madVR - high quality video renderer (GPU assisted)

[kalston]

I've had the lip sync issue as well when using the "overshoot max frame latency" with any file type (.mkv, .m2ts etc) before I was even using LAV at all so I'd blame madVR rather than LAV :) And it was with older builds of madVR too, a few months from now.

Anyway I don't need this option anymore since the old fullscreen mode works well enough for me, no matter what I can't get smooth glitch free playback with the new mode and recent nvidia drivers so I've given up.

[6233638]

If changing madVR version fixes his sync issues, how is that the splitters fault? :DRazoola's "overshoot max frame latency" issues may well have been caused by 0.80, but I've not been using that setting with madVR at all as it caused a lot of problems on my system when it was added. (it's for low powered cards that can barely run the software)


I've been experiencing lip-sync errors, specifically with Blu-ray playback, when using the .bdmv file to start the film since before December though. (madVR 0.80 was released on the 18th) That's why I was skeptical about it being something introduced in 0.80

What I was wondering, was if Plutotype had also updated LAV Filters as well as madVR recently, or changed from loading with an mpls or m2ts file to the bdmv (which is by far the easiest thing to do) and thought that maybe that was causing the lip-sync errors as his playback was OK in 0.79 when mine was not. It could be entirely unrelated though. I'm definitely interested in hearing whether dropping to 0.79 fixes things for him.


I had no issues prior to using .bdmv files to start playback which I only started using recently when I heard that LAV Filters had support for it. (is that a recent addition, or just something I only found out about recently?) Before then I was mostly loading the m2ts file directly, or mpls if it was a branching title. (though I can't remember if that was OK for lip-sync or not, that was broken for a long time too, but it may have been prior to switching to LAV from Haali)


I'm not saying LAV Filters is definitely the cause, but that's why I was wondering how Plutotype was starting playback of the film, whether it was the bdmv, mpls or m2ts file. It's possible that since he was updating madVR he has also updated LAV filters at the same time, or changed from one filetype to another. (I have a habit of doing that, and sometimes MPC-HC & ffdshow too, which is really great for troubleshooting problems if they arise...)

I haven't had a chance to properly test things myself yet, as I haven't had much time to watch films lately, or do testing which involves waiting 20–30 minutes to see if sound goes out of sync. (which means I can't use the PC for anything else)


Either it's a problem with madVR that goes back further than 0.80, an issue with LAV Filters opening .bdmv (and possibly .mpls) files, or something else entirely. At least with non-branching titles, it's easy enough to just open the .m2ts directly which does not seem to have lip-sync problems for me.

The only reason I'm suspecting LAV here at all, is because it changes depending on which file I open. (or at least it seemed to from my limited tests)

[nevcairiel]

I had no issues prior to using .bdmv files to start playback which I only started using recently when I heard that LAV Filters had support for it. (is that a recent addition, or just something I only found out about recently?) Before then I was mostly loading the m2ts file directly, or mpls if it was a branching title. (though I can't remember if that was OK for lip-sync or not, that was broken for a long time too, but it may have been prior to switching to LAV from Haali)

The bdmv support is as old as any Blu-ray support. Opening a mpls file has exactly the same effect as opening the bdmv file, except that it specifys which track to run, while bdmv runs the longest track.
In general, the demuxing is basically the same, the bdmv support only lets it figure out which m2ts to use.

I've been watching alot of BDs, all through bdmv playback, and never had any issues. :d
Its possible that some isolated BD titles have a odd playlist setup that could potentially cause issues, but i have no concrete bugs reports of anything like that.

[kalston]

I've been experiencing lip-sync errors, specifically with Blu-ray playback, when using the .bdmv file to start the film since before December though. (madVR 0.80 was released on the 18th) That's why I was skeptical about it being something introduced in 0.80

What I was wondering, was if Plutotype had also updated LAV Filters as well as madVR recently, or changed from loading with an mpls or m2ts file to the bdmv (which is by far the easiest thing to do) and thought that maybe that was causing the lip-sync errors as his playback was OK in 0.79 when mine was not. It could be entirely unrelated though. I'm definitely interested in hearing whether dropping to 0.79 fixes things for him.

