madVR - high quality video renderer (GPU assisted)

[QBhd]

I've discovered that my TV (Sony 55W905) can only display 4:4:4 chroma in Game or Graphics modes but only properly displays 24fps input without judder in Cinema mode. So I need to make a choice between the two unfortunately.

Am I correct in thinking that if I sacrifice 4:4:4 chroma there's no point in using NNEDI3 for chroma upscaling with madVR as it'll just be blurred by the TV anyway?

I have a 4:2:2 Plasma and I notice enough of a difference with NNEDI3 on chroma upscaling to keep using it.

Wait, so the effects of madVR are only active with certain hardware acceleration options like DXVA2?

I have never used DXVA or any other HW accel and all options working fine

QB

[Nui]

I have a ridiculous problem and one of you might be able to help me.
Basically I want to use the APL pattern using madTPG. But, my television, panasonics sloppiest plasma ever made, prevents me from using it. Reason is, that the lines containing the measurement field MUST be black or white. Otherwise they will have an IRE depended colorshift.

Is there a way I can black out the sides of the pattern? An extra application for example?

[Asmodian]

Have you played with the "Image area:" and "Background:" sliders? The Background slider allows you to change the color of the background between black, shades of gray, and white. The Image area slider changes the size of the active test pattern. If you double click in the image madVR Test Pattern Generator will go full screen.

[Nui]

I know what they do. I need the "background" to be confined to the lines above and below the image area, not on the same lines. Special requirement, because my display is a bit stupid ;)

[StinDaWg]

That thread isn't about a bug with AR for downscaling causing aliased text with NNEDI3.

That is an issue caused by using NNEDI3 doubling for luma but not for chroma which causes chroma misalignment artifacts which are then exaggerated by AR and/or LL downscaling with the sharper downscaling options. Use chroma doubling too and they go away completely.

Catmul-Rom+AR+LL is good because it has the lest artifacts and you get the benefits of LL downscaling.

Edit: Chroma misalignment probably isn't the right way to describe it, if the luma is sharper than the chroma that means the chroma has been blurred more so it bleeds further into the surrounding pixels but it is still perfectly aligned with the luma.
Chroma doubling is both a GPU killer and a waste quality wise, so until/if this "bug" is fixed, I'm not using AR on downscaling.

I don't prefer Catmul-Rom for downscaling because it's softer, and it's also aliased if you use AR without chroma doubling just like Lanczos. To each their own.

[Asmodian]

Chroma doubling is both a GPU killer and a waste quality wise, so until/if this "bug" is fixed, I'm not using AR on downscaling.

I don't prefer Catmul-Rom for downscaling because it's softer, and it's also aliased if you use AR without chroma doubling just like Lanczos. To each their own.

Absolutely; it is perfectly reasonable to turn off AR since you don't see the artifacts from using NNEDI3 doubling only on the luma even with a very sharp down scaling algorithm. It isn't really a bug though, or at least it probably isn't. It is an artifact of the way all the math interacts but everything is working "as designed" if not as planned.

I was objecting to the identification of where the problem was, not the workaround for it. ;)

[Supun92]

Guys. I tried to setup madVR for potplayer in my laptop. But madVR works properly when a select power-saver mode from AMD Switchable Graphics(running on Intel graphics), but when i select Performance mode from AMD Switchable Graphics (running on AMD Radeon) potplayer just shows black screen & audio works.

My System specs:
Model- HP ProBook 450 G1
CPU- Intel(R) Core(TM) i5-4200M CPU @ 2.50GHz
Ram-4GB DDR3
GPU-2GB AMD Radeon HD8750M
HDD-750GB

Does anyone of you know what is wrong with the system& is there any way to solve this?

