(s) does it happen? Does it happen with ALL your DVDs? Is the delay always the same value?

With MPC-HC it is impossible to load separate subtitles when I see DVD subtitles. MPC-HC only allow to load subtitles included in the DVD but not subtitles which are not included in the DVD (it is the same thing with ffdshow). Consequently vobsub is the only choice. For example, I buy japan DVD anime from japan and there isn't subtitles in the DVD. I extract subtitles from.mkv (available in sites like tokyotosho.com) and I retime (if necessary) to see japan dvd with english subtitles (using MPC-HC and vobsub). But between XP and vista there is a delay of neraly 1610 milliseconds (If I compare the time of the start of the subtiles (available in.srt or.txt file), I have noticed that on vista, there is a delay of 1610 milliseconds (subtitles are always display at the same time but video start with neraly 1600 milliseconds of delay). It is the same delay for all dvd I tested.

My video renderer is libmpeg2 (because i use ffdshow). I tried to use integrated mpeg2-video codec of MPC-HC but there is the same problem of delay of subtitles display with vista. Yes it happen with all dvd with separated subtitles under vista. I give you 20 seconds of DVD japan anime (NADESICO) You must decompressed in a folder (video_TS) and use launch dvd function in MPC and open folder (video_TS) (not the.vob or.ifo, just the folder) I give you also the subtitles files to see subtitles (you must use vobsub to display these subtitles) You can test under xp and under vista and compare EDIT: With the 1.1.604 version separated subtitles can't be loaded when we want to see a dvd with launch DVD function of MPC-HC (load subtitle doesn't work). With the version more recently (1.2.908 and superior) load subtitle function works and I can load separated subtitles of DVD.

But the delay between voice and subtitle display is still present (as described above). I noticed another bug. Load subtitle function works only with VMR9 and not with EVR custom and VMR9 renderer is very bad under vista. One Tiny Bug I noticed, might as well mention it: In the properties dialog of for example, the mpeg2 filter, if 'Set interlaced flag in output media type' is set to on and click ok, when the dialog is loaded again, the Deinterlacing option is not greyed out.

Also one feature request: My goal in my htpc would be complete automation, so in time I would hope to have colour space conversion done and detected automatically. One thing I find myself doing alot however is choosing audio output in many dvds especially music dvds like queens greatest hits. Alot of these dvds default to stereo when they actually have a 5.1 track. Could a feature be added sometime in the future to allow people to add preferences for say 5.1 >2.1 or DTS >DD, and for mpc-hc to choose the audio track based on these simple rules. This would be fantastic and I'm sure bring alot of people along the way in terms of automating common tasks!!!! Under Windows XP - DXVA can't work with EVR, only VMR9/7. DXVA work with EVR/EVR Custom only in Vista Well thats not correct.

DXVA does work in Win XP with EVR. Just choose output EVR or EVR Custom Presenter and you will see that on an ATI CARD MPC-Videodecoder shows that its using the 'VC1-Bitstream decoder'. Also the CPU usage is about 10% compared to 70-90% when not using DXVA. I do not know what happens internally - but i also do have all the evr renderer options when i go to the property page of the video renderer. But i have to retract my previous post somewhat.

It also works with 934! Do not really know why it didn*t work before, but after a reboot it worked just like in all the old versions. Regards, Casshern. VC-1 stutters with internal splitter? I'm so close to getting complete playback of BD straight from the disc, with DXVA disabled, so I can get a Gamma control, which my projector desperately needs. The biggest problem is VC-1 is choppy with the MPC-HD MPEG PS/TS/PVA splitter - which 'sees' TrueHD and names the streams. It's choppy with any external decoder (I use ffdshow pointing at WMV9 currently).

VC-1 is not choppy with the Haali Splitter, but that doesn't see TrueHD and doesn't name the streams. Using the MPC splitter in another player gives the same result - it's the splitter. With the internal VC-1 decoder (but not DXVA) I get flashing blocks. I know the VC-1 choppiness issue has been mentioned, but it's one of two last stumbling blocks to full BD playback from disc (full DTS-MA in ffdshow audio is the other).so if you guys could fix it. I'LL BE YOUR BEST FRIEND FOREVER!!!

Sorry to pollute the development thread, I didn't want to, but it's driving me nuts. All I need is VC1 support fixed and everything can played from disc. VC-1 stutters with internal splitter?

I'm so close to getting complete playback of BD straight from the disc, with DXVA disabled, so I can get a Gamma control, which my projector desperately needs. The biggest problem is VC-1 is choppy with the MPC-HD MPEG PS/TS/PVA splitter - which 'sees' TrueHD and names the streams. It's choppy with any external decoder (I use ffdshow pointing at WMV9 currently). VC-1 is not choppy with the Haali Splitter, but that doesn't see TrueHD and doesn't name the streams. Using the MPC splitter in another player gives the same result - it's the splitter.

With the internal VC-1 decoder (but not DXVA) I get flashing blocks. I know the VC-1 choppiness issue has been mentioned, but it's one of two last stumbling blocks to full BD playback from disc (full DTS-MA in ffdshow audio is the other).so if you guys could fix it.

I'LL BE YOUR BEST FRIEND FOREVER!!! Sorry to pollute the development thread, I didn't want to, but it's driving me nuts.

All I need is VC1 support fixed and everything can played from disc. I too concord that the VC-1 issue (MPEG PS/TS/PVA internal splitter) should become a priority! It's the only MAJOR problem keeping MPC to become a VERY GOOD player!:thanks: I do appreciate dev's hard work and all I'm saying it's that this should be at the TOP of the bug list! It's there a FLAC support that I can't find?:helpful. I have a large problem with MPC:HC that doesn't occur with MPC. I am not sure if it's a misconfigured setting or a problem with the app itself but whenever I stream media from my computer to my television using HDMI from a radeon 4550, MPC:HC creates audio static, audio lag at times and even once in a blue moon loss of video signal for a few seconds.

This does not happen with MPC or windows media 11. I can reproduce this error on two computers with completely different setups. The only thing they share in common is the software used to playback media (MPC:HC / ffdshow) and the video card (radeon 4550).

Using realtek hdmi driver 2.09 if that makes a difference. MPC-HC is very good player for DXVA. More accurately, he is only one, which can plays files with dxva acceleration.:) Im using Win XP SP3. But I have two problems: 1) I can not using subtitles (Im not english speaking naturarely, I need to see subtitles (eg. *.srt, *.sub) at movie. I have two choices.

Movie with subtitles, but without dxva. Or dxva using, but without subtitles.:confused: When i switch from EVR renderer to VMR9 renderer, I can see subtitles, but dxva decoder is fail in = movie is jerky (I see only P-frames?). 2) DXVA is fully functional only for standard resolution (eg. 720x576, etc). When I am using nonstandard resolution, eg. 576x320px, I get strong tearing in fullscreen mode. Note: When I switch to VMR9 Renderless and playing MP4 file, subtitles are corrrectly displayed and DXVA is functional.

But MKV is still problematic. I think, that internal MKV splitter in MPC is PROBLEM. The main motivation for this proposal is to enable hardware accelerated video decode at various entry-points (VLD, IDCT, Motion Compensation etc.) for the prevailing coding standards today (MPEG-2, MPEG-4 ASP/H.263, MPEG-4 AVC/H.264, and VC-1/VMW3). Extending XvMC was considered, but due to its original design for MPEG-2 MotionComp only, it made more sense to design an interface from scratch that can fully expose the video decode capabilities in today's GPUs. The current interface is focused on video decode only and is window system independent, so that potentially it can be used with graphics sub-systems other than X.

Media Player Classic HomeCinema (x86), svn 939 (Media Player Classic HomeCinema (x64), svn 939 (Supported languages: CN, CZ, DE, ES, FR, HU, IT, KR, PL, RU, SK, TR, UA Changes log (Updated FFmpeg; FIXED: Bugs on presentation graphics, missing subtitles, better synchronization. I try this version and I noticed that the bug of subtitles display blocked has desappeared: 2) I notice a bug with blu-ray subtitles in this version. Sometimes a subtitles is displayed but doesn't disappeared. Consequently the following subtitles aren't displayed. When I use inverse telecine filter to deinterlace my video (because japan anime are often interlaced), the second subtitle ('burning') of utawarerumono is blocked and don't disappeared. The following subtitles aren't displayed. This problem didn't existed with 1.2.923 version Here is the video 'Here is the link for video (utaweruromo 1 min with subtitles) with subtitles embedded (265 MB) the other link is a txt file with all subtitles (1 kB) and here is the parameters that I use for inverse telecine (is telecide with avisynth) (But synchronization of subtitles isn't still perfect in utawarerumono.

Maybe this is a good time to mention the fact that haali media splitter is closed source, and all developement that is not related to coreavc seems to have stopped. I've actually been hoping for an open-source alternative. Maybe we can unify the splitter in MPC-HC and allow it to be used seperately (like the dxva decoders) ideally it would get its own sub-section in the SVN (unlike how its semi hidden now) to attract more dev's. Also it would be cool if it was a smart as XBMC's splitter(which almost seems to detect anything in any extention/container).