I had no issues prior to using .bdmv files to start playback which I only started using recently when I heard that LAV Filters had support for it. (is that a recent addition, or just something I only found out about recently?) Before then I was mostly loading the m2ts file directly, or mpls if it was a branching title. (though I can't remember if that was OK for lip-sync or not, that was broken for a long time too, but it may have been prior to switching to LAV from Haali)

I'm not saying LAV Filters is definitely the cause, but that's why I was wondering how Plutotype was starting playback of the film, whether it was the bdmv, mpls or m2ts file. It's possible that since he was updating madVR he has also updated LAV filters at the same time, or changed from one filetype to another. (I have a habit of doing that, and sometimes MPC-HC & ffdshow too, which is really great for troubleshooting problems if they arise...)

I haven't had a chance to properly test things myself yet, as I haven't had much time to watch films lately, or do testing which involves waiting 20–30 minutes to see if sound goes out of sync. (which means I can't use the PC for anything else)

Either it's a problem with madVR that goes back further than 0.80, an issue with LAV Filters opening .bdmv (and possibly .mpls) files, or something else entirely. At least with non-branching titles, it's easy enough to just open the .m2ts directly which does not seem to have lip-sync problems for me.

The only reason I'm suspecting LAV here at all, is because it changes depending on which file I open. (or at least it seemed to from my limited tests)

This is odd, before using LAV I did have troubles playing Blu-rays where the film is split in several .mt2s (on top of my head : Inception, King Kong (2005) & Conan the Barbarian (1982) UK edition). In this situation I had to use the corresponding playlist file but it resulted in lip sync issues and a noticeable hiccup when the transition between the files happened. So I ended up ripping each one of them to a .mkv file.

But with LAV I've had no such issue and playback of all my Blu-rays is absolutely flawless.

[Razoola]

Are you only having lip-sync issues with Blu-ray? If so, are you starting the video with the "index.bdmv" file, or the .m2ts video file?

While using LAV Filters has enabled me to start a film using the index.bdmv file (previously it wouldn't work) I find that it causes lip-sync problems, when playing the .m2ts file directly does not, so I suspect it might be a LAV Filters issue rather than madVR. (but I haven't tested either in isolation so I couldn't say for sure)

What driver version are you using?

I'm definitely back to having problems getting smooth playback again, but then I am also on beta Nvidia drivers. I really should test to see whether it's the drivers or a change in madVR (by coincidence, new versions of madVR are often close to driver releases) but it's annoying either way, as I can't stick to WHQL drivers if I want to play the latest games.

I'm using 290.53 drivers but Im sure they are not the problem as going back to 079 fixes the issues for me.

[6233638]

Well I've just done about four hours of testing... and nothing is showing lip-sync errors for me now—including discs that were giving me problems before—with LAV Filters 0.44 (CUVID decoding) madVR 0.80 and MPC-HC 3895. (haven't been updating it nearly as frequently these days, don't know that it makes a difference when I'm only using external filters now) Can't explain it, it was definitely a repeatable issue back at the end of November.


I've also noticed that with the latest settings I've been trying out, I haven't had any presentation glitches yet after 45 minutes into the current clip. (just brought up the stats before pausing the film)

I have both CPU & GPU Queue size at 4, am using D3D11 for presentation, have 3 frames presented in advance in exclusive mode, and have "limit rendering times to avoid glitches" enabled. All other rendering options are at defaults.

That testing was all with Blu-ray at 24Hz, so I can't say it's definitely glitch-free playback until I try some 25@50 or 30@60. My guess is that they probably will show glitches.