[Asmodian]

That sounds like the same problem newer Optimus (Nvidia's version of switchable graphics) users have. I don't believe anyone has found a solution yet, sorry. :(

[generalmx]

Long overdue follow-up on http://forum.doom9.org/showthread.php?p=1686141#post1686141 --- while I can't say for certain, I'm pretty sure the original problem of pan&scan problems was due to use of ReClock and the fact I was doing anime, where it can be much more easily spotted. I've mostly eliminated my problems thanks to some configuration and hardware changes I've since made:

- Using Toasty's CRU, I successfully added 1920x1200@48Hz to my main monitor (it's a workstation monitor, so no EDID 24Hz or 30Hz), and set up rules in madVR to switch to this as necessary.
- ReClock is set to default to 24Hz and "Auto".
- I setup <480p, 480p, 720p, and 1080p rules in madVR, which helped significantly.
- I further setup 720p@24 and 720p@30 rules, as it seems doing 720p@30 frames at 60Hz can be just a bit too taxing for my single watercooled R9 290X (NNEID w/ Luma Doubling upscaled to 1200p).

I'd also like to confirm as I'm sure others have in the past that madVR is brilliant with the few of us using 4K HDTVs. As my previous cheapo 32" 1080p TN HDTV had serious trouble doing both 24Hz and 30Hz on a PC (some sort of timing issue I couldn't be bothered with), and as I can certainly use the extra desktop space for programming, I decided to splurge a little on the 39" 2160p MVA Seiki, which can also do 1080p@120Hz (using CRU), as well as up to 4K@30Hz. Changing my madVR rules only a little, I now have madVR doing most of the work on scaling up content to 4K@30Hz and 4K@24Hz (NNEID, of course), which looks a hell of a lot better on mixed <1080p content then the display's built-in scaler.

The rules one can setup in madVR aren't bad at all. I even wonder if the next step might be to open it up a bit more such that madVR could call an external program upon exit that looks at the last file's recorded final stats (same as CTRL+J) and tries to intelligently make up rules based on that and previous user configuration. A start would be to let madVR record these stats per-file in a simple text file, then have someone make a parser that suggests rules. I hesitate to suggest this quality-balancing feature to be added outright to madVR, since I bet it would be an absolute hell to support, partly as it's ultimately up to user preference.

Oh, and I'd like to query which is better to code for in OpenCL and for video decoding --- Nvidia or AMD? Since I've been hearing that while Nvidia pushes CUDA, OpenCL with AMD is actually a bigger PITA to code for (I think already mentioned in this thread, but I can't find it). I'm actually considering making my next upgrade either getting a different CPU with an Intel GPU to try QuickSync, and/or an Nvidia GTX 750 Ti secondary; I think I really liked CUVID support for running @ 60 FPS, which eliminated the motion problems pretty much completely.

And as always, a huge thank you to madShi :)

[StinDaWg]

Absolutely; it is perfectly reasonable to turn off AR since you don't see the artifacts from using NNEDI3 doubling only on the luma even with a very sharp down scaling algorithm. It isn't really a bug though, or at least it probably isn't. It is an artifact of the way all the math interacts but everything is working "as designed" if not as planned.

I was objecting to the identification of where the problem was, not the workaround for it. ;)
Ya, I just don't know why most people still recommend using AR for downscaling if they use something other than NNEDI3 doubling for chroma. The jaggedness it causes on diagonal lines (text, signs, cartoons, ect) is so obvious, it kind of defeats the purpose of using NNEDI3 in the first place.

Lanczos 3 with the anti-ringing filter:
http://abload.de/img/l3-arw9euh.png

Lanczos 3 without the anti-ringing filter:
http://abload.de/img/l3v4f05.png

[Asmodian]

We aren't "recommending AR for down-scaling", we are recommending catmull-rom + AR + LL for down scaling. That particular test image will probably look better (i.e. sharper) with something else but if you want LL down-scaling with minimal artifacts catmull-rom is the preferred option. I don't recommend Lanczos3+AR for down-scaling, I don't think anyone has have they?

If sharpness > all then Lanczos 3 or 4 without AR is probably a good choice. However, I want my night skies fully populated. ;)

[StinDaWg]

We aren't "recommending AR for down-scaling", we are recommending catmull-rom + AR + LL for down scaling. That particular test image will probably look better (i.e. sharper) with something else but if you want LL down-scaling with minimal artifacts catmull-rom is the preferred option. I don't recommend Lanczos3+AR for down-scaling, I don't think anyone has have they?