The internal splitter in MPC-HC definitely needs to be improved to reduce dependence on Haali splitter, which is not open source. At the moment, it doesn't support ordered chapters in mkv(correct me if wrong, been trying some samples and doesn't work for me) which makes it less useful as a single standalone player which doesn't need installing that you can just throw on a USB flash drive. Also because Haali is not open source and hasn't been updated for months, there's no way to determine whether it will be forward compatible with future OS like Windows 7 and the next major release, Windows 8(?) or whatever it's gonna be called. Maybe this is a good time to mention the fact that haali media splitter is closed source, and all developement that is not related to coreavc seems to have stopped. I've actually been hoping for an open-source alternative. Maybe we can unify the splitter in MPC-HC and allow it to be used seperately (like the dxva decoders) ideally it would get its own sub-section in the SVN (unlike how its semi hidden now) to attract more dev's.

Also it would be cool if it was a smart as XBMC's splitter(which almost seems to detect anything in any extention/container). The splitters are already available as standalone filters. They have been for years. And I doubt that Haali has stopped development. Considering he's given me new beta versions of HMS/HR a few weeks ago, I can confirm that the development is far from over.especially with MKV 2.0 around the corner;) Is there any information out there about these 2? Maybe a SVN changelog or roadmap?

Apart from the?private? Build you seem to have both of these have not been officially updated for about half a year. Still even if they have been haali splitter is still closed source, and features/changes/fixes are only added when haali wants to add them.

I'm not saying haali splitter is a bad thing, in fact its great. But i prefer not to put all my eggs in one basket.

Is there any information out there about these 2? Maybe a SVN changelog or roadmap? No whatsnew section, so I dunno what's changed in HMS.just the version number is 1.8.365.21 well it's not like you have much options anyhow.Gabest's MKV splitter doesn't support 1.5mbit DTS tracks(random silences) and neither does it support multiple soundtracks or seamless branching.the good side being that it provides better jitter in Reclock than HMS. The only other MKV splitter I know of is KPMlayer's but they don't release external filters, so it only works in KMP(except if someone knows how to extract it from the main exe?). Do you have some samples to provide which exhibit these issues?

Could you test this: 1) DTS - 5.1 - 1536kbps (2) AVC + AC3 2ch & AC3 6ch ((sound starts at ~23s) because those samples seem to be ok for me w/ the rev 940. I dunno what rev it is, but I'll try w/ MatroskaSplitter_20081210.zip and I'll report back:cool: last time I tried, I had some very random silences on 1.5mbit DTS(like if the frames were too long?) and when there was several soundtracks it would just play them all at once:eek. Media Player Classic HomeCinema (x86), svn 939 (Media Player Classic HomeCinema (x64), svn 939 (Supported languages: CN, CZ, DE, ES, FR, HU, IT, KR, PL, RU, SK, TR, UA Changes log (Updated FFmpeg; FIXED: Bugs on presentation graphics, missing subtitles, better synchronization. I try this version and I noticed that the bug of subtitles display blocked has desappeared: 2) I notice a bug with blu-ray subtitles in this version. Sometimes a subtitles is displayed but doesn't disappeared.

Consequently the following subtitles aren't displayed. When I use inverse telecine filter to deinterlace my video (because japan anime are often interlaced), the second subtitle ('burning') of utawarerumono is blocked and don't disappeared. The following subtitles aren't displayed. This problem didn't existed with 1.2.923 version Here is the video 'Here is the link for video (utaweruromo 1 min with subtitles) with subtitles embedded (265 MB) the other link is a txt file with all subtitles (1 kB) and here is the parameters that I use for inverse telecine (is telecide with avisynth) (But synchronization of subtitles isn't still perfect in utawarerumono I give another example of video with subtitle embedded (same anime but not the same episode: utawarerumono 30seconds) where subtitles are not well synchronized.

Here is video 00001.m2ts - 125.30MB (here is subtitle file to see good timing http://www.zshare.net/download/7d4bd6/. I've been using FFDShow MT (experimental) for 2 weeks and I'm in shock how fast it is on multi core system (MPEG-2 and H.264) Despite 'experimental' label I haven't had any problems with decoding. There is only small problem during seeking. If you rewind to none-key frame you get distorted image till new key frame.

I can live with that because speed is more important for me now. I get on my Q6600@3Ghz 80+fps with insane footage (1920x1080p 59.94fps @ 50Mbps)!!! My 4850 has small problems with decoding. I'm not surprised because 59.94fps with FullHD belongs to level 5.0:) i cheated with level profile.

I forgot to ask:) Can we get true multi-threaded mpeg-2/H.264 decoding in MPC-HC? I am having the same flashing problem with my PCI-e ATI 3650 card.

Running XP and Catalyst 8.10 and MPC build 906 Picture flashes back a bit, sometimes macroblocks appear, sometimes pictures flashes back and forth and get stuck in a loop. Any solution for that? I also have this problem with a ATI 2600 HD PRO AGP under WINXP and VMR9/EVR.

But it is not certain that this really is a problem of the MPC HC DXVA decoder or the ATI drivers. The 4xxx series seems to work better. Under Vista the problem is reduced - turning of aero and all graphical effects under evr seems to work. But you can use VMR7 or OVERLAY with DXVA this works fine (i use the catalyst 8.5 drivers as they are the most stable ones). Unfortunatly you cant display subtitles or use shaders. Maybe the devs could try to make dxva work under vmr7 renderless.

Also on the source repository there is some talk that mpc hc might initilalize VMR9 wrong for DXVA so maybe there is a solution after all. Regards Casshern. Well thats not correct. DXVA does work in Win XP with EVR. Just choose output EVR or EVR Custom Presenter and you will see that on an ATI CARD MPC-Videodecoder shows that its using the 'VC1-Bitstream decoder'. Also the CPU usage is about 10% compared to 70-90% when not using DXVA. Regards, Casshern No, sorry, there is no support for DXVA with EVR in Windows XP.

Even if the codec property pages say bitstream the cpu still does the processing. There is no support whether VC-1 or h.264 is used. I use an ati card as well (4850) and used the mpc codecs with the typical results - no support for dxva with EVR in XP. No, sorry, there is no support for DXVA with EVR in Windows XP. Even if the codec property pages say bitstream the cpu still does the processing.

There is no support whether VC-1 or h.264 is used. I use an ati card as well (4850) and used the mpc codecs with the typical results - no support for dxva with EVR in XP. No - the CPU does not do the decoding.

And Yes its DXVA. The ATI2x00 and 4x00 use a different DXVA engine, maybe thats why its not working for you. But i doublechecked with another 2x00 owner and it works for him too. We both use Catalyst 8.5, winxp sp3 and the evr property dll somebody posted here a long time ago.

But at the moment you do not loose much, as this mode behaves similar to VMR9 in that DXVA shows some corruption and 'jumping' frames. That's why i suspect it falls back to some VMR9 compatibility mode. It works better under VISTA. Wow EVR working with DXVA in MPC-HC under XP, wonder if that also works for Nvidias Driver, did you tried it with another DXVA Decoder like Arcsoft, Cyberlink, Elecard?:) Strange it seems to only work with VC-1 i wonder why H.264 gets blocked except if your stream gets rejected anyways by MPC-HC to get processed by the GPU and seeing it's Anime i guess the Encoder didn't take the restrictions into account and so it falls back to Software Decoding:) How is the CPU and Kernel utilization with EVR (DXVA) under XP compared to VMR9 (DXVA)? It seems like seeking of m2ts muxed from 264 video files puts the a/v out of sync. At least for me.

But if m2ts is muxed from mkv file seeking is ok. I tried MPC-HC 908 and 940 with internal, coreavc, ffdshow-mt decoders with old renderer, overlay, VMR7/9, haali, and EVR. I couldn't test DXVA, old video card. WMP 11 and PowerDVD 8 seeking it stays in sync.

I posted some examples here (only the encoded link is useful, the BDRB.m2ts is problematic, the others were muxed from mkv files and are ok. DXVA with EVR and EVR custom seems to works with Nvidia also. No corruption or other problems, only thing is, subs and the 'Display stats' dont work, then again, subs + Display stats work when DXA is not enabled. I got the evr.dll from microsofts.net frame 3.0. But it shows the EVR renderer only with nonDXVA, with DXVA enabled its VMR.

So i think it falls back to vmr7 maybe when the file is DXVA compliant. Later edit: this is only for H.264, as nvidia cards dont support VC-1 bitstream decoding. No - the CPU does not do the decoding. And Yes its DXVA. The ATI2x00 and 4x00 use a different DXVA engine, maybe thats why its not working for you. But i doublechecked with another 2x00 owner and it works for him too. We both use Catalyst 8.5, winxp sp3 and the evr property dll somebody posted here a long time ago.

But at the moment you do not loose much, as this mode behaves similar to VMR9 in that DXVA shows some corruption and 'jumping' frames. That's why i suspect it falls back to some VMR9 compatibility mode. It works better under VISTA. No offense but this is a lot of misinformation.

