madVR - high quality video renderer (GPU assisted)

[bugmen0t]

Hi Madshi!

I want to make a log for the smooth motion black frame bug but I don't want the log to be written to my desktop/ssd (only if you still need one, of course).
How can I change the log path?
Since it will probably take several movies and days until I get a succesful log, what is the best way of doing it? I guess I should delete any old log before starting the next movie and stop it immediately after a black screen happens. Do you need the complete log then or only the last seconds/minutes? Any advice here?

[madshi]

I want to make a log for the smooth motion black frame bug but I don't want the log to be written to my desktop/ssd (only if you still need one, of course).
How can I change the log path?
Since it will probably take several movies and days until I get a succesful log, what is the best way of doing it? I guess I should delete any old log before starting the next movie and stop it immediately after a black screen happens. Do you need the complete log then or only the last seconds/minutes? Any advice here?
You can use this to redirect the log file:

http://en.wikipedia.org/wiki/NTFS_symbolic_link
http://lifehacker.com/5496652/how-to-use-symlinks-in-windows

Yes, you should *definitely* delete the log file before starting a new movie playback. And it would be very cool if you could close the media player as fast as possible after the black frame appeared. It would also be nice if you could maybe count the seconds it took from the black frame to closing the media player, to help me find the right place to look for in the log. You don't have to upload the whole log. If you look at the first column of the log, it contains the time difference from the start of media playback until each logged event. So e.g. if you needed 10 seconds from black frame to closing the media player, then just make sure the last e.g. 30 seconds are still in the log file. You can delete the rest. But you can also upload the whole log file, if you prefer it that way. Just make sure you 7zip it.

Thanks...

[khanmein]

@madshi, right clicked > filters > "mpc matroska splitter (low merit)" for all .mkv files. when i tick "matroska" at option > source filters it won't show again.

what's the meaning of low merit? thanks bro!

[madshi]

@khanmein, I don't see what this has to do with madVR?

[ryrynz]

That's too bad, really. I fear without a log (which ends very near to the black screen) this might be hard to fix.


It's strange you haven't encountered it considering.. I could certainly grab a log of it. With bugmen0t keen on capturing it I'll wait and see how his log fares and chip in if needed.
Having someone log it that experiences this a lot more often would be advantageous, anyone?

[cyberbeing]

@khanmein

That means you neither have a external MKV splitter installed, nor do you have the MPC-HC internal MKV splitter enabled. MPC-HC never truly disables its internal filters, instead assigning them a "low merit" so they can be used as a last resort when no other connectable filters are available. At least that has been my observation.

As madshi mentioned, this behavior shouldn't have anything to do with madVR. Install a MKV Splitter like Haali or LAV Filters and it should go away.

[madshi]

It's strange you haven't encountered it considering..
Well, I do actual movie watching on my HTPC/projector, which I drive with native frame/refresh rate. So smooth motion FRC is off there. I do have smooth motion FRC turned on on my development PC because my computer monitor can only do 60Hz. But on the dev PC I'm not watching movies, just doing quick checks. Never seen the black frame problem here. But it makes sense since it seems to occur only rarely and I only play video files for seconds or very few minutes at a time on the development PC.

[dansrfe]

madshi, out of curiosity, which projector do you use?

[chros]

Of course Anti-ringing costs extra performance. Maybe your GPU is too slow for that?
Hhmmm, I thought of that, but the strange thing is that the GPU utilization is around 5% (I see it in GPU-Z). I'll check it in this evening again ...

[karamancho]

Is this a new problem with v0.86.3? Or do you have this problem with all older madVR builds, too?

just checked v0.86.1 and the behavior is the same. queues dropping resulting in a lot of dropped/delayed frames whenever the subs are present on the screen. happens only with MPC-HC internal subs render and only in windowed mode. normal behavior with DirectVobSub or with subs disabled. MPC-HC internal subs render in FSE mode works fine too.

I guess I haven't noticed this before since I never watch movies in windowed mode

[kiccolsd]

Try doing a very small zoom with 1080p mkvs, like 1% smaller or bigger. Does that make the problem go away? And yes, please also try windowed mode, just to make sure. Do you happen to have a digicam which might be able to capture the artifact you're talking about? It would be easier for me to help if I knew exactly how the artifact looks like.