If sharpness > all then Lanczos 3 or 4 without AR is probably a good choice. However, I want my night skies fully populated. ;)
Catmull-Rom + AR + LL for down scaling causes the same jagged text. It is no better than Lanczos in this scenario. I don't know what night skies has to do with anything. I see the same detail retention in real world content between CR and Lanczos, just Lanczos being a bit sharper overall.

[Asmodian]

I see, I would be wiling to compromise on Catmull-Rom + LL without AR for down-scaling after NNEDI3 doubling the luma. I find ringing harder to notice when down-scaling while aliasing is more obvious.

I like LL because it keeps bright details on a dark background much better, try down-scaling a night sky with and without LL. Lanczos with LL rings too much even for me and catmull-rom is sharp enough for down-scaling. To each their own though, there is a reason madVR has options. :)

[6233638]

Catmull-Rom + AR + LL for down scaling causes the same jagged text. It is no better than Lanczos in this scenario. I don't know what night skies has to do with anything. I see the same detail retention in real world content between CR and Lanczos, just Lanczos being a bit sharper overall.No, it does not cause jagged text.

Using NNEDI3 Image Doubling without enabling Chroma Doubling, or using a different number of taps for Image/Chroma doubling results in this aliasing.

It is the quality of the downscaling with CR+LL+AR which allows this to be more easily seen. It is not caused by CR+LL+AR.


If you are using anything else for downscaling, you should disable the linear light option. It just looks bad.

[ajp2k11]

@MadVR or anybody else for that matter

I've had problems with MadVR for a long time now where I would only get a black screen with the audio playing. Sometime, only on rare occasions, the file starts playing after up to 30 seconds. It seems a bit random but apparently MadVR can't determine the refresh rate of my laptop screen because it says 0Hz.

Anyway, today I discovered that if I play a file and then move MPC-HC's window randomly over the screen (50% zoom works best) the video will start to play after a few seconds and it says refresh rate 120Hz just like it should. I've tested this many times now and each time it works...

Just wanted to share my findings.... weird huh?

(Win 8.1, AMD 6650M/Catalyst 13.12 I think from Windows update)

EDIT: Of course, when I rebooted it doesn't seem to work anymore... even weirder. :-S

[StinDaWg]

No, it does not cause jagged text.

Using NNEDI3 Image Doubling without enabling Chroma Doubling, or using a different number of taps for Image/Chroma doubling results in this aliasing.

It is the quality of the downscaling with CR+LL+AR which allows this to be more easily seen. It is not caused by CR+LL+AR.


If you are using anything else for downscaling, you should disable the linear light option. It just looks bad.
I think we've already established that. My point is you can't use AR on any of the downscaling options if you don't also use chroma doubling. It will cause jagged text on anything from CR all the way up to Lanczos. I've tried them all, and it is noticeable on all. If that can't be "fixed" because of the way doubling works, then I would recommend for everyone to turn off AR on downscaling unless you want jagged lines. Chroma doubling is too resource intense for me to turn on, so it's not an option. Truth be told, I don't see any more ringing with 720p content with it on or off, but I can see the jagged text easily. It's the lesser of two evils.

Good news though, it looks like this issue is going to be fixed in the next build...

http://forum.doom9.org/showpost.php?p=1692399&postcount=21

FYI, this problem should be fixed in the next build. The problem occurred because the latest madVR build is trying to correct the NNEDI3 0.5 pixel shift, without adjusting the AR filter accordingly. It will be some time before the next build is released, though, since I'm quite busy with other stuff atm...

[Asmodian]

Good news! It really is a bug after all too. ;)

[dansrfe]

@madshi,

Although I understand that the underlying "black/frozen screen" problem with integrated/dedicated gpu machines and madVR may be the fault of an external library, is it possible to at least pinpoint where the problem is rather than ignore it altogether?

Multiple people with 7/8xx GPUs have cited a problem with getting madVR/MPC-HC to render out with the NVidia GPU selected and no one really knows where the problem is occurring. DirectX is backwards compatible with DirectX 9.0 and Windows 8.1 comes with DirectX 11.2 so attempting to install an older version of this (which can't be done anyway) doesn't seem to be the right solution.

Is there any information that I and others experiencing this problem can provide that would aid in pinpointing the problem and closing in on a potential solution? Will remote access to a machine with this problem help?