If EVR is not in the filter chain it DOES NOT work. Based on some of the other responses, it looks like even though you select EVR it is defaulting to VMR7 for playback (which supports DXVA in XP). Please show your screenshots of EVR working with DXVA in XP. To reiterate EVR with DXVA does NOT work with XP.

Of course, it works better in Vista BECAUSE it actually works;) Not trying to be confrontational but I can see someone reading that kind of statement and spending hours troubleshooting and researching why they can't get DXVA with their ATI 2x000 cards working in XP. No offense but this is a lot of misinformation. If EVR is not in the filter chain it DOES NOT work. Based on some of the other responses, it looks like even though you select EVR it is defaulting to VMR7 for playback (which supports DXVA in XP). Please show your screenshots of EVR working with DXVA in XP.

To reiterate EVR with DXVA does NOT work with XP. Of course, it works better in Vista BECAUSE it actually works;) Not trying to be confrontational but I can see someone reading that kind of statement and spending hours troubleshooting and researching why they can't get DXVA with their ATI 2x000 cards working in XP. Before posting - just try it. Or look at the screenshots px posted here. That's the same i am getting and the same the other posters are seeing. EVR is in the filter chain - what it does internally is still up to discussion, but it DOES NOT fall back to VMR7 (windowed or not). More likely it falls back to some sort of VMR9 compatibility mode, if it falls back at all - as the EVR property page is fully functional.

Please stay to the facts, everything else is misinformation. LOL yeah that is really strange now, except maybe Casimir666 implemented also a DXVA force so when you select DXVA as Decoder it uses the Renderer capable of providing it ignoring your selection and if it can't find any falls back to the Renderer you selected and Software Decoding would make perfect sense:D Actually the EVR renderer is in the filterchain. So any fallback has to be internal to the EVR renderer. Internally EVR might have some VMR9 compatibility mode.

On the other hand the EVR property page works. What EVR under XP does internally is still up for discussion, but from a users perspective it just means he has an additional renderer option to use DXVA with VC1-Bitstreams. Does anyone else experience blurring at the bottom of the video (last 8 lines or so) when accelerating H.264 or VC-1 videos with ATI cards? The problem does not show up on most Blu Ray videos, because most have black bars at the top and bottom. The ones that do have full frame video have this problem. Playing the discs in PowerDVD does not show the problem, however using PowerDVD's DXVA filter in MPC does show the problem, so I don't think MPC HC's DXVA implementation is to blame. I also have the problem with my x264 encodes of TV shows I have recorded, since they are mostly full frame (1920x1080).

I had this problem with my Radeon 3450, and was happy to see that a GeForce 8500 did not have this problem (this card won't bitstream VC-1 though). I recently upgraded to a Radeon 4650, hoping that this issue had been fixed (on ATI's side) and dreaming of one day getting a receiver with HDMI audio capability. To my surprise, this blurring issue still remains!

Needless to say I was not happy, and am currently awaiting a new GeForce 9400 Motherboard hoping that Nvidia will continue to work better. I am using Windows XP SP2, and the System default renderer.

I tried other renderers, and the problem seems to go away if I use VMR9, but I get bad tearing when I use that one. Anyone else experiencing this? Are you sure your using DXVA (check the stats overlay). Also what CPU do you have (because i think it's too slow for 22mbits of h.264 thus why you have your problem) Just thought I would report back on my curious lack of success due to regular glitching playing h264 files from HD DVD with DXVA. Yes, I am certain DXVA is working as the CPU usage is down in the single digits to teens. Processor is an X2 @ 2.4G.

The files play fine with software decoding (and on my Popcornhour media extender). I also determined that there is no issue with VC-1 HD DVD rips as I use the WM decoder+Haali to play those. After trying to mux Babel various ways with various tools the end result is always the same. Just tried the Transformers HD DVD with the same results, this time using h264info to remove the pulldown instead of eacto, i.e. Regular glitches with the MPCvideo decoder+Haali but plays fine in software.

Using VMR9 renderless in MPC-HC but have tried Overlay with no effect. If someone has their h264 HD DVD rips playing nice with MPC-HC and Nvidia graphics, I would appreciate it if they would share any detailed settings in MPC-HC, the decoder, or graphics drivers that could getting me going in the right direction. I have a problem playing an MKV file that has an unusual characteristic - it has two audio streams, one DTS and one 'Vorbis'. The encoder made the Vorbis stream as the default, evidently, because it comes up with that one playing. I am using Haali Splitter and when I select the DTS track, it does not work.

My guess is that since DTS requires a different group of filters, it would seem that Haali Splitter audio selection does not change the filters loaded by MPC. I tried enabling the Internal Audio Switcher, but it only shows one choice of 'MKV file audio'. ( I can play the file with the Vorbis audio without problems, and I can play other MKV files that only have DTS audio tracks, but I cannot change from Vorbis audio to DTS audio using Haali Splitter. ) Any thoughts? I have my HTPC connected to an analog monitor and to my Sanyo Z2000.

I have two sound cards installed, first, the motherboard one without spdif, configured as 'system default' and connected to my monitor speakers. The second one, a C-Media with six channels output and SPDIF out connected to my receiver by analog and optical. My HTPC is my work PC too, so I have 2 copies of MPC-HC configured to use.ini files. My work copy, don't have D3D activated so I can test files and see simple things and configured to use the 'system default' sound so I can hear things through my monitor speakers. My HTPC copy, has D3D activated, send the full screen to my second monitor (projector) and configured to use the C-Media audio device to send, normally, DD or DTS through SPDIF to my receiver. The problem is that if I set the output of the AC3/DTS filter to spdif in one copy, the other copy get this configuration too.

It seems that the preferences of this filter is not saved to the.ini file so every time I 'work' with my work copy of MPC-HC and change the preference of the AC3/DTS filter to analog output, when I want to use my HTPC copy, I have to open it without load any file, and change the settings of the audio filter prior to load the file I want to play. It would be very useful if this preferences are saved to.ini files too. I had reported some time ago the following: I have noticed that using VMR9 exclusive full screen the slider to move around the movie diseapears after only 1s even if the mouse pointer is still on top of it.

With old releases (more than 6 months ago), one had to move away the mouse pointer to make the slider disappear. I would be grateful if you could put back the old behaviour or at least to offer an option as it is now very hard to use the slider.

I didn't get any feedback. Am I the only one who has this problem? I would be grateful if you could post a fix or give a workaround since because of this 'bug', it makes quite hard to use the latest version with all the great new features. Thanks again for your great product and happy new year 2009!

According to x264 wiki, we can use 5 refs with a resolution of 1920x816 and keep dxva support. I am trying it, but mpc video decoder is not working very well. It runs with dxva, but some of the frames show corrupted squares. In software mode, it works fine, no corrupted squares.

I uploaded the file here: I am using XP SP3, mpc-hc v1.2.940.0, GF8600GT, 178.24 drivers. Would this be a mpc decoders bug, a drivers bug, a x264 bug, or simply I can only use 4 refs? With 4 refs, it works fine. I just read on wikipedia that the radeon 4300/4500 have a minor revision of UVD called UVD2.2 According to the AMD website and wikipedia article and several forums: -These card support full bitstream decoding of MPEG2(in UVD instead of the previous shader assisted decoding) -These cards support Assisted decoding of MPEG1 and DIVX -These cards fix compatibility with h264/vc1/mpeg2 files (possibly allowing for L5.1 decoding) Does anyone have access to one of these cards? (the 4300 series is not recommended for any deinterlacing/post-processing testing). Media Player Classic HomeCinema (x86), svn 940 (Media Player Classic HomeCinema (x64), svn 940 (Supported languages: CN, CZ, DE, ES, FR, HU, IT, KR, PL, RU, SK, TR, UA Changes log (Updated FFmpeg Silly question of the day. Every time I download one of these there are a lot of dlls with the exe.

Do I need to do anything with the DLL or just have them in the same folder I unpack the mpcplayer? I have a problem playing an MKV file that has an unusual characteristic - it has two audio streams, one DTS and one 'Vorbis'. The encoder made the Vorbis stream as the default, evidently, because it comes up with that one playing. I am using Haali Splitter and when I select the DTS track, it does not work. My guess is that since DTS requires a different group of filters, it would seem that Haali Splitter audio selection does not change the filters loaded by MPC. I tried enabling the Internal Audio Switcher, but it only shows one choice of 'MKV file audio'. ( I can play the file with the Vorbis audio without problems, and I can play other MKV files that only have DTS audio tracks, but I cannot change from Vorbis audio to DTS audio using Haali Splitter.

) Any thoughts? Would be nice to have some of the bugs here acknowledged by the devs. You mean: ' Well, we're not going to fix that. Oh, that one we're not going to fix. No way I'm going near that one. ' The problem with open source is that everyone is a volunteer, and ultimately there is no compulsion or authority, and developers are free to waste thousands of hours with DXVA that will be totally obsolete in 2-3 years, and is largely impractical right now due to the large number of inconsistencies and nonconformities. But then again, it's been over 25 years, and Microsoft still hasn't fixed the major bug of windows stealing focus either.:rolleyes.