For now, I've given up on entirely glitch-free playback, but at least films seem to be all right again, which is what I care about the most.


As for your problems Razoola, they definitely seem to be related to 0.80, and likely the "overshoot max frame latency" setting. Out of interest, what card are you using? That setting is intended for low-power HTPC stuff if I remember correctly.

[Plutotype]

What I was wondering, was if Plutotype had also updated LAV Filters as well as madVR recently, or changed from loading with an mpls or m2ts file to the bdmv (which is by far the easiest thing to do) and thought that maybe that was causing the lip-sync errors as his playback was OK in 0.79 when mine was not. It could be entirely unrelated though. I'm definitely interested in hearing whether dropping to 0.79 fixes things for him.
(or at least it seemed to from my limited tests)

Hi 6233638,
Thanks for your input. I update every LAVfilters release immediately as it gets released. Im not storing bdmv structures but always only the better version of the movie as one untouched m2ts file.
I hadnt lipsync issue for a year now, but when I noticed it yesterday and saw people also mentioning it here, so I asked if there were any changes to the last version 0.80 in this regard.

Im going to watch a 150min movie right now with LAVfilters 0.44 and 0.80, will report it here if any lipsync issues will be present during ( at the end ) of the movie.
Pluto

[6233638]

Just as an update on the above settings... finished watching a 75 minute 30p video at 60Hz there without any presentation glitches so it seems to be working for both 24@24 and 30@60. (and presumably also 25@50 as that's typically the same as 30@60) Might be worth a shot for anyone on Nvidia experiencing glitches with the latest drivers.

It seems so random as to what does/doesn't work depending on drivers/cards/madVR versions though, that I really don't know if it'll help.

[ADude]

- hardware accelerated video decoding (DXVA) is currently not supported

So is that still true ? (And so, the references to DXVA in the changelog only refer to assistance to the renderer and not decoding ?)

[fastplayer]

^ http://forum.doom9.org/showthread.php?p=1536640#post1536640

[ADude]

Thanks for the info in the link.

[starlight2]

Hi guys, I have a Nvidia GTS 450.
I tried to do MADNV12test, the result ':

Windows 7 x64 Service Pack 1
NVIDIA GeForce GTS 450

D3D9 Surface StretchRect:
D3DFMT_R8G8B8: creating GPU texture failed
D3DFMT_A8R8G8B8: lossy (0-0)
D3DFMT_X8R8G8B8: lossy (0-0)
D3DFMT_A8B8G8R8: creating GPU texture failed
D3DFMT_X8B8G8R8: creating GPU texture failed
D3DFMT_A2R10G10B10: lossy (0-0)
D3DFMT_A2B10G10R10: lossy (0-0)
D3DFMT_A16B16G16R16: StretchRect failed
D3DFMT_A16B16G16R16F: lossy (0-0)
D3DFMT_A32B32G32R32F: lossy (0-0)

D3D9 Surface VideoProcessor:
D3DFMT_R8G8B8: creating GPU texture failed
D3DFMT_A8R8G8B8: lossy (0-0)
D3DFMT_X8R8G8B8: lossy (0-0)
D3DFMT_A8B8G8R8: creating GPU texture failed
D3DFMT_X8B8G8R8: creating GPU texture failed
D3DFMT_A2R10G10B10: lossy (0-0)
D3DFMT_A2B10G10R10: lossy (0-0)
D3DFMT_A16B16G16R16: lossy (0-0)
D3DFMT_A16B16G16R16F: lossy (0-0)
D3DFMT_A32B32G32R32F: lossy (0-0)

DXVA Surface StretchRect:
D3DFMT_R8G8B8: creating GPU texture failed
D3DFMT_A8R8G8B8: lossy (0-0)
D3DFMT_X8R8G8B8: lossy (0-0)
D3DFMT_A8B8G8R8: creating GPU texture failed
D3DFMT_X8B8G8R8: creating GPU texture failed
D3DFMT_A2R10G10B10: lossy (0-0)
D3DFMT_A2B10G10R10: lossy (0-0)
D3DFMT_A16B16G16R16: StretchRect failed
D3DFMT_A16B16G16R16F: lossy (0-0)
D3DFMT_A32B32G32R32F: lossy (0-0)