As said before I'm a perfect noob, Can you help me with zoom? In MPC-HC I can find only 50%, 100%, 200%, Automatic. BTW yes I have an iPhone5 to make a video when is ready I'll PM you.
Thx for your support!!

Edit: I tried with 200% zoom and windowed mode (after unchecking exclusive mode in madvr settings)... nothing changes. I'm going to record the issue. Check your PM soon.

[dansrfe]

madshi, is there any potential fix for the secondary screen becoming unusable once a video is in exclusive mode on the primary screen? No window can be focused unless the video is in windowed mode on the primary. Should I fill out a bug report for this or is this behavior expected?

Also, this may or may not be related but I was playing around with DisplayFusion which creates the taskbar on all secondary monitors and the unresponsive monitor problem shows up on all screens, regardless of which screen the video is playing on in exclusive mode.

[truexfan81]

while my internet was down for an hour i decided to experiment with the madvr deinterlacer. I noticed when i went into the nvidia control panel and disabled inverse telecine, madvr stopped dropping frames and the gpu usage went down by 30%. so my question is what is it about inverse telecine that causes it to use so much power?

[khanmein]

@madshi, thanks.

@cyberbeing, i did installed lav filters since the beginning. now i need to ticked the matroska at source filters to get rid of low merit. thanks.

[madshi]

madshi, out of curiosity, which projector do you use?
JVC X3.

Hhmmm, I thought of that, but the strange thing is that the GPU utilization is around 5% (I see it in GPU-Z). I'll check it in this evening again ...
That's weird. Jinc3 should push GPU up *much* more than just 5%.

just checked v0.86.1 and the behavior is the same. queues dropping resulting in a lot of dropped/delayed frames whenever the subs are present on the screen. happens only with MPC-HC internal subs render and only in windowed mode.
Not sure what to say. Maybe your PC is simply not fast enough for this. Or maybe the MPC-HC internal sub renderer does something weird. I don't really know. I do recommend using fullscreen exclusive mode...

As said before I'm a perfect noob, Can you help me with zoom? In MPC-HC I can find only 50%, 100%, 200%, Automatic.
Just press the "9" key in the num block. That will very slightly zoom into the image.

Edit: I tried with 200% zoom and windowed mode (after unchecking exclusive mode in madvr settings)... nothing changes. I'm going to record the issue. Check your PM soon.
So this issue only occurs with 720p videos but not with 1080p videos? Sounds weird. Does it occur with *every* 720p video? And does it *not* occur with every 1080p video? And you double checked that VMR9/EVR don't show this problem with the same 720p video, correct?

madshi, is there any potential fix for the secondary screen becoming unusable once a video is in exclusive mode on the primary screen? No window can be focused unless the video is in windowed mode on the primary.
What happens if you start a game in fullscreen exclusive mode on the primary monitor? Can you then focus windows on the secondary screen?

while my internet was down for an hour i decided to experiment with the madvr deinterlacer. I noticed when i went into the nvidia control panel and disabled inverse telecine, madvr stopped dropping frames and the gpu usage went down by 30%. so my question is what is it about inverse telecine that causes it to use so much power?
Does that have anything to do with your bug #85? It's hard for me to handle bug reports if you write something to the bug tracker and something to the forum...

[truexfan81]

Does that have anything to do with your bug #85? It's hard for me to handle bug reports if you write something to the bug tracker and something to the forum...

no madshi this is unrelated, just a general question

[madshi]

while my internet was down for an hour i decided to experiment with the madvr deinterlacer. I noticed when i went into the nvidia control panel and disabled inverse telecine, madvr stopped dropping frames and the gpu usage went down by 30%. so my question is what is it about inverse telecine that causes it to use so much power?
Well, NVidia and AMD (not sure about Intel) support per pixel film mode detection and inverse telecine. It's not too easy to do this well, so it does cost quite a bit of GPU performance. madVR's IVTC (which is inverse telecine + decimation) runs on the CPU instead (for now). It performs better and has better quality, IMHO, compared to what NVidia and AMD are doing. However, it's more limited in functionality: It's not per-pixel and it doesn't (yet) auto-detect whether the content is video or film.