Hi, In recent builds the default colorspace was changed from YV12 to YUV2 in order to help hardware acceleration this however made performance impact on video playback when using some video cards. Is it possible to make a little checkbox which colorspace to 'prefer'? I'm also not sure why output pin reports a different resolution than the video itself. This has been tested with rev901 vs rev940 on a laptop (GMA950) with a native resolution of 1280x768 using VMR9 (Windowed) using standalone filters in MPC-HC rev940. Rev901 Input pin: Video: MPEG4 Video (H264) 1440x1080 (16:9) 50.00fps Output pin: Video: YV12 1536x1080 (16:9) 50.00fps A bit choppy but at least it plays somewhat rev940 Input pin: Video: MPEG4 Video (H264) 1440x1080 (16:9) 50.00fps Output pin: Video: YUY2 2048x1080 (16:9) 50.00fps Slideshow Keep up the good work //Danne.

Hello, i plan to buy a new grahic card explicitly for using dxva with mkv movies. My question now is what should i buy, ati or nvidia? I am having a nvidia 8600 GT but i am not contently with my card and dxva.

Some movies get to blocky with dxva. I would like to buy an ati hd4350, so i hope to get a better dxva support. Or is a newer nvidia 9xxx card better? DXVA has limitations, and on h.264 encodes that exceed these limitations you'll get blockiness. See here for more info.

And I suggest you stick with your current card and just use a software decoder for those troublesome files and save yourself some money. Unless your cpu can't handle 720p/1080p h264 decoding? DXVA has limitations, and on h.264 encodes that exceed these limitations you'll get blockiness. See here for more info. I suggest you stick with your current card and just use a software decoder for those troublesome files and save yourself some money.

Unless your cpu can't handle 720p/1080p h264 decoding? 720p isn't a problem for my cpu, just some 1080p files with higher bitrates. So you think, i won't improve dxva for mkv files with a newer card, like ati 4xxx? I am using the lastest version of ffdshow-mt and for my nvidia 8600gt i downloaded the lastest beta driver. It was around here also note the post below stating 11 reference frame limitation. I'm not sure that ATi can even playback L5.1 streams to begin with (my last ATi card was an x800 XT) I don't think it will be better switching camps. With regards to nVidia cards, afaik all their 8xxx + series with the exception of a few have the same DXVA features with regards to GPU decoding.

See here: Doesn't include the 9xxx series or the 2xx's. Note: Unless I'm wrong, Purevideo = DXVA So the short answer is, it won't get better by upgrading your GPU. Perhaps give coreavc a try? Although based on my own benchmarks, the speed gap between coreavc and ffdshow-mt has considerably lessened recently. Additional info regarding Purevideo from Neuron2's development dialog with nvidia Quote nvidia engineer: 'When getting full VP utilization, without any 3D bottleneck, the perf of the 8500GT should be very close to a 9800GTX, since IIRC the video processor is clocked at the same speed on both, though this could depend on the manufacturer (usually it is running at 450MHz, but some low-end GPUs are only running VP at 400MHz).' That is for CUDA h264 decoding, but assuming that DXVA utilizes the same video processor they are talking about, then you would get a minor improvement by upgrading to a more expensive nvidia card.

It was around here also note the post below stating 11 reference frame limitation. I'm not sure that ATi can even playback L5.1 streams to begin with (my last ATi card was an x800 XT) I don't think it will be better switching camps. With regards to nVidia cards, afaik all their 8xxx + series with the exception of a few have the same DXVA features with regards to GPU decoding.

See here: Doesn't include the 9xxx series or the 2xx's. Note: Unless I'm wrong, Purevideo = DXVA So the short answer is, it won't get better by upgrading your GPU. Perhaps give coreavc a try? Although based on my own benchmarks, the speed gap between coreavc and ffdshow-mt has considerably lessened recently.

Additional info regarding Purevideo from Neuron2's development dialog with nvidia Quote nvidia engineer: 'When getting full VP utilization, without any 3D bottleneck, the perf of the 8500GT should be very close to a 9800GTX, since IIRC the video processor is clocked at the same speed on both, though this could depend on the manufacturer (usually it is running at 450MHz, but some low-end GPUs are only running VP at 400MHz).' That is for CUDA h264 decoding, but assuming that DXVA utilizes the same video processor they are talking about, then you would get a minor improvement by upgrading to a more expensive nvidia card. Ok, thank you for your explaining to me. I think the best for me is keeping my current video card. I noticed today that I get about 80ms jitter with EVR CP with all recent builds, anyone else noticed this?

Last version that I know that works fine is rev 922. I also get 80ms jitter with EVR Custom Presenter with rev 934 or later (including rev 948) when using ffdshow. Jitter is 0ms up to and including rev 929. 80ms jitter also when using internal MPC-HC AVC ffmpeg decoder. However using DXVA I get 0-1ms jitter (rev 948). This is on Vista x64, all the other renderers are unaffected. Note: my testing is based on available builds from xvidvideo.ru so it maybe be possible this bug occurred anywhere between 929 and 934.

I also get 80ms jitter with EVR Custom Presenter with rev 934 or later (including rev 948) when using ffdshow. Jitter is 0ms up to and including rev 929.

80ms jitter also when using internal MPC-HC AVC ffmpeg decoder. However using DXVA I get 0-1ms jitter (rev 948). This is on Vista x64, all the other renderers are unaffected. Note: my testing is based on available builds from xvidvideo.ru so it maybe be possible this bug occurred anywhere between 929 and 934.

You mean the jitter displayer with 'Ctrl+J'? Can you post a screenshot? It's not only displayed with CONTROL + J, but also in the EVR properties, in quality stats. Also displayed with CTRL + 4, in the statistics. EDIT: screenshot added (Ok i have changed the fps calculation for EVR Custom in rev 931, because on my machine the jitter curve was not centered. Now it's ok for me but not for other:mad. This is only a display problem, and should affect render.

Ok i have changed the fps calculation for EVR Custom in rev 931, because on my machine the jitter curve was not centered. Now it's ok for me but not for other:mad.

This is only a display problem, and should affect render. Casimir666 have you succeed to synchronize subtitle of blu-ray because even in the last version (1.2.948) some subtitles are still not synchronized with voice?

Sometimes there is 2 to 4 seconds of delay between start of sentences and subtitles display? And I notice that 'load subtitle' function doesn't work with evr custom selected. With vmr9 renderless rendere selected 'load subtitle' function works but not with evr custom renderer selected. Subtitle synchronization or load subtitle function?

Because on some Blu-ray synchronization is perfect (ghost in the shell for example) but in some blu-ray (utawarerumono or die buster) I have a delay for some subtitles between voice and the display of subtitle. For load subtitle function, when i select vmr9 renderless, subtitle are displayed but with evr custom renderer selected subtitles are not displayed or very bad under xp and under vista and with 2 different configs. Load subtitle.

I never experienced any problems with that. ADude I am using Haali Splitter and when I select the DTS track, it does not work. Have similar problem, selecting different stream in Haali ends with no sound at all, and with internal splitter player tries to render both streams simultaneously (ffdshow audio decoder tray icon doubled) and also no sound at all.

Even AVI files with two&more streams plaqued. The 'Audio' item in 'Play' menu grayed out, internal audio switcher not work (seems like it not included into last builds or not inserted by player into graph). MPC-HC build 940 from xvidvideo.ru, build 908 from sourceforge also affected. It works fine for me. Are you sure you have the internal audio switcher enabled in MPC? It seems to be entirely random whether anyone actually reads the problem reports in this thread.

I posted the following twice before: I have a problem playing an MKV file that has an unusual characteristic - it has two audio streams, one DTS and one 'Vorbis'. The encoder made the Vorbis stream as the default, evidently, because it comes up with that one playing. I am using Haali Splitter and when I select the DTS track, it does not work. My guess is that since DTS requires a different group of filters, it would seem that Haali Splitter audio selection does not change the filters loaded by MPC. I tried enabling the Internal Audio Switcher, but it only shows one choice of 'MKV file audio'. ( I can play the file with the Vorbis audio without problems, and I can play other MKV files that only have DTS audio tracks, but I cannot change from Vorbis audio to DTS audio using Haali Splitter.

) Any thoughts? I've been meaning to ask this for awhile since I upgraded to Vista a few months back. When Options >Player >'Keep history of recently opened files' is checked, opening a file via File >Open file.

Or ctrl-o will show the current or last opened filename in the Open file dialog. However as soon as I click Browse in the Open file dialog, it defaults to the last folder I used in Firefox or whatever app I used last that uses the open or save file dialog. I remember when I was using XP (and an older build of MPC-HC) the browse button would point to the folder of the currently opened file or last played file.

Basically I would like the old behavior back where browse would point to the folder that contained the last opened file or currently opened file. I'm not sure if this behavior is related to Vista or the later builds of MPC-HC. In every movie i got the same problem like in the attachement Asus ATI 3450 Catalyst 8.12, VMR 9. This is a know bug, as far as i know no-one has been able to find the cause/solution. It has to do with the way MPC-HC gives the video-stream to the videocard, ATI expects a different input from nvidia.