DXVA Surface VideoProcessor:
D3DFMT_R8G8B8: creating GPU texture failed
D3DFMT_A8R8G8B8: lossy (0-0)
D3DFMT_X8R8G8B8: lossy (0-0)
D3DFMT_A8B8G8R8: creating GPU texture failed
D3DFMT_X8B8G8R8: creating GPU texture failed
D3DFMT_A2R10G10B10: lossy (0-0)
D3DFMT_A2B10G10R10: lossy (0-0)
D3DFMT_A16B16G16R16: lossy (0-0)
D3DFMT_A16B16G16R16F: lossy (0-0)
D3DFMT_A32B32G32R32F: lossy (0-0)

D3D9 Surface speed test:
NV12: upload 1083 fps, download 918 fps, trick download failed
YV12: upload 1166 fps, download 987 fps, trick download failed
A8R8G8B8: upload 314 fps, download 265 fps, trick download failed

DXVA Surface speed test:
NV12: upload 1209 fps, download 1079 fps, trick download failed
YV12: upload 1166 fps, download 1008 fps, trick download failed
A8R8G8B8: upload 317 fps, download 237 fps, trick download failed

A8R8G8B8 Texture speed test:
default: upload 213 fps, download 306 fps
dynamic: upload 284 fps, download 26 fps, trick download 210 fps





Help me please? I'm not sure what to do and what is this procedure.
Thank you for your help.
Thanks

[Mangix]

that nv12 test is a bit outdated. don't worry about it at all.

[starlight2]

Thanks
then the bug of the levels video has been resolved with nvidia?

[Remicade]

If I have ATI the only solution for DXVA in madVR is Cyberlink Video Decoder ? With HAM or DXVA ?

[Andy o]

If it even works well (never did for me). With DXVA deinterlacing though, you can use almost any modern and even old-ish CPU to decode. What do you have?

[Mangix]

Thanks
then the bug of the levels video has been resolved with nvidia?

IIRC, that test was made by madshi to test how good nvidia and amd hardware were at handling nv12 at different texture formats. like i said, nothing to worry about since madshi does not use nv12(I assume it gets converted to YV12) to do any processing. nv12 is however required for proper DXVA video decoding with deinterlacing(unless i missed something).

anyways, nothing to worry about.

[Remicade]

Andy it's works very well no artifacts, I watch only progressive material, I will compare the quality of Cyberlink decoder versus MPC HC. Somebody in another forum told me Cyberlink was a litle bad in image quality in his system.

[Mangix]

If it even works well (never did for me). With DXVA deinterlacing though, you can use almost any modern and even old-ish CPU to decode. What do you have?

i have an old computer with a radeon hd2400 pro and a 2.6ghz pentium 4. i will tell you that it drops frames on 480i content when DXVA deinterlacing is used on XP but works smooth when IVTC is applied(software i believe). so not always smooth.

[italospain]

Hi madshi

here a 3 examples of my NTSC DVD Library where i am not able to IVTC with madVR or DScaler mod.

https://rapidshare.com/files/576191865/Examples.zip

greetings

[LigH]

Without downloading your sample, just a remark: Sometimes NTSC is indeed plain interlaced NTSC-TV...

[italospain]

Without downloading your sample, just a remark: Sometimes NTSC is indeed plain interlaced NTSC-TV...

all 3 examples are classic cinema movies shot in 35mm

[Andy o]

Andy it's works very well no artifacts, I watch only progressive material, I will compare the quality of Cyberlink decoder versus MPC HC. Somebody in another forum told me Cyberlink was a litle bad in image quality in his system.