[ajp2k11]

New unreleased Intel driver for Ivy/Haswell for those who want to try, don't have Intel graphics myself...

http://asder00.blogspot.se/2013/06/intel-hd-graphics-15.31.12.3204.html

[dansrfe]

What happens if you start a game in fullscreen exclusive mode on the primary monitor? Can you then focus windows on the secondary screen?

As odd as it may be, I don't actually have any games installed to test this with :/

But, I do have VMware installed and according to this: http://pubs.vmware.com/workstation-9/index.jsp?topic=%2Fcom.vmware.ws.using.doc%2FGUID-848D482F-C93C-461C-9A9A-C3FF51A0F03E.html I can go into exclusive mode with virtual machines.

So, when I try this with virtual windows and linux machines on primary or secondary monitor, the mouse becomes locked to that screen which prevents me from changing focus unless I use the keyboard which then causes VMware to take the virtual machine out of exclusive mode and into windowed fullscreen mode instead.

According to other gaming forums, a similar behavior is exhibited while playing games in exclusive mode, but I can't say for sure.

[madshi]

So, when I try this with virtual windows and linux machines on primary or secondary monitor, the mouse becomes locked to that screen which prevents me from changing focus unless I use the keyboard which then causes VMware to take the virtual machine out of exclusive mode and into windowed fullscreen mode instead.

According to other gaming forums, a similar behavior is exhibited while playing games in exclusive mode, but I can't say for sure.
So this does not seem to be a madVR problem, but a general limitation of FSE mode, correct?

[dansrfe]

So this does not seem to be a madVR problem, but a general limitation of FSE mode, correct?

I suppose so. madVR is better in that respect I guess. It frees up the primary monitor while in FSE mode on the secondary.

[hdboy]

Is it possible for madvr to tell when the current refresh rate is 59.99hz and not 59.94hz? I set madvr to switch to 1080p59 but when the display is at 60hz already, madvr does not switch to 59hz. ie it doesn't distinguish between the two. If the display is currently at 23hz, otoh, it does switch to 59hz.

[pie1394]

Just found a better setting which runs much stable with madVR unless the GPU is overloaded.

CPU queue size: 16
GPU queue size: 16
back buffers: 8 (both window and FSE)
present several frames in advance: [checked]
after last render step: [flush & wait(sleep)] or [flush]
others: [don't flush]

It prevents frame dropping at dynamic GPU frequency change during the playback which even happens to the powerful HD7970.

On my ION system, the tearing does not happen to window mode now. It is verified with the observation on the quick horizontal panning scene for 24fps contents at 60Hz output. (If smooth-motion is ON, it still happens...)

Some people notice the present queue number keeps dropping or low even GPU loading value is not at 100%. But I can say the root cause is often likely overloaded GPU. It happens to my ION system with 1920x1080i60 contents at any window size. If GPU deinterlacing job is required on such so-so GPU, it causes frame dropping when the GPU loading goes up about 70~%. (I think...)


ps: The above setting is verified ok on my 2 systems: (OS is win7x64 sp1)

Mini-server: Core2Duo P9700(1.6 ~ 2.8GHz) + 8GB DDR3-1333 + ION-LE GF9300 256MB share-mem (Geforce 314.22) + Bi-linear + Window mode ==> Dell U2412M 1920x1200p60

HTPC: Core i5-3570K(1.6 ~ turbo 3.8GHz) + 8GB DDR3-2400 + HD7970 3GB DDR5-5500 (Catalyst 13.6 beta) + Jinc3_AR + FSE + auto-refresh-rate changer ==> Sony 65HX920 1920x1080p24/p50/p60

[Vyral]

Hi,
I did some tests yesterday with GPU-z to check GPU load and there is something strange. Here are the result :
1080p videos : 0% GPU load ; ~7ms rendering time
720p videos : 30% GPU load ; ~11ms rendering time

Here are my specs :
Hardware :
Intel Core i3-2100 @ 3.10GHz
AMD Radeon HD 6850
CCC :
screen : 1080p60 ; color 32bits
color & quality settings : all unchecked
pixels : RGB 4:4:4 (full range)
everything else is default
LAV :
default settings
xy-VSFilter :
default settings
madVR :
display already calibrated ; BT 709 ; ppc 2.20
Jinc 3 AR / Jinc 3 AR / Catmull-Rom AR,LL
CPU 12
GPU 8
backbuffer : 4 (both window and FSE)
use a seperate device for presentation
smooth motion ON if judder
everything else is default

Even if the video match the screen, GPU load should not be 0%, right ?