What makes things even worse is that ATI tweaks these settings between driver releases causing different results everytime. Nero Vision Portable Ita Download Firefox. Seeing as PowerDVD can playback all the media perfectly on any driver version(without updating the PowerDVD.exe) there must be a 'safe' way of giving the slices to ATI cards. I get these errors for both H264 and VC1. If you would also have read this thread then you would have seen that I replied to jffulcrum's post.:mad: You should use a different tone if you want people to help you. What are you using to decode DTS? I use ffdshow.

Yes, but you asked jffulcrum a question which was answered in my post. My post was posted twice, and jffulcrum's post was a reply to my post. In all three of the problems I've posted over the past year that no one has responded to, I have already played the file in another player before posting, so I don't need help. The post is a problem report - because you can bet that for every one problem report, 100 people like jffulcrum have also encountered the problem and most don't bother to report it. For a program to be quality, bugs need to have higher priority than new features.

I appreciate that all the development is by kind volunteer effort, but fixing bugs and reading bug reports needs to have a better system than 'whatever posts happens to be noticed because they are interesting'. I don't decode DTS, all 5.1 is passed to the receiver intact on the SPDIF by 'SPDIFER' open source filter. Vorbis is decoded to PCM, and PCM is handled by 'ACFILTER' open source filter. I am not sure how dual audio is suppoed to work, either all the filters are supposed to be loaded at startup for all audio, or else changing audio in Haali is supposed to cause different filters to be loaded, but neither is happening. For a program to be quality, bugs need to have higher priority than new features. I appreciate that all the development is by kind volunteer effort, but fixing bugs and reading bug reports needs to have a better system than 'whatever posts happens to be noticed because they are interesting'.

There is a real bug-tracker available om the site: Now i see bugs that have not been touched at all yet that date back to september 2007. Maybe its a good idea to add the bugtrack to the startpost Same for the feature request tracker: Every bug discussed here should have a tracker ID If someone is unsure if the problem in question is a bug or a user error they can post here, but as soon as it appears to be a bug a ticket should be made before continuing, any updates to the situation can be added to the ticket.

That does mean however than all the active DEV's will have to clean up both lists. Well, after digging deeply inside ffdshow audio decoder settings i solved problem with 'missed in action' MCP internal audio switcher. My error was an 'Connect to' option in 'Output' page of decoder, it must be 'any filter'. But MPC still tries to render all streams simultaneously ('Audio' item in 'Play' menu show only one stream), and still no sound, as well as after changing stream in Haali or in ffdshow decoder internal stream switcher (i not knew that them has it). When i receive good internet connection i will upload my samples and ask ffdshow-tryout team about it.

Hello everyone, I finally taken the time to setup MPC-HC for development so I can fix a couple of issues that have been irritating me for a long time. What is the easiest way to get changes to the devs so they can give input and maybe include the changed code in SVN? I'm guessing that SVN submit access is out of the question at this stage. I have been working on the following areas: Custom EVR The sync for the custom EVR was practically nonexistent. Aside from being unable to drop frames when the movie fps is about equal to the screen refresh it also uses a non-robust method for syncing.

I have now fixed the sync to be as perfect as is possible. For example when you run a 23.976 movie in 24 fps instead of stuttering several frames as the Haali renderer does it only stutters once to fix the sync. I have added a new statistics graph to visualize audio/video sync in addition to the jitter display. In debugging I also found out that the sample queue didn't fill up as quickly as it should. I thought that this was because you didn't get any samples from the mixer while waiting for VBlank in Present, so I moved the getting of samples from the mixer to a separate thread. Now I'm not sure if this was an issue, but it shouldn't hurt.

In addition I had to create a separate lock for the sample queue, otherwise the thread would not be able to work at the same time as wating for VBlank. In addition to this the decoder is not able to lock new surfaces while waiting for VBlank, so I had to do a manual sync to VBlank and remove the VSync in Present. This relies on m_pD3DDev->GetRasterStatus. I'm not sure if this is available on all graphics cards. But if it is, it should fix all sync to vblank issues and remove the need for the lock surface hack. I also had trouble playing VC1 content with the software decoder (DXVA hangs after a couple of seconds and the computer hangs shortly after). Because the internal software decoder was broken for m2ts files I used the default VMR decoder.

This caused 13 fps playback with lots of dropped frames after I had fixed the sync. Turns out GetSampleTime was unreliable and the frame times were switched around. I fixed this by correcting the sample times if they were outside a specific range. Not sure if this could break other things. Also GetSampleDuration always returned 1, so I changed to calculate the subPicQueue fps from m_rtTimePerFrame instead. All these fixes made it possible to playback VC1 content in software without issues where it would start to stutter before because of the sample queue not being filled because of threading issues.

It also fixed the same issues with other software decoders that require much CPU. File streaming performance I changed MINPACKETS in base splitter to 100 and MINPACKETSIZE to 256*1024 and DEFAULT_CACHE_LENGTH from 2048 to 64k. This made it possible for me to stream bluray content over my 1GBit network with the internal splitter. This would only work before with the haali splitter, but the haali splitter doesn't support MLP and other hd audio stream formats.

Hdmv LPCM Changed a calculation in ProcessHdmvLPCM that only worked for power of 2 number of channels. Further work I'm planning on adding the following features and fixes in the future: * Fix the broken sync for DTS passthrough on Radeon HD cards. This is because the sound time source is unstable, should be possible to smooth it out if ATI can't bother to care. * Add support for automatically changing display refreshrate through PStrip with configurable mode lines. * Add support for automatically changing the number of audio output channels based on content. This is because my receiver requires the correct number of channels to use the correct expansion modes to 7.1.

* Add support for DTS-MA and MLP streams passthrough. Questions Why is m_rtTimePerFrame twice the time it should be? Have to divide by 2 on many places? Is it possible/how to pause sound from EVR?

I would need to do this when video is too far behind audio. Have anyone researched DTS-MA and MLP passthrough on Radeon HD cards?

Is it even supported? Is it much work to put together the correct streams, and is there information about how the receiver expects the data? Yes, but you asked jffulcrum a question which was answered in my post. My post was posted twice, and jffulcrum's post was a reply to my post.You assume jffulcrum's problem has the exact same cause, which is not necessarily true. In all three of the problems I've posted over the past year that no one has responded to, I have already played the file in another player before posting, so I don't need help.It means nobody knows a solution or they simply don't care. The more details you supply, the bigger chance somebody can help. For a program to be quality, bugs need to have higher priority than new features.

I appreciate that all the development is by kind volunteer effort, but fixing bugs and reading bug reports needs to have a better system than 'whatever posts happens to be noticed because they are interesting'.Now only if people would also submit proper bug reports. I don't decode DTS, all 5.1 is passed to the receiver intact on the SPDIF by 'SPDIFER' open source filter. Vorbis is decoded to PCM, and PCM is handled by 'ACFILTER' open source filter. I am not sure how dual audio is suppoed to work, either all the filters are supposed to be loaded at startup for all audio, or else changing audio in Haali is supposed to cause different filters to be loaded, but neither is happening.SPDIF is the cause. It is not really a bug in MPC or Haali Media Splitter, it is rather more a limitation of DirectShow. Haali dynamically changes the decoding filter in the filter graph. The benefit of that is that only one decoder will be loaded at a time.

Downside is that changing audio will cause MPC's internal switcher to disappear from the graph. Not really a big deal, unless you use it for volume normalization or channel remapping. MPC's internal splitter simply exposes all audio pins and thus it loads several decoders. The audio switcher connects to all of their output pins and makes sure that only one is active. The audio switcher connects to the audio renderer. If you then switch to a decoder that uses SPDIF, things go wrong. Some of the fixes are implemented in the VMR9 renderer as the EVR inherits from.

I could try to implement all fixes in VMR9 as well. Would it be worth the effort to implement them in any other renderer? Actually all the renderers are usefull to some people. But if you could completely fix VMR9 i think the other ones would become obsolete (a lot of people are forced to use VMR7 due to problems with VMR9) For completeness sake it would be 'nice' to have them all fixed, i suggest focusing on VMR9 and EVR only for the moment though. Actually all the renderers are usefull to some people.

But if you could completely fix VMR9 i think the other ones would become obsolete (a lot of people are forced to use VMR7 due to problems with VMR9) For completeness sake it would be 'nice' to have them all fixed, i suggest focusing on VMR9 and EVR only for the moment though I concur. VMR9 is the default renderer for Windows XP (and VMR is the only renderer where DXVA is guaranteed to work) so it's widely used, I think - not to mention that you have to install.NET 3 and regsvr32 a DLL to use EVR on XP.:) np: Other People's Children - Suicide Common (A Number Of Small Things (Disc 2)). Hello everyone, I finally taken the time to setup MPC-HC for development so I can fix a couple of issues that have been irritating me for a long time. All new people who can fix bugs and improve mpc-hc are welcome:-) Could you send me your patches in PM to have a look on it? [/INDENT]Why is m_rtTimePerFrame twice the time it should be?