I'd be skeptical about that claim. Decoders usually don't mess with quality that way unless they're broken, and if they are, AFAIK it's very obvious stuff like blocking or horrible color shifts.

[Andy o]

i have an old computer with a radeon hd2400 pro and a 2.6ghz pentium 4. i will tell you that it drops frames on 480i content when DXVA deinterlacing is used on XP but works smooth when IVTC is applied(software i believe). so not always smooth.

I think on that machine the CPU and/or GPU are probably the bottlenecks.

[mindbomb]

I have some performance problems, maybe someone can clear up what is going on.

Using mpc hc 3917 as the player and to display subtitles. Using lav video with cuvid mode as the video decoder. MadVr .80 as the video renderer.

graphics card is an 256mb 8600gt.


Everything is perfect, except when you ask the player to display vobsub subtitles, and then all the queues dramatically fall, and the playback becomes very poor.
This doesn't happen with text or bluray subtitles.

Does anyone know what is possibly going on?

[golagoda]

I have some performance problems, maybe someone can clear up what is going on.

Using mpc hc 3917 as the player and to display subtitles. Using lav video with cuvid mode as the video decoder. MadVr .80 as the video renderer.

graphics card is an 256mb 8600gt.


Everything is perfect, except when you ask the player to display vobsub subtitles, and then all the queues dramatically fall, and the playback becomes very poor.
This doesn't happen with text or bluray subtitles.

Does anyone know what is possibly going on?
Make sure you don't have any custom settings in the NVIDIA Control Panel for MPC-HC or globally setting anti-aliasing to anything but application-controlled (as all it will do is thrash your CPU and GPU with no real benefit).

If you don't have anything like that you can probably try DirectVobSub instead of the internal subtitle renderer (I recommend using xy-vsfilter (https://code.google.com/p/xy-vsfilter/))

[mindbomb]

oh wait, i spoke too soon, it affects all image subs, but text subs are unaffected.

you know what, forget everything, cause i can't even nail down what is going on.

[mikewando]

Is there any way the deinterlacing and IVTC used in madVR could be leveraged for encoding?

I've been thinking about this a little more, all it would really take is a directshow filter very much like madVR, but with an output pin that could be used in a graph for DirectShowSource in an avisynth script. I've never written a directshow filter, so I have no idea how trivial writing this from scratch or adapting madVR would be, but http://msdn.microsoft.com/en-us/library/cc307941.aspx certainly makes it seem possible with DXVA 2.

At any rate, I still think this would be immensely helpful for speeding up the encoding process, so if such a filter is feasible, there is at least some demand for it.

[6233638]

Another deinterlacing sample. This is a weird one, don't know if it will be useful or not. The film itself was perfect, staying in 2:2, but when the credits came up, it went to unknown cadence, and then switched to 4:2:2. (which is wrong)

http://www.mediafire.com/?l85rwl1ss2cghv5

Just thinking out loud, something that would be nice would be the option to have IVTC (film mode) deinterlacing enabled by default—anything I have that requires deinterlacing, would make use of the IVTC mode.

The option to log timestamps for cadence changes would be great too. If I'm in the middle of watching a film, I'm not going to stop it and make a note of when it is (as I have to leave exclusive mode to see the time) unless it stays in the wrong cadence for a long time. There was one point in a film was watching earlier where it switched to a weird (wrong) cadence briefly and I can't seem to find it now as the film has a lot of similar/repeating shots. (though I suspect this one will be fixed with the improved logic in an upcoming version)

[Mangix]

I think on that machine the CPU and/or GPU are probably the bottlenecks.

it's the gpu. piece of ****. have many problems with it.

[Andy o]

Man, even at the time the 2400pro was weaaaak. I remember I started really getting into the HTPC thing with that generation of ATI cards, but got a 2600pro after the 2400pro and XT were deemed too weak.