[chros]

Hhmmm, I thought of that, but the strange thing is that the GPU utilization is around 5% (I see it in GPU-Z). I'll check it in this evening again ...
Sorry, it was my fault. MadVR had used intel GPU and not nVidia.
I have renamed mpc-hc.exe to game.exe (thanks for the hint) and now it's working flawlessly!

[digitech]

Just found a better setting which runs much stable with madVR unless the GPU is overloaded.

CPU queue size: 16
GPU queue size: 16
back buffers: 8 (both window and FSE)
present several frames in advance: [checked]
after last render step: [flush & wait(sleep)] or [flush]
others: [don't flush]

It prevents frame dropping at dynamic GPU frequency change during the playback which even happens to the powerful HD7970.

On my ION system, the tearing does not happen to window mode now. It is verified with the observation on the quick horizontal panning scene for 24fps contents at 60Hz output. (If smooth-motion is ON, it still happens...)

Some people notice the present queue number keeps dropping or low even GPU loading value is not at 100%. But I can say the root cause is often likely overloaded GPU. It happens to my ION system with 1920x1080i60 contents at any window size. If GPU deinterlacing job is required on such so-so GPU, it causes frame dropping when the GPU loading goes up about 70~%. (I think...)


ps: The above setting is verified ok on my 2 systems: (OS is win7x64 sp1)

Mini-server: Core2Duo P9700(1.6 ~ 2.8GHz) + 8GB DDR3-1333 + ION-LE GF9300 256MB share-mem (Geforce 314.22) + Bi-linear + Window mode ==> Dell U2412M 1920x1200p60

HTPC: Core i5-3570K(1.6 ~ turbo 3.8GHz) + 8GB DDR3-2400 + HD7970 3GB DDR5-5500 (Catalyst 13.6 beta) + Jinc3_AR + FSE + auto-refresh-rate changer ==> Sony 65HX920 1920x1080p24/p50/p60

I'm curious which are your madvr settings (scaling chroma,image, resolution,etc) that you use with your nvidia ion i have one too for my htpc with an atom 330 and i haven't been able to experience a totally smooth reproduction in madvr yet.

[pie1394]

I'm curious which are your madvr settings (scaling chroma,image, resolution,etc) that you use with your nvidia ion i have one too for my htpc with an atom 330 and i haven't been able to experience a totally smooth reproduction in madvr yet.

Scaling : Bi-linear
Subtitle: ASS or SSA
Player: MPC-BE x86

1280x720p24 H.264 Hi10p -> 1920x1080 [ok]
1920x1080p24 H.264 Hi10p -> 1920x1080
[ok] for < 6 ~ 8 Mbps source (I think)
[frame drop] for >= 10 Mbps source

1440x1080i60 / 1920x1080i60 MPEG2-HD -> 1920x1080p60
[frame drop] if GPU deinterlace is in use
[ok] if GPU deinterlace is turned off

For the contents with ION chipset's DXVA decoding via LavFilter's video, it should just work fine with Atom330 CPU.

With Atom330, FFMPEG in LavFilter can only decode some low-bit rate SD H.264 Hi10 contents. Atom330 is too weak to decode any high bit-rate HD one.

The C2D P9700@2.8GHz (6MB L2 cache) is about 4 times as powerful as Atom330 on 1~2 threaded program. It is just fast enough to decode regular Hi10 FHD 24fps contents. In addition my ION MB's memory sub-system runs at dual-ch 128-bit DDR3-1333 mode, not the single-ch 64-bit DDR2-667, DDR3-800/1066 one on most Atom330+ION MB or Mac-mini. (ps: MB is AOpen GP7A)

With the H.264 4K contents at 60~100Mbps or HEVC 4K, it needs Core i5-3570K@3.4GHz (4core 6MB L3) computing power to do that. My previous HTPC was C2D E8400@3.6GHz (2core 6MB L2). It can only handle those < 15 Mbps ... if I don't remember it wrong.