Have to divide by 2 on many places? All new people who can fix bugs and improve mpc-hc are welcome:-) Could you send me your patches in PM to have a look on it? I will, after I have implemented everything in VMR9 as well. Do you have a database of sample clips that I could use to make sure that everything is working correctly? This code works if FramesPerSecond are in 'field per second', but it didn't if it's 'frames per second' (constant should be 10000000I64 instead of 20000000I64). Problem: - MPCVideoDec always set 'interleaved' flag in output media type (ffdshow and cyberlink do the same, it's a trick than cannot be changed easily).

EVR automaticaly double the fps in this case, but leave the InterlacedMode to progressive. That the reason some people have noticed a jitter in latest builds. - Microsoft WindowsMedia DMO decoder didn't set interlaced flag, so my code is wrong. So yes there is a bug here, another solution should be found to have fps. Did you patch need the exact fps to works?

Because until now the only purpose was to display jitter curve. The only way i know it to rewrite an audio renderer, and resample sound to follow the video. That's more or less what reclock do. It's mostly for the stats, but it was used in the sync calculation in the EVR code before. Also I used it for setting the subtitle FPS. I guess you could check for the interlaced flag and calculate the frame time correctly?

This is interesting, because reclock doesn't work all that great. I have my display fixed at 60hz, nothing i can do about this, i would like to have the audio resampled to 60hz(or the closest devision) without changing the pitch like reclock does. Reclock works perfectly fine, it's just MPC HC that doesn't work well with it(random frame drops, random freezing on seeks, massive frame drops after 45/60' due to poor presenting code apparently). You can't quite ask for 60Hz w/o changing the pitch. Either you accelerate 23.976fps to 24fps@48Hz/25fps@50Hz, or you live with 24@60 judder:p if Beliyaal wants to improve Reclock support, I'm sure I can find him some dedicated betatesters:D.

Reclock works perfectly fine, it's just MPC HC that doesn't work well with it(random frame drops, random freezing on seeks, massive frame drops after 45/60' due to poor presenting code apparently). You can't quite ask for 60Hz w/o changing the pitch.

Either you accelerate 23.976fps to 24fps@48Hz/25fps@50Hz, or you live with 24@60 judder:p if Beliyaal wants to improve Reclock support, I'm sure I can find him some dedicated betatesters:D In my case it either says that it cannot work and remains red, or it works and its like i'm watching alvin and the chipmunks In my book thats broken:D My problem is: screen = 60HZ Media is 23.97, 24, 25 and 30. So what i need is: Any >30, without pitch resampling. Reclock works perfectly fine, it's just MPC HC that doesn't work well with it(random frame drops, random freezing on seeks, massive frame drops after 45/60' due to poor presenting code apparently). You can't quite ask for 60Hz w/o changing the pitch. Either you accelerate 23.976fps to 24fps@48Hz/25fps@50Hz, or you live with 24@60 judder:p if Beliyaal wants to improve Reclock support, I'm sure I can find him some dedicated betatesters:D I have just tried Reclock and it seems to be working really well with the new sync code.

Even fixes the unstable DTS clock on ATI cards when you pass spdif through it. In my case it either says that it cannot work and remains red, or it works and its like i'm watching alvin and the chipmunks In my book thats broken:D My problem is: screen = 60HZ Media is 23. Broadcom Netxtreme Bcm5782 Gigabit Ethernet Controller Driver Windows Xp. 97, 24, 25 and 30. So what i need is: Any >30, without pitch resampling At least in the latest beta of Reclock it stretches instead of pitiching the sound. Although this isn't ideal either. If you really want to fix it without changing the sound you will just have to resample the video instead.

I have long been thinking about if it would be possible to do this with motion detection. Maybe if you do the motion detection off line and have it as a separate stream it could be feasible with good quality:).

I have just tried Reclock and it seems to be working really well with the new sync code. Even fixes the unstable DTS clock on ATI cards when you pass spdif through it. Well it can't get any worse I'd say:D j/k the biggest problem was random freezing on seeks, it's been reproduced on many different computers in either VMR/EVR or HR.open a bunch of files and seek a lot, it will happen, eventually. Then we got the random drop problem in VMR9/EVR, this is what JohnAd told me about it: EVR is sensitive to the time taken on the longest frames, the default implementation it is not very compatible with software processing on compressed material as you need to keep well under 40ms per frame peak, this can and should be handled outside the renderer. And last issue, reproduced by many ppl as well: run a 23.976 movie at 47.952/48.000Hz in custom EVR/VMR9, enable the CTRL+J infos.wait 45/60' and it will drop frames like crazy. Apparently it's due to the same problem JohnAd told me about.this doesn't happen with HR;).

Well it can't get any worse I'd say:D j/k the biggest problem was random freezing on seeks, it's been reproduced on many different computers in either VMR/EVR or HR.open a bunch of files and seek a lot, it will happen, eventually. Will check if I can reproduce. Then we got the random drop problem in VMR9/EVR, this is what JohnAd told me about it: and last issue, reproduced by many ppl as well: run a 23.976 movie at 47.952/48.000Hz in custom EVR/VMR9, enable the CTRL+J infos.wait 45/60' and it will drop frames like crazy.

Apparently it's due to the same problem JohnAd told me about.this doesn't happen with HR;) These problems should be fixed. Processing has been moved to a new thread, and the old code wasn't even able to drop frames, it just said it did but didn't.

Not sure what you mean by wait 45/60? Will check if I can reproduce.

These problems should be fixed. Processing has been moved to a new thread, and the old code wasn't even able to drop frames, it just said it did but didn't. Not sure what you mean by wait 45/60?

45/60 mins it will start nicely at 24.000 in the CTRL+J infos(with 48.000Hz in Reclock), then it will drop frames from time to time(eg. You get a small wave in the jitter infos).then after 45/60 minutes it will get worse and worse and drop frames constantly, requiring a reseek:eek: ideally it shouldn't drop frames AT ALL.HR doesn't:) Apparently using MPC HC in exclusive D3D mode would fix this problem, but this is not convenient for everyday use.plus if you enable the CTRL+J infos you still get big waves from time to time and it's not dead on 24.000 I dearly hope that you'll be able to reproduce the random freezing problem on seeks, as it's really the biggest issue right now with using Reclock in MPC HC:o:thanks. Hello everyone, I finally taken the time to setup MPC-HC for development so I can fix a couple of issues that have been irritating me for a long time. That's some awesome work - what a great contribution:-) I too would like to add a little - just a few extra commands to the MpcApi interface, but I'm having trouble building the code. I presume I need to revert to the June 2007 version of DirectX, because later ones don't have d3d.h, but now I'm stuck on libavcodec_gcc.lib - where did you get it, or how is it built?

And if it's built with gcc, why can't it be built with VS2008? It is not really a bug in MPC or Haali Media Splitter, it is rather more a limitation of DirectShow. Haali dynamically changes the decoding filter in the filter graph.

The benefit of that is that only one decoder will be loaded at a time. Downside is that changing audio will cause MPC's internal switcher to disappear from the graph. Not really a big deal, unless you use it for volume normalization or channel remapping. MPC's internal splitter simply exposes all audio pins and thus it loads several decoders. The audio switcher connects to all of their output pins and makes sure that only one is active. The audio switcher connects to the audio renderer.

If you then switch to a decoder that uses SPDIF, things go wrong. Halli Splitter does not change the filter that is loaded when I switch from default Vorbis audio track to the DTS audio track. This is the same whether Internal Audio Switcher is enabled or not. Is there any other MKV splitter that can handle two audio streams? PS I previously had a problem with Haali Splitter, posted a thread in Doom9 and zero replies. Hello everyone, I finally taken the time to setup MPC-HC for development so I can fix a couple of issues that have been irritating me for a long time. [INDENT]The sync for the custom EVR was practically nonexistent.

Aside from being unable to drop frames when the movie fps is about equal to the screen refresh it also uses a non-robust method for syncing. I have now fixed the sync to be as perfect as is possible.

For example when you run a 23.976 movie in 24 fps instead of stuttering several frames as the Haali renderer does it only stutters once to fix the sync. Some of the fixes are implemented in the VMR9 renderer as the EVR inherits from. I could try to implement all fixes in VMR9 as well.

Would it be worth the effort to implement them in any other renderer? Would it be possible to fix the Haali renderer, too? Or is the stuttering code inside of the Haali renderer, so that you can't fix it? Generally I like the idea of the Haali renderer. It simply bypasses all the ugly graphics manufacturers' ideas of how video could be 'optimized' by not using the usual video pipeline. It's also the only way I can get proper BTB and WTW with my ATI card.

So if you could optimize the Haali Renderer, as well, as that would be much appreciated! * Add support for automatically changing the number of audio output channels based on content. This is because my receiver requires the correct number of channels to use the correct expansion modes to 7.1. Do you mean for LPCM transport over HDMI? And your change would result in that the receiver correctly detects e.g.

2.0 for stereo content, 5.1 for 5.1 content, 6.1 for 6.1 content and 7.1 for 7.1 content? That would be VERY nice, if possible! Is it possible/how to pause sound from EVR? I would need to do this when video is too far behind audio. Too far behind because the decoder isn't fast enough? Or too far behind due to audio and video clocks differing? The first case is very bad and I don't see any good way to solve it.