[Mangix]

i read that it was equivalent in performance to an 8500 gt from nvidia. i find that hard to believe having owned one since i could run madVR at decent settings just fine.

anyways, sorry about going a bit off-topic.

[e-t172]

Just curious: why isn't madVR compatible with "native" DXVA decoders (e.g. ffdshow DXVA, MPC DXVA)? Why is EVR the only renderer that can do this? Is there some kind of fundamental problem or is it just that madshi hasn't got to it yet?

[nevcairiel]

I'm sure it would be possible, he just hasn't done it yet.

[CruNcher]

DXVA + Custom Shader support in MadVR would rock the boxors ahh yeah and a nice Realtime Sync Graph ;)

[Remicade]

LAV Video Decoder ATI DXVA works in madVR...

[Thunderbolt8]

^^ does it support VC-1 and interlaced VC-1?

interlaced vc-1 is really the last pain in the ass stuff I cannot play properly with madvr (high bitrate BBC docs etc.)

[Paladin77]

^^ does it support VC-1 and interlaced VC-1?

interlaced vc-1 is really the last pain in the ass stuff I cannot play properly with madvr (high bitrate BBC docs etc.)

http://forum.doom9.org/showpost.php?p=1551231&postcount=8147

[e-t172]

LAV Video Decoder ATI DXVA works in madVR...

Yeah, but unless I'm missing something, LAV Video DXVA copies GPU-generated frames back into CPU RAM, while pure DXVA doesn't, so the latter is potentially faster.

[iop]

When the possibility to take screenshots with alt+i with MPC will be added?

[pirlouy]

I'm sure it would be possible, he just hasn't done it yet.
I don't know what happens recently, but for some weeks, DXVA is everywhere. CUVID, QuickSync, threads comparing "hardware" solution. I've read Potplayer offers DXVA solution. Now, you betatest a new DXVA decoder.
I think Madshi will have to change his roadmap to follow the mode (or rather developers interest right now). :D

For those like me who does not know a lot of things: if madshi were to implement DXVA, does pure DXVA decoders (like those offered by MPC-HC) could improve quality, or software solution + shaders by madVR will always offer the best quality ?

Because right now, I don't understand all this interest in hardware solutions now that we have really powerful CPU, and software solutions seem to be more stable than hardware ones (with their buggy drivers)...

[bjd]

LAV Video DXVA copies GPU-generated frames back into CPU RAM

if madshi were to implement DXVA

I am not 100% certain on this but in theory if this was implemented on the madvr decoders there would be no need to send the decoded frames back to the CPU memory, they would just be added to the renderer queue on the GPU memory which should offer better performance.

[dansrfe]

Bug: When wallpaper automatically changes (e.g. every 5 mins), multiple frames are dropped in exclusive mode.

Bug: Refresh/reset of exclusive mode when changing player to other display takes many minutes.

Feature Request: Support for IVTC and according refresh rate changes.

Feature Request: Hotkey combination to turn off refresh rate changer for current instance. (Useful for switching screens for a short while and don't want to wait for screen resolution and refresh rate to change). In my case I have a screen that has to change from 2048 x 1152 to 1080p in order to go to 24Hz.

IVTC/Video Deinterlacement works very well for me.

[e-t172]

I am not 100% certain on this but in theory if this was implemented on the madvr decoders there would be no need to send the decoded frames back to the CPU memory, they would just be added to the renderer queue on the GPU memory which should offer better performance.

With a native DXVA decoder and a native DXVA renderer such as EVR, frames are never transferred back to CPU memory.

[Remicade]

Because right now, I don't understand all this interest in hardware solutions now that we have really powerful CPU, and software solutions seem to be more stable than hardware ones (with their buggy drivers)...
I have Phenom II 920 if I use madVR alone without DXVA (LAV or Cyberlink HAM) CPU is ~ 30% (2800 Mhz) and the cooler is quite noisy. If I enable DXVA the noise is very low and the CPU works at 800-1600 Mhz. I want madVR (+Reclock) for me it's the same quality of picture as EVR but the motion it's more fluid.