[huhn]

Multiple people with 7/8xx GPUs have cited a problem with getting madVR/MPC-HC to render out with the NVidia GPU selected and no one really knows where the problem is occurring. DirectX is backwards compatible with DirectX 9.0 and Windows 8.1 comes with DirectX 11.2 so attempting to install an older version of this (which can't be done anyway) doesn't seem to be the right solution.

you have to install dx 9 for dx 9 games to. you need the libs anyway and dx10-11 doesn't have them. because the GPU is backwards compatible doesn't mean you doesn't need the libs...

and i highly recommend you to ask nvidia

[DragonQ]

Direct 9.0c hasn't been updated in like 4 years, I'm amazed Microsoft don't just include the entire runtime with Windows by now (like they do with .NET).

[dansrfe]

you have to install dx 9 for dx 9 games to. you need the libs anyway and dx10-11 doesn't have them. because the GPU is backwards compatible doesn't mean you doesn't need the libs...

and i highly recommend you to ask nvidia

I installed the DX9 runtime as well from the web installed. It downloaded all the updates since 2005 and copied all the DX9 files yet there's still a black screen.

[leeperry]

^ Captain obvious to the rescue but it works fine for pretty much everyone else, might consider reinstalling the OS and/or try different graphic drivers versions after a clean uninstall.

[flashmozzg]

Do you have any iGPU's? try disabling them.

[dansrfe]

^ Captain obvious to the rescue but it works fine for pretty much everyone else, might consider reinstalling the OS and/or try different graphic drivers versions after a clean uninstall.

It works fine indeed for desktops that have the iGPU disabled or laptops which don't have a dedicated late nvidia graphics chip.

As obvious as it should be, bottom line is that it's not working and many people with this particular setup are having this issue but may have given up trying to make it work altogether due to posts like yours.

[dansrfe]

Do you have any iGPU's? try disabling them.

Unfortunately laptop bioses don't give an option to disable the iGPU.

[flashmozzg]

Unfortunately laptop bioses don't give an option to disable the iGPU.

Can't you just disable it in Device Manager? I did.

[Anime Viewer]

^ Captain obvious to the rescue but it works fine for pretty much everyone else, might consider reinstalling the OS and/or try different graphic drivers versions after a clean uninstall.
There is nothing to indicate either the OS or the gpu drivers. People who've had the black screens in games and MPC have tried the last resort clean OS installs, and it doesn't work. Its not a case of a few users having the black screen, so far reports are that every 800 series user who tries to use their Optimus Nvida gpu (and possibly new ATI switchable GPU systems (http://forum.doom9.org/showthread.php?p=1692218#post1692218) have the issue. 800 users have tried driver versions old and new, and (unlike like the Nvidia OpenCL bug) currently none of the versions tested have resolved the issue. The 800 users are using the same (ex: 340.52 default Nviida) drivers as 600 (and other series) users. If the drivers were the problem people with models other then the 800 would have an issue just like what happened with the Nvidia OpenCL bug.


Do you have any iGPU's? try disabling them.
Can't you just disable it in Device Manager? I did.

I seriously doubt you disabled the iGPU on a optimus equipped system. That's not how Optimus technology works - the iGPU in Optimus systems is always on. Even when the system uses the Nvida gpu to process programs the pictures are still transmitted to the screen(s) by the Intel gpu. If you were to disable the Intel gpu in device manager it would just default to a limited generic driver much like how a nonfuctioning, corrupt, safe mode boot, or driver uninstall results in a generic (limited resolution) gpu driver being loaded still running through the gpu.
Rumors are that it may have something to do with how shapes (like rectangles) are processes for antialasing in the chip coding. If that's the case it seems likely that if a solution is going to come about its probably going to require someone to develop a workaround similar to how one was created with coding to get around the Nvidia OpenCL driver bug that Nvidia never fixed. (http://forum.doom9.org/showthread.php?p=1675628#post1675628)

[dansrfe]

Exactly, I recently did a clean re-install and disabled the iGPU in device manager. It defaulted to the basic Microsoft driver and nvidia control panel refused to launch citing that the display isn't connected. The iGPU and dedicated GPUs are linked and everything still goes through the iGPU though it may or may not be processed by the nvidia GPU.