In the 2nd case can't you repeat video frames to match video/audio back together? I'm aware that this will likely cause visible stuttering. But that's the same we get if video comes in too fast, compared to audio, right? I think I'd prefer that over manipulating the audio data. Another option would be to write an Audio Renderer (as Casimir has suggested). But I'm thinking: It should also be possible to write an intermediate filter which accepts PCM, DTS, AC3 etc and outputs them again.

Such an intermediate filter would that manipulate the audio data as needed (while communicating with the video renderer). Should work just fine, too.

But manipulating audio data is not pretty. If you do bitstream, you can only repeat/drop full audio frames, which can be very clearly audible (especially during dialog and music). With PCM it's a bit better, but still it's not easy.

You can resample (which changes the pitch) or you can try to stretch somehow without changing the pitch (which can introduce artifacts on its own). But I really don't like the general idea of tampering around with audio, after we spent so much money and space on getting lossless audio from Blu-Ray.

One tiny video stutter every few minutes would be preferable in my opinion. And it might not even be noticed at all, if it happens to happen in a scene where there's not much motion. Anyway, this is just my opinion.

I have a DVR I use to record from TV, and the media I use to record makes a diference to MPC-HC. If I burn something to a DVD-R, I can seek the video, but if I burn the same thing to a DVD+R I can't seek it. This happens in MPC-HC (1.2.908.0 and the latest 948) and MPC. There is though a difference between the DVD-R and the DVD+R. The DVD-R has a VIDEO_TS folder, but the DVD+R has a VIDEO_TS and a VIDEO_RM folders. Could this be the cause?

Here are samples of the same clip recorded to DVD-R and DVD+R. Hope this helps finding the problem.

PS: I can seek the DVD+R in PowerDVD and VLC, but I prefer using MPC-HC/MPC;). Would it be possible to fix the Haali renderer, too? Or is the stuttering code inside of the Haali renderer, so that you can't fix it? What is wrong with HR? I use it all the time with Reclock(and so do many friends of mine), it never drops frames and never stutters.

Another option would be to write an Audio Renderer (as Casimir has suggested). A new Reclock basically? If that's because you guys are losing A/V sync after a while, doesn't Reclock in 'original speed' fix the problem?

You can't play 24 fps in 60Hz smoothly, but Reclock can still sync the A/V frames AFAIK. What is wrong with HR? I use it all the time with Reclock(and so do many friends of mine), it never drops frames and never stutters. With ReClock maybe.

Beliyaal reported problems when graphics card output refresh rate doesn't 100% match source frame rate. Probably without ReClock. A new Reclock basically?

If that's because you guys are losing A/V sync after a while, doesn't Reclock in 'original speed' fix the problem? You can't play 24 fps in 60Hz smoothly, but Reclock can still sync the A/V frames AFAIK.

I'm not a fan of ReClock because it *constantly* manipulates the audio data. I don't consider that a good idea. Plus, ReClock doesn't work very well with bitstream audio (audible artifacts). I'd prefer a solution which keeps its hands of video/audio data as long as possible, and only does something to bring sync back in line when it's really necessary (e.g. Sync mismatch >50ms).

And I prefer manipulating video playback instead of audio. My way of thinking comes from that I really try to match audio and video clocks as good as possible. So the clock difference is really minor in my case. As a result the constant audio data manipulation of ReClock is just not needed. Maybe I need a dropped/repeated video frame once or twice in a movie.

I'd much prefer that over constant resampling/stretching of the audio data throughout the whole movie. With ReClock maybe. Beliyaal reported problems when graphics card output refresh rate doesn't 100% match source frame rate. Probably without ReClock. I'm not a fan of ReClock because it *constantly* manipulates the audio data. I don't consider that a good idea. Plus, ReClock doesn't work very well with bitstream audio (audible artifacts).

I'd prefer a solution which keeps its hands of video/audio data as long as possible, and only does something to bring sync back in line when it's really necessary (e.g. Sync mismatch >50ms).

And I prefer manipulating video playback instead of audio. My way of thinking comes from that I really try to match audio and video clocks as good as possible. So the clock difference is really minor in my case. As a result the constant audio data manipulation of ReClock is just not needed.

Maybe I need a dropped/repeated video frame once or twice in a movie. I'd much prefer that over constant resampling/stretching of the audio data throughout the whole movie.

Getting smooth playback w/o resampling and w/o changing the playback speed on a Windows PC would require a miracle I think:D you can still output 47.952Hz and use Reclock in pass-through S/PDIF mode.or better get a RME soundcard and change the masterclock speed in its drivers(no resampling!) quite frankly yesgrey3 has been constantly updating the Reclock resampler, and the latest version in 'excellent' mode is both very CPU efficient & very good sounding:) I personally don't find dropping/duping frames acceptable, and I'm not the only one;) Reclock+HR in 48Hz is where HTPC's shine! No dropped frame WHATSOEVER & perfectly butter smooth playback:cool. Now I'm stuck on libavcodec_gcc.lib - where did you get it, or how is it built? And if it's built with gcc, why can't it be built with VS2008?

I gave up trying to get MinGW functional. I just changed the project to use VS2008 in release as well as debug. The reason GCC is used is because asm optimization only works there, but because I don't use the internal decoders it doesn't matter for me. Also it's useful to have a slow decoder so you can test when the render is starved:) Would it be possible to fix the Haali renderer, too? Or is the stuttering code inside of the Haali renderer, so that you can't fix it?

Generally I like the idea of the Haali renderer. It simply bypasses all the ugly graphics manufacturers' ideas of how video could be 'optimized' by not using the usual video pipeline. It's also the only way I can get proper BTB and WTW with my ATI card. So if you could optimize the Haali Renderer, as well, as that would be much appreciated! I'm not sure, but i think the source for Haali render is closed, so I wouldn't be able to fix it? I'm not sure what BTB and WTW means, but I might be able to fix it in the VMR9 and EVR renderes. Do you mean for LPCM transport over HDMI?

And your change would result in that the receiver correctly detects e.g. 2.0 for stereo content, 5.1 for 5.1 content, 6.1 for 6.1 content and 7.1 for 7.1 content? That would be VERY nice, if possible! Yes exactly, I guess that the best bet here is creating a new audio render with WASAPI exclusive mode in Vista, if this doesn't work I would simply change the output mode configured in Windows. Too far behind because the decoder isn't fast enough?

Or too far behind due to audio and video clocks differing? The first case is very bad and I don't see any good way to solve it. In the 2nd case can't you repeat video frames to match video/audio back together? I'm aware that this will likely cause visible stuttering. But that's the same we get if video comes in too fast, compared to audio, right? I think I'd prefer that over manipulating the audio data.

The case where it was a problem was when the decoder wasn't fast enough. I managed to fix this yesterday after much head scratching with EC_STARVING and overriding the GetState method of the OuterEVR (took a while to figure out this was needed). My way of thinking comes from that I really try to match audio and video clocks as good as possible. So the clock difference is really minor in my case. As a result the constant audio data manipulation of ReClock is just not needed. Maybe I need a dropped/repeated video frame once or twice in a movie.

I'd much prefer that over constant resampling/stretching of the audio data throughout the whole movie. You will be happy to know I feel exactly the same way and will work toward making it as good as possible.

Getting smooth playback w/o resampling and w/o changing the playback speed on a Windows PC would require a miracle I think:D you can still output 47.952Hz and use Reclock in pass-through S/PDIF mode.or better get a RME soundcard and change the masterclock speed in its drivers(no resampling!) quite frankly yesgrey3 has been constantly updating the Reclock resampler, and the latest version in 'excellent' mode is both very CPU efficient & very good sounding:) I personally don't find dropping/duping frames acceptable, and I'm not the only one;) Reclock+HR in 48Hz is where HTPC's shine! No dropped frame WHATSOEVER & perfectly butter smooth playback:cool: Even in 'Original speed' mode when rendering spdif data it will drop and add audio frames and slightly change the speed of the audio.

If it didn't it would be perfect, unfortunately it's not open source so I can't go in and fix the issues or add WASAPI exclusive mode. It also fails to reliably detect my refresh rate.

The only feature that I need from it is it's ability to fix the broken ATI DTS passthrough clock and possibly 25->24 fps movie conversion. Reclock seems to cause many other issues so it might be good to have a new audio render that only does the least amount of damage to the sound. Some more updates: * Managed to fix pausing of the graph with EC_STARVING as mentioned above.

* VMR9 renderer doesn't handle sync itself so it's more work to add the other fixes than I originally though. It should be possible however as the Haali renderer seems to handle it itself. Will look at this later. * Fixed rtTimePerFrame detection by using the VMR9 method as the Media Foundation api seems to be broken and it doesn't correctly report the interlace flags. * There is still some issues that I have found that I want to fix before submitting code. * Found the Reclock deadlock and implemented a workaround. Getting smooth playback w/o resampling and w/o changing the playback speed on a Windows PC would require a miracle I think:D I'm not sure about that.