[Chortos-2]

madshi, can’t you convert the subtitles from PC to TV and then blend them onto the video instead of converting the video to PC, blending and then converting back to TV?

[pie1394]

Sony PS4 list price --- US$399

It will be interesting if MPC-BE + LavFilter + madVR can be run on it. Just like most x86 PC, this box with x86 CPU+AMD GCN GPU obviously runs with the regular Win8 OS (Sorry, not quite sure if it is the same with XBox One ... most likely a custom-built OS derived from PS3 OS. The only chance is another Win7/Win8 OS installation.). PS4's GPU should be capable of running Jinc3_AR scaling algorithm at 60 fps for SD(480i) ~ FHD(1080i/p) contents. :p

I don't think any custom-built HTPC with the same price is capable of such GPU performance. US$600~800 might be the minimum cost. :rolleyes:

[digitech]

Sony PS4 list price --- US$399

It will be interesting if MPC-BE + LavFilter + madVR can be run on it. Just like most x86 PC, this box with x86 CPU+AMD GCN GPU obviously runs with the regular Win8 OS. PS4's GPU should be capable of running Jinc3_AR scaling algorithm at 60 fps for SD(480i) ~ FHD(1080i/p) contents. :p

I don't think any custom-built HTPC with the same price is capable of such GPU performance. US$600~800 might be the minimum cost. :rolleyes:

Nice point and that would be an awesome-best bang for your buck as a media player, but i doubt our favorites movie players-decoders-renders will be easilily ported without waiting a LOT, i don't think sony will give us a piece of cake crackable console, (got to learn from the past) but anything is possible, time will tell.

[bugmen0t]

Hi madshi!

I want to report back that I now have a bit over 100gb of madvr logs but without any black frames. The last one I had was last week with v0.86.2 before I started logging. With v0.86.3 I only watched one movie in release mode, all others in debug mode and not one black frame (or I missed them). Since other people reported this bug with the current version, however I guess it's still there.

What to do now? Should someone else try it? Is it possible it doesn't happen in debug mode? I can continue making logs but I'm not sure it will help.

[Snowknight26]

When creating a graph with madVR in it, GraphStudioNext doesn't have a close button on the video window (it does with other renderers). Would that be a madVR quirk?

[ryrynz]

What to do now? Should someone else try it? Is it possible it doesn't happen in debug mode? I can continue making logs but I'm not sure it will help.

I've enabled it a few times but haven't had it happen with debug mode enabled either. Bad luck? IDK. But I did consider the possibility that the debug build wasn't affected by this also.

[sheppaul]

I cannot run the settings window of madVR. Is it impossible to call madVR perferences' window before starting playback?
Is there any reason to check madVR instance while trying to configure madVR before running?


---------------------------
madVR instance not found.

Please make sure that your firewall doesn't block madVR.
---------------------------

[nevcairiel]

I cannot run the settings window of madVR. Is it impossible to call madVR perferences' window before starting playback?

No, that is not currently possible.

[sheppaul]

No, that is not currently possible.

Then, please allow this if possible...

:)

[flashmozzg]

Which OS are you using? Could you try different madVR builds to see if this was introduced in a recent build? You'll find download links for older madVR builds on the first post in this thread.


Sorry, missed that post and replying only now. I'm using Win7 Ult. sp1 x64 . I believe it was introduced when DXVA2 was implemented (0.85), but i night be wrong. I'll check now.
Edit: Checked. I'm on 0.84.7 now. 0.84.6 works fine, 0.85 and over has this problem.

[ryrynz]

Running the new 9.18.10.3204 drivers (15.31.12.64.3204) for Haswell/Ivy and the issues experienced with MadVR's seekbar not disappearing are fixed.

[bugmen0t]

Tried one movie in release mode and had a black frame within the first few minutes so at least I can confirm it's still there. :rolleyes:

[TheShadowRunner]

Most likely dumb question but where do I find this XySubFilter..?