[huhn]

There is nothing to indicate either the OS or the gpu drivers. People who've had the black screens in games and MPC have tried the last resort clean OS installs, and it doesn't work.

so games have the problem to? so may ask nvidia?

[EZS]

Hello

I think MADVR has a bug with rendering TSC2 codec, which is one of Techsmith Camtasia capture codec.

Because when I set it back to MPC-HC default, or the Enhanced Video Renderer, the problem goes away.

Also, if you look closely you can see it's kind of repeating the playback, like there are 2 layers or something going, very blocky and blinking, unwatchable.

It's not that of a problem but it's a pretty normal bug.

Every other detail, a demo, including mediainfo, settings, etc, is in this video here I put on mediafire:

https://www.mediafire.com/?zsuxjujyabvvkhz

Oh, hopefully this is the correct thread for sending bug report ... wasn't sure. can it be it's own thread perhaps?

[michkrol]

Hello

I think MADVR has a bug with rendering TSC2 codec, which is one of Techsmith Camtasia capture codec.

Because when I set it back to MPC-HC default, or the Enhanced Video Renderer, the problem goes away.

Also, if you look closely you can see it's kind of repeating the playback, like there are 2 layers or something going, very blocky and blinking, unwatchable.

It's not that of a problem but it's a pretty normal bug.

Every other detail, a demo, including mediainfo, settings, etc, is in this video here I put on mediafire:

https://www.mediafire.com/?zsuxjujyabvvkhz

Oh, hopefully this is the correct thread for sending bug report ... wasn't sure. can it be it's own thread perhaps?

Welcome to the forum.

Is there any reason, you can't just share the file that doesn't work? We could then see if it's a problem with your setup or really a bug in madVR. A short sample of the video is usually enough and being able to reproduce a bug makes it easier for the developer to fix.

If the problematic file is something you don't want to/can't share publicly, you could probably send a link directly to madshi, but bear in mind he's quite busy right now, so you might not hear back from him soon.

This video doesn't help that much and honestly, it's inconvenient to work with. A series of screen-shots or just text info would be much easier to use. Debug info (CTRL+J) shows many useful information (a screen shot is enough). Check if you're using the same filters (decoders) with both madVR and EVR.

Backup madVR's settings and reset them to defaults, just to make sure it's not some configuration problem.

PS. Yes this is the correct thread for bug reports.

[StinDaWg]

What is the biggest cause for high deinterlace stats? I have a 7850, and deinterlacing is set to vector adaptive. Sometimes max stats (5s) will get into the 20-27 range which causes occasional dropped frames. Movie frame interval is 20.85ms. Is there any settings I can change to lighten the load?

[EZS]

Welcome to the forum.

Is there any reason, you can't just share the file that doesn't work? We could then see if it's a problem with your setup or really a bug in madVR. A short sample of the video is usually enough and being able to reproduce a bug makes it easier for the developer to fix.

If the problematic file is something you don't want to/can't share publicly, you could probably send a link directly to madshi, but bear in mind he's quite busy right now, so you might not hear back from him soon.

This video doesn't help that much and honestly, it's inconvenient to work with. A series of screen-shots or just text info would be much easier to use. Debug info (CTRL+J) shows many useful information (a screen shot is enough). Check if you're using the same filters (decoders) with both madVR and EVR.

Backup madVR's settings and reset them to defaults, just to make sure it's not some configuration problem.

PS. Yes this is the correct thread for bug reports.


Well I can create a sample no problem, it's just that most of the recordings were quite long.

I just reinstalled everything around MPC-HC as I cleaned out some codec packs earlier, this was a month ago including the video, I know about the stats display now.

It's not file specific, it's codec specific, any kind of recording has this bug.

I spent quite some time doing those settings so unless I can save to profile I don't feel like setting it all to default, as a matter of fact I can give you half-a-word that it didn't work from the beginning, it never did with madvr.

https://www.mediafire.com/?g8txs1rhg6ytz9x

Oh, i just found a clue, i went frame step and seems like some of the frames work, most probably I-frames, that's why fast seek under pause seems to work nicely, until you start playing back.