[Razoola]

An update on my problem with 080. It seems my dropped frame and queue issues only affect 25hz material on a secondary display at 50hz (720p TV rips). 23.976 material on a 23.976 secondary display is fine.

The only fix I have found is to enable the treat 25p material as 24p (requires reclock).

[Gser]

For those like me who does not know a lot of things: if madshi were to implement DXVA, does pure DXVA decoders (like those offered by MPC-HC) could improve quality, or software solution + shaders by madVR will always offer the best quality ?

Because right now, I don't understand all this interest in hardware solutions now that we have really powerful CPU, and software solutions seem to be more stable than hardware ones (with their buggy drivers)...

Software decoders support higher profile levels and are more flexible feature wise. All "properly" implemented decoders produce the same picture hence DXVA's popularity is only a fad and in MadVRs case a completely useless feature since MadVR needs a very good GPU to run properly - users that tend to have a good GPU also tend to have a good CPU which is more than capable decoding h.264. DXVA is only useful on very old PC's or some mobile devices non of which can use MadVR - it would even reduce performance on crappy GPUs.

[entrecour]

I'd appreciate folks advice on the optimum settings in NVIDIA Control Panel when using LAV CUVID with madvr or Custom EVR. I have searched the forum and found bits and pieces, but I wanted to pull it altogether to make sure I haven't missed anything.

This is what I have now, using a preset in 3D Settings, with my comments in italics.

NVIDIA Control Panel

3D Settings

Adjust Image settings with preview
- let 3D application decide
- Use the advanced 3D image settings
- use my preference emphasizing:Quality /Performance

Manage 3D settings
- Global settings / Program settings
Ambient occlusion - off
Anisotropic filtering - application controlled
Antialiasing - gamma correction - off
Antialising - mode - application controlled
Antialiasing -setting - application controlled
Antialiasing - Transparency - Off
CUDA -GPU's - All
Maximum prerendered frames - 3
Multi-display/mixed GPU acceleration - Single display performance
Texture filtering - Ansiotropic sample optimization - Off
Texture filtering - Negative LOD bias - Clamp
Texture filtering - Quality - Quality
Threaded Optimization - Auto
Triple buffering - Off
Vertical sync - use 3d application setting

Set PhysXconfig - Auto-select

Display

- Change resolution - user hardware dependent
- Adjust desktop color settings
2. Choose how color is set. Use NVIDIA settings: Defaults
3. Apply following enhancements Digital vibrance: +50% Hue: +0 (defaults)
- Adjust desktop size and position, Scaling - No scaling (in my case I let TV scale), Deinterlacing Use inverse telecine
Video

- Adjust video color settings - With the NVIDIA settings - Color - defaults, Gamma - Defaults, Advanced, Dynamic range: Full 0-255 if using monitor or TV can be set to 0-255, if not Limited 16-235
- Adjust video image settings - Edge enhancement - use the NVIDIA setting 0%, noise reduction - use the NVIDIA setting 0%


Some specific questions

1. Is max prerendered frames OK left at 3?
2. Do you recommend leaving Digital Vibrance at the default (+50%)?
3. Should I specifically set all Antialiasing settings to OFF or can I leave them as "application controlled"

I use both mpc-hc / madvr and DVBViewer / Custom EVR.

By the way I am have a 8600GT / 512M with the 290.36 driver on Vista 32bit connected to a TV via DVI / HDMI cable if that makes any difference.

:thanks:

[Budtz]

Yes I also feel at bit in the dark with regards to setting Nvidia settings for optimum PQ.

[leeperry]

As much as switching refresh rates on the fly is usually OK during playback, doing so in pause mode usually ends up in PotP locked up hard here on XPSP3...it would be great if that wasn't the case :)