[TOM_SK]

Most likely dumb question but where do I find this XySubFilter..?

xy-vsfilter not XySubFilter:
http://code.google.com/p/xy-vsfilter/wiki/Downloads?tm=2

[DarkSpace]

Most likely dumb question but where do I find this XySubFilter..?
xy-vsfilter not XySubFilter:
http://code.google.com/p/xy-vsfilter/wiki/Downloads?tm=2
I'm fairly sure he did mean XySubFilter, as that it the version that sends the subpictures to madVR directly instead of blending them onto the video.
Unfortunately, as far as I'm aware, XySubFilter is still work in progress and a public beta for testing purposes is not yet released, so the only way to get it is to compile it yourself from the source.

[TheShadowRunner]

I'm fairly sure he did mean XySubFilter, as that it the version that sends the subpictures to madVR directly instead of blending them onto the video.
Unfortunately, as far as I'm aware, XySubFilter is still work in progress and a public beta for testing purposes is not yet released, so the only way to get it is to compile it yourself from the source.
Yes indeed, I did mean XySubFilter, guess I'll just wait, thanks for the info Dark.

[Buckster]

solved my performance issues with my 5670 and MadVR

seemed to be a combination of having Jriver set to use "Hardware Decoding where possible" on - which I've changed to off

and I had MadVR set to auto-switch to 1080p60 rather than 1080p59 (now instead of saying a frame repeat every 4 mins its 50 mins)

overall result, Jinc3 (upscaling), and Bicubic (chroma upscaling) - seems to work with no dropped frames

but even better Jinc3 (upscaling) and Jinc (chroma upscaling) - also now works with no dropped frames

:)

[madshi]

Is it possible for madvr to tell when the current refresh rate is 59.99hz and not 59.94hz? I set madvr to switch to 1080p59 but when the display is at 60hz already, madvr does not switch to 59hz. ie it doesn't distinguish between the two. If the display is currently at 23hz, otoh, it does switch to 59hz.
Which OS and which madVR build have you testing this with? The very latest build (v0.86.3) contains a special fix for Windows 8 which works around issues like this.

Just found a better setting which runs much stable with madVR unless the GPU is overloaded.

CPU queue size: 16
GPU queue size: 16
back buffers: 8 (both window and FSE)
present several frames in advance: [checked]
after last render step: [flush & wait(sleep)] or [flush]
others: [don't flush]
That doesn't look so very much different from the defaults? FWIW, I would recommend to use a CPU queue slightly higher than the GPU queue. E.g. try a CPU queue of 24 with everything else the same.

I did some tests yesterday with GPU-z to check GPU load and there is something strange. Here are the result :
1080p videos : 0% GPU load ; ~7ms rendering time
720p videos : 30% GPU load ; ~11ms rendering time

Even if the video match the screen, GPU load should not be 0%, right ?
The 6850 is quite fast, but even the 6850 should not be able to do Jinc3 AR chroma upsampling with 0% GPU. So I would say there's probably something wrong somewhere. Does the madVR OSD report "NV12" (YCbCr 4:2:0)?

madshi, can’t you convert the subtitles from PC to TV and then blend them onto the video instead of converting the video to PC, blending and then converting back to TV?
Converting the subtitles like this would mean that I'd have to modify every pixel of the bitmaps XySubFilter is sending me. Doing so would noticeably increase CPU usage. So I don't think this is a good approach.

Ok, I could also convert the subtitles on the GPU, but that would be more difficult than doing it on the CPU, and it might not be any faster than the current solution. XySubFilter delivered subtitles can contain of many small parts. I'd have to convert every single one of them separately...

I want to report back that I now have a bit over 100gb of madvr logs but without any black frames. The last one I had was last week with v0.86.2 before I started logging. With v0.86.3 I only watched one movie in release mode, all others in debug mode and not one black frame (or I missed them). Since other people reported this bug with the current version, however I guess it's still there.

What to do now? Should someone else try it? Is it possible it doesn't happen in debug mode? I can continue making logs but I'm not sure it will help.
I've enabled it a few times but haven't had it happen with debug mode enabled either. Bad luck? IDK. But I did consider the possibility that the debug build wasn't affected by this also.
Please read (and post feedback, if you can) here:

http://bugs.madshi.net/view.php?id=84

When creating a graph with madVR in it, GraphStudioNext doesn't have a close button on the video window (it does with other renderers). Would that be a madVR quirk?
madVR does not draw any buttons on the video window. I don't think the other renderers are doing that, either. I suppose that's probably a bug in GraphStudioNext? But I don't really know.