EDIT: these security questions feel kind of too cryptic, i know the land, but not exactly this site, popular mpeg-4 asp encoder ?

[huhn]

Well I can create a sample no problem, it's just that most of the recordings were quite long.

I just reinstalled everything around MPC-HC as I cleaned out some codec packs earlier, this was a month ago including the video, I know about the stats display now.

It's not file specific, it's codec specific, any kind of recording has this bug.

I spent quite some time doing those settings so unless I can save to profile I don't feel like setting it all to default, as a matter of fact I can give you half-a-word that it didn't work from the beginning, it never did with madvr.

https://www.mediafire.com/?g8txs1rhg6ytz9x

Oh, i just found a clue, i went frame step and seems like some of the frames work, most probably I-frames, that's why fast seek under pause seems to work nicely, until you start playing back.

EDIT: these security questions feel kind of too cryptic, i know the land, but not exactly this site, popular mpeg-4 asp encoder ?

you are not using windows xp aren't you? there was another issue with an xp PC and a desktop recoding tool. but most post got deleted (someone was not that "nice"...).

your sample works totally fine what should I see go wrong there?

there is a bug tracker that's a good place for bugs too:
bug tracker:

http://madVR.bugs.madshi.net

[vivan]

It's not file specific, it's codec specific, any kind of recording has this bug.madVR is a video renderer, not decoder - it's doesn't know anything about codecs. However, there're still few source-related differences with EVR (EVR doesn't support 4:4:4 chroma subsampling your file has - so LAV decoder have to convert it to 4:2:2).
Try disabling 4:4:4 formats in LAV decoder and check if you still have that issue with madVR.

https://www.mediafire.com/?g8txs1rhg6ytz9xI don't see any artifacts here too.

EDIT: these security questions feel kind of too cryptic, i know the land, but not exactly this site, popular mpeg-4 asp encoder ?those questions are outdated, but the answer is xvid.

[EZS]

Allright, seems like it's a LAV issue, i found out due to seeing that MPCHC internal LAV decoder icons were not loading, then i found out it's not the icons, it's they're not being loaded when I run tsc2 video, works fine with rest.

So it appears to be, for not, LAV's problem, been troubleshooting with #mpc-ch over at freenode and will report back.

Will try the 4:4:4 thing as well.

[huhn]

would be nice to know if you are using win xp just to know if this is really the same issue reported be for.

[nevcairiel]

How is it a LAV issue if LAV isn't even used? :p

[huhn]

How is it a LAV issue if LAV isn't even used? :p

I call this an issue!

[nevcairiel]

With his configuration though, LAV supports his sample file just fine.

[EZS]

LAV Splitter: 0.62.0.0
LAV Video Decoder: 0.62.0.0
LAV Audio Decoder: 0.62.0.0
MPC-HC (standalone, not from any pack): 1.7.6

Win7 x64
Intel Core i7 3820 (Sandy bridge E)
Radeon HD 5850 (14.4 catalyst)
Asus P9X79 Motherboard

https://pastee.org/5cze7

scroll down for pin info

EDIT:
EVR mode AVI decompresser pin info here, first link only madvr mode

https://pastee.org/e5azs


Jeeb from #Mpc-Hc says it's some weird stuff that the avi or lav might be throwing out and MADVR is just taking and rendering it out, not concluded yet.

EDIT2:

EVR Mode "color space converter" pin info https://pastee.org/zzwns

[boday]

This problem seems really weird to me...

Just bought a Intel NUC D54250WYK powered by i5 4250U with HD 5000. Installed Windows 8.1 with all system updates, all drivers are recent versions (July 2014).

Playing any video with MPC-HC with madVR gets unreasonably high rendering time, and thus tons of dropped frames. For example, playing a normal 10-bit 1080p video without any image scaling gives roughly 40 ms of average rendering time, which on my laptop with HD 4000 this should be well within 10 ms.

I have the latest version of KCP which contains the following components:
- MPC-HC version 1.7.6
- LAV filters 0.62.0
- madVR 0.87.10
- XySubFilter 3.1.0.697

I have tried the following but won't help:

- Use different versions of graphics driver as old as 10.18.10.3412 in Feb 2014 to the latest version;
- Change all scaling algorithms as low as bilinear or even nearest neighbor;
- Disable smooth motion;
- Enable/disable "windowed overlay", "fullscreen exclusive mode", "use separate device for presentation/DXVA processing";
- Enlarge CPU/GPU queue size;
- Any settings under "trade quality for performance".