We have HDMI output now which combines video + audio. I'd imagine that the audio and video clock are now related somewhat, because the audio is sent over HDMI in between the pictures. I don't know for sure how the audio clock is created by those HDMI enabled audio devices, though. Quite frankly yesgrey3 has been constantly updating the Reclock resampler, and the latest version in 'excellent' mode is both very CPU efficient & very good sounding:) Resampling audio is similar to not having 1:1 pixel mapping, regardless of how good a resampler you're using. I personally don't find dropping/duping frames acceptable, and I'm not the only one;) I don't like it, either, but if there's only one or two frame drops/duplicates per movie then I like that a lot better than having all audio manipulated *all the time*. Reclock+HR in 48Hz is where HTPC's shine! No dropped frame WHATSOEVER & perfectly butter smooth playback:cool: You're drooling over how good video is then.

But your audio is being resampled all the time. Maybe you're not bothered by that, but I am. And I'm another. Reclock sounds BETTER than the other audio renderers on my system due to Kernel streaming (well, it did on XP, but I'm not so sure about Vista). Now that you mention it, I'd *love* to have an audio renderer which works like any other audio renderer, but uses kernel streaming. Something like ReClock, but without all the audio manipulation stuff.

I'm not sure, but the source for Haali render is closed, so I wouldn't be able to fix it? I'm not sure what BTB and WTW means, but I might be able to fix it in the VMR9 and EVR renderes. If the problem is inside the Haali Renderer then you can't fix it. I thought maybe it's in the MPC code calling the Haali Renderer. Well, seemingly not. BTB and WTW means 'blacker than black' and 'whiter than white'. With RGB video there are two different 'levels', either 'PC levels' or 'video levels'.

When using PC levels, black is RGB '0, 0, 0' and white is '255, 255, 255'. With video levels, black is '16, 16, 16' and white is '235, 235, 235', IIRC.

The values below 16 are BTB. Now ATI is doing some ugly stuff in their drivers. With VMR9 and EVR regardless of what I do, all values below '16, 16, 16' are clipped by the ATI driver. Which is already bad in itself. But it's also bad because it means that the ATI driver internally stretches the YCbCr data coming from the decoder to PC levels and then later converts PC levels to video levels. Doing that can introduce banding problems. The Haali Renderer bypasses all this ATI driver sh*t by pretending to not display video but by displaying games (I think).

So ATI isn't tempted to do violence to the 'video' data, anymore. The nice thing about the Haali Renderer is that is bypasses all these ugly video manipulations (face color 'improvements', non-defeatable noise 'reduction' and detail 'enhancement', color 'correction', etc etc etc) which otherwise are hard or even impossible to turn off. That isn't limited to ATI, NVidia and Intel also have similar problems, but not as bad as ATI, I think. Yes exactly, I guess that the best bet here is creating a new audio render with WASAPI exclusive mode in Vista, if this doesn't work it would simply change the output mode configured in Windows. That would be VERY nice. I had already thought about this problem, too, and wondered about how to fix it.

I was thinking about changing output modes in the OS settings programmatically. But if you can do it inside of the audio renderer, that would be of course much better. BTW, how much time do you have for working on MPC? I'm asking because writing an audio renderer is probably quite complex, from what I've seen in the MSDN documentation, because the audio renderer is the one responsible for the clock of the whole DirectShow graph. Of course writing an audio renderer gives you the most control you can get, so it is tempting.:) You will be happy to know I feel exactly the same way and will work toward making it as good as possible.

I'm also happy to see that you seem to be working on exactly the 'right things', from my point of view, to make MPC HC the ultimate home cinema playback solution.:). If the problem is inside the Haali Renderer then you can't fix it. I thought maybe it's in the MPC code calling the Haali Renderer. Well, seemingly not.

BTB and WTW means 'blacker than black' and 'whiter than white'. With RGB video there are two different 'levels', either 'PC levels' or 'video levels'.

When using PC levels, black is RGB '0, 0, 0' and white is '255, 255, 255'. With video levels, black is '16, 16, 16' and white is '235, 235, 235', IIRC. The values below 16 are BTB. Now ATI is doing some ugly stuff in their drivers.

With VMR9 and EVR regardless of what I do, all values below '16, 16, 16' are clipped by the ATI driver. Which is already bad in itself. But it's also bad because it means that the ATI driver internally stretches the YCbCr data coming from the decoder to PC levels and then later converts PC levels to video levels. Doing that can introduce banding problems. The Haali Renderer bypasses all this ATI driver sh*t by pretending to not display video but by displaying games (I think). So ATI isn't tempted to do violence to the 'video' data, anymore. The nice thing about the Haali Renderer is that is bypasses all these ugly video manipulations (face color 'improvements', non-defeatable noise 'reduction' and detail 'enhancement', color 'correction', etc etc etc) which otherwise are hard or even impossible to turn off.

That isn't limited to ATI, NVidia and Intel also have similar problems, but not as bad as ATI, I think. This must be the D3DPRESENTFLAG_VIDEO flag. We should be able to add an option to turn this flag off, this should have the same effect as the Haali renderer. BTW, how much time do you have for working on MPC? I'm asking because writing an audio renderer is probably quite complex, from what I've seen in the MSDN documentation, because the audio renderer is the one responsible for the clock of the whole DirectShow graph. Of course writing an audio renderer gives you the most control you can get, so it is tempting.:) Not sure, but I guess that I could spend at least a weekend trying. I have written a whole software sound mixer/engine with 3D and reverb etc support for a game engine (PC, PS3 and Xbox 360) before, so I think I can manage.:).

You could download MinGW GCC from here: or here: I got that far, but couldn't get it to build successfully. Something about make toolchain I think.

When I have time later I will try to figure it out, and also I'm thinking about integrating the gcc build process from Visual Studio. A step by step guide of how to setup the build environment with the various libraries etc for building would be great for reducing the barrier of entry for contributing to MCP-HC! This must be the D3DPRESENTFLAG_VIDEO flag. We should be able to add an option to turn this flag off, this should have the same effect as the Haali renderer. To be honest, I've never looked in detail at how renderers work, so I can't really say much about it. But I guess if you use EVR or VMR, the graphics driver will know that it's video regardless of whether the D3DPRESENTFLAG_VIDEO flag is set or not. So I'd guess that all the bad driver manipulations probably depend on whether EVR/VMR is used and not on the D3DPRESENTFLAG_VIDEO flag.

Haali uses neither EVR nor VMR, so the graphics driver thinks a game is being played. Not sure, but I guess that I could spend at least a weekend trying. I have written a whole software sound mixer/engine with 3D and reverb etc support for a game engine (PC, PS3 and Xbox 360) before, so I think I can manage.:) Cool. Well, we will take whatever you can give us, of course.:). I too would like to add a little - just a few extra commands to the MpcApi interface, but I'm having trouble building the code. I presume I need to revert to the June 2007 version of DirectX, because later ones don't have d3d.h, but now I'm stuck on libavcodec_gcc.lib - where did you get it, or how is it built?

And if it's built with gcc, why can't it be built with VS2008?VS2008 does not support the inline assembly that libavcodec contains. You need to build it with MinGW GCC to get decent decoding performance. You can find an easy to use installer for MinGW in the first post of the large ffdshow topic. To be honest, I've never looked in detail at how renderers work, so I can't really say much about it. But I guess if you use EVR or VMR, the graphics driver will know that it's video regardless of whether the D3DPRESENTFLAG_VIDEO flag is set or not. So I'd guess that all the bad driver manipulations probably depend on whether EVR/VMR is used and not on the D3DPRESENTFLAG_VIDEO flag. Haali uses neither EVR nor VMR, so the graphics driver thinks a game is being played.

I have investigated this now and in ATI drivers >8.10 the video controls works ONLY when rendering with DXVA. I might be wrong but Haali renderer does not support DXVA? In that case you shouldn't have any problems if you use newest ATI drivers and disable the DXVA codecs.

Also as you said the D3DPRESENTFLAG_VIDEO doesn't have any effect. It might in older ATI drivers though that works through the presented surface instead of in the DXVA decoder.

I have investigated this now and in ATI drivers >8.10 the video controls works ONLY when rendering with DXVA. I might be wrong but Haali renderer does not support DXVA? In that case you shouldn't have any problems if you use newest ATI drivers and disable the DXVA codecs. Also as you said the D3DPRESENTFLAG_VIDEO doesn't have any effect. It might in older ATI drivers though that works through the presented surface instead of in the DXVA decoder.

Does adding the presentflag to all renderers cause consistent behaviour regardless of DXVA? (Currently all renderers behave differently). I have investigated this now and in ATI drivers >8.10 the video controls works ONLY when rendering with DXVA. I might be wrong but Haali renderer does not support DXVA? In that case you shouldn't have any problems if you use newest ATI drivers and disable the DXVA codecs. Also as you said the D3DPRESENTFLAG_VIDEO doesn't have any effect.

It might in older ATI drivers though that works through the presented surface instead of in the DXVA decoder. In my experience the video controls only work when you feed the renderer with the NV12 colorspace. Of course it works with DXVA, because DXVA needs NV12 to work. I don't know about Haali's renderer, but with EVR is like that. That's also the reason why Microsoft's decoders output NV12.