Then, please allow this if possible...
http://bugs.madshi.net/view.php?id=83

Sorry, missed that post and replying only now. I'm using Win7 Ult. sp1 x64 . I believe it was introduced when DXVA2 was implemented (0.85), but i night be wrong. I'll check now.
Edit: Checked. I'm on 0.84.7 now. 0.84.6 works fine, 0.85 and over has this problem.
Hmmmm... Weird. Can you please try resetting your MPC-HC settings? My best guess is that your color (saturation, hue etc) controls in MPC-HC are screwed up.

Running the new 9.18.10.3204 drivers (15.31.12.64.3204) for Haswell/Ivy and the issues experienced with MadVR's seekbar not disappearing are fixed.
Sounds good! I'll double check this with my Intel contact...

solved my performance issues with my 5670 and MadVR

seemed to be a combination of having Jriver set to use "Hardware Decoding where possible" on - which I've changed to off

and I had MadVR set to auto-switch to 1080p60 rather than 1080p59 (now instead of saying a frame repeat every 4 mins its 50 mins)

overall result, Jinc3 (upscaling), and Bicubic (chroma upscaling) - seems to work with no dropped frames

but even better Jinc3 (upscaling) and Jinc (chroma upscaling) - also now works with no dropped frames
Did you try 1080p60 with 23.976fps content or with 24.000fps content? Both exists. Usually the best solution is to allow madVR to switch to both 1080p59 and 1080p60, then madVR will choose which fits better to the source.

[madshi]

madVR v0.86.4 released

http://madshi.net/madVR.zip

* added support for drawing remote controlled calibration test patterns
* renamed madHcNet.dll to madHcNet32.dll
* drop/repeat estimate now works up to 240Hz
* tray icon appears faster now
* fixed: #81: Repeated frames count increases when video is paused
The key reason for this release is the added support for remote controlled calibration test patterns, in the hope that ArgyllCMS + HCFR will make use of that.

Caution: The file "madHcNet.dll" was renamed to "madHcNet32.dll" in this release. This might be important if you repackage the files somehow or have some batch files to copy the files or whatever. Before anyone asks: No, this is not a sign of a soon-to-come madVR 64bit build. That is still not planned anytime soon. However, there might be a "madHcNet64.dll" soon to make it possible for 64bit calibration software to show madVR calibration test patterns, too.

[Thunderbolt8]

how to understand remote controlled test patterns? Im sure you dont mean that I can control to run test patterns with my remote. but dont have much more of a clue how this can be used and for what.

[madshi]

That means that other software, e.g. ArgyllCMS or HCFR (or Calman or ChromaPure) can tell madVR to show a specific color test pattern. This works on the local PC, or also via LAN. E.g. you can run HCFR on your notebook, and run madVR on your HTPC, and HCFR on your notebook will be able to tell madVR (running on your HTPC) to show a specific color test pattern. This is what I mean with "remote controlling".

Please note that HCFR etc do not support this *yet*, but I hope they'll support it soon.

Of course for LAN remote controlling to work you need to enable LAN network access in the madVR tray icon configuration.

[Buckster]

Did you try 1080p60 with 23.976fps content or with 24.000fps content? Both exists. Usually the best solution is to allow madVR to switch to both 1080p59 and 1080p60, then madVR will choose which fits better to the source.

hi madshi - I'd set by mistake both 23.976 and 24 to switch to 1080p60, as the only auto switch modes I'd set MadVR to were 1080p50 and 1080p60

I didn't even realise that there was a different 1080p59 setting tbh, and in the ATI driver settings if I switch to 59hz it auto-chooses 60 - which is odd

but MadVR somehow is managing to switch to 59 anyway :) my frame repeat stats went from 4 mins to 50 mins + :)

so no dropped frames now, excellent picture quality, and much better motion than how my TV handles as stock 23.976 - so great

thanks :)

[mrkazador]

Running the new 9.18.10.3204 drivers (15.31.12.64.3204) for Haswell/Ivy and the issues experienced with MadVR's seekbar not disappearing are fixed.

Where did you get .3204 drivers? I'm using an Intel i3 NUC and just installed 9.18.10.3165 and the seekbar is not disappearing for me and can't find the drivers you listed.