Video playback with madVR disabled (using EVR-CP), or using other video players (mpv and VLC) seems perfectly fine.

This is a very strange issue since this NUC model has been out for more than half a year, and I've seen quite a few people on the Internet said they're using this as an HTPC without any complaint. Reviews from AnandTech and MissingRemote also report it works reasonably well with madVR. Moreover, the current madVR version has been released quite a while, and I've seen no issue reported about madVR working with Intel GPUs or Intel's recent versions of graphics driver.

Maybe the NUC is somehow broken?

I have no clue for what is wrong here, and I appreciate any ideas.:(

[Superb]

boday: Try the newer driver that Intel released a few days ago. It's in the Intel Download Center. v3907.
Get rid of KCP and install the stuff manually! Maybe KCP changed something w/ the defaults.
Check without XySubFilter.
Reset madVR's settings. There is a bat file in the madVR folder to do that.
Reset MPC-HC's settings either from the Options panel, during uninstallation, or during manual installation.
Don't use external LAV. Use the one that come w/ MPC-HC.
If none of the above helps, try playing w/ the flush settings under madVR's settings->rendering->exclusive mode settings.

[huhn]

@EZS

I guess I found the problem or at least something that will lead to it.

first of all MPC-HC interal LAVaudio and video isn't used for this sample only the splitter. in my case the external filter are used (I usually external lavfilter and MPC-BE). so I guess the decoder/transformfilter that is used instead is the problem. and the issue is only visibly with madVR because it can handle all kinds of colorspaces.

can you give as a screen like this?
http://abload.de/img/avifileaiqen.png

please try mpc->BE< with there internal filter and list what you have installed for codecs.

I guess the problem lies with internal LAVfilter in mpc-hc that can't deal with that file for reasons I don't know yet.

and we may use a new thread if madVR is not the proglem in this case.
@boday

can you please check the powerstate of your gpu with gpu-z or a program like that?
and a screen of the OSD (control + j)

[EZS]

Okay here it is, http://i.imgur.com/ifB9Htb.jpg

Also my replies will be slow now, got some stuff in these couple of days.

I didn't finish with JEEB from #MPC-HC , he said it's doing some stuff either avi decompresser or splitter and it's not putting correct output or what ..

Or rather he said that he sees VFW codecs being forced in instead of LAV so MADVR doesn't work right, we don't know yet who's to blame.

[huhn]

i can say for sure that the video works right with lavvideo.

so install lavfilter and make sure it is loaded. i don't know why the internal mpc-hc lavfilter isn't decoding the codec with there lavfilter. you can try to block the avi decompressor to.

I guess the problem ti's the avi decompressor that tries to send a colorspace which it can't handle properly

[boday]

Try the newer driver that Intel released a few days ago. It's in the Intel Download Center. v3907.
Get rid of KCP and install the stuff manually! Maybe KCP changed something w/ the defaults.
Check without XySubFilter.
Reset madVR's settings. There is a bat file in the madVR folder to do that.
Reset MPC-HC's settings either from the Options panel, during uninstallation, or during manual installation.
Don't use external LAV. Use the one that come w/ MPC-HC.
If none of the above helps, try playing w/ the flush settings under madVR's settings->rendering->exclusive mode settings.
Thanks for the reply!

Just updated again to the latest Intel driver v3910. No improvement.

Also tried uninstall KCP and installed the latest build of 32-bit MPC-HC and madVR, so now there is no external LAV and XySubFilter. I can see the performance is sort of better, as now it can play for 30 s or so without problem, but then it begin dropping frames again when the cooling fan goes up and I can see the CPU is working at a restricted frequency (no turbo boost).

I think the improvement is expected as I'm not using XySubFilter now. And I don't think the over-heating is the problem here, because I can play movies with mpv for 2 hours without any dropped frames, while the CPU usage was maintained at ~50%. So over-heating is more like the consequence, not the cause. (Well after did the test below with the screenshots, I doubt there is any over-heating of the chips because I can play with mpv without problem, and CPU frequency went up to 1.8 GHz right after "over-heated" with madVR).

And again, even within the first 30 s when playing with MPC-HC and madVR, the average rendering time was ~25 ms which is still higher than expected.

@boday

can you please check the powerstate of your gpu with gpu-z or a program like that?
and a screen of the OSD (control + j)
Thanks for your suggestion. Attachments are the screenshots when playing the same file using MPC-HC+madVR or mpv together with Windows task manager and GPU-Z. GPU load is maintained at ~70% during playback with madVR and ~40% with mpv (using vo=opengl-hq). Note that there is no imaging scaling in both situation, just chroma upsampling.

[huhn]

I think the improvement is expected as I'm not using XySubFilter now. And I don't think the over-heating is the problem here, because I can play movies with mpv for 2 hours without any dropped frames, while the CPU usage was maintained at ~50%. So over-heating is more like the consequence, not the cause. (Well after did the test below with the screenshots, I doubt there is any over-heating of the chips because I can play with mpv without problem, and CPU frequency went up to 1.8 GHz right after "over-heated" with madVR).

madVR doesn't really use the CPU

And again, even within the first 30 s when playing with MPC-HC and madVR, the average rendering time was ~25 ms which is still higher than expected.
is the file interlaced or high FPS?

Thanks for your suggestion. Attachments are the screenshots when playing the same file using MPC-HC+madVR or mpv together with Windows task manager and GPU-Z. GPU load is maintained at ~70% during playback with madVR and ~40% with mpv (using vo=opengl-hq). Note that there is no imaging scaling in both situation, just chroma upsampling.
the screens can take some time a mod has to look at them first.
you can simply upload them to a image hoster in the future and post a link.

you get 40% gpu usage with mpv? i get about 10 ms at 350 mhz with an old HD 4000 and a 1080p blu ray backup. there is something else eating your GPU. have you disabled the image things under video in the intel driver?(there is a drop down menu if you click on "video" for more options)

i get 8 ms with a 23p blu ray in full screen and my chroma/dither/SM settings. with 40 % gpu usage at 350-650 mhz

[boday]

madVR doesn't really use the CPU
Yeah, but I just thought this might indicate the chips are over-heated, but seems it's not the case.

is the file interlaced or high FPS?
No, it's a normal 23.976 fps, progressive 1080p movie file.

the screens can take some time a mod has to look at them first.
you can simply upload them to a image hoster in the future and post a link.
Sorry...still new to this forum. Here are the links:
https://www.dropbox.com/s/1w42vcywzcdoo8z/Capture.jpg?dl=0
https://www.dropbox.com/s/ihc0h154r4r5kzc/Capture1.jpg?dl=0

you get 40% gpu usage with mpv? i get about 10 ms at 350 mhz with an old HD 4000 and a 1080p blu ray backup. there is something else eating your GPU. have you disabled the image things under video in the intel driver?(there is a drop down menu if you click on "video" for more options)

i get 8 ms with a 23p blu ray in full screen and my chroma/dither/SM settings. with 40 % gpu usage at 350-650 mhz
That reminds me. I just tried playing the file on my laptop with i7-3520m and HD 4000, and GPU load is only 25% for mpv, and around 70% for KCP (same config as I originally mentioned) with high settings, but the rendering time is only 8-9 ms, though.

Yes I have disabled all the video processing functions in the Intel driver setting, like color correction, skin tone enhancement, noise reduction etc.

So, it seems even with mpv the GPU is not working as expected, but just because mpv is less GPU-demanding so it appears to be OK.

[huhn]

That reminds me. I just tried playing the file on my laptop with i7-3520m and HD 4000, and GPU load is only 25% for mpv, and around 70% for KCP (same config as I originally mentioned) with high settings, but the rendering time is only 8-9 ms, though.

Yes I have disabled all the video processing functions in the Intel driver setting, like color correction, skin tone enhancement, noise reduction etc.

So, it seems even with mpv the GPU is not working as expected, but just because mpv is less GPU-demanding so it appears to be OK.

I meant 10 % usage with mpv not 10 ms.