madVR - high quality video renderer (GPU assisted)

[MariaX9]

Aww man, put the deringer on already, it's one of the most mind-blowing features I've seen in mVR to the point that I'm wondering how I could watch anything before this popped up http://forum-images.hardware.fr/images/perso/1/anefay.gif

Wasen`t Deringing producing some artifacts with NGU? And ppl recommended it to turn off with NGU?

[ryrynz]

Whenever I'm creating or deleting a profile the scroll bar resets back to the stop, madshi can you fix this, also that one character input bug (http://bugs.madshi.net/view.php?id=406) has been a pain in my ass for about a year now..
Would've liked for simple things like to to be fixed before upgrading to 0.91.7 but oh well, Anyway I'm doing this finally on the HTPC so I can give NGU AA a proper test with my setup.

4) NNEDI3 256-taps is better than NGU Anti-Alias Very High for chroma upscaling.

TBH all I'm seeing here is sharpness difference here, not anything to do with quality. I'm interested to see if NGU soft/AA can rival super-xbr 100-125 though.

Now that I've had a quick play with NGU AA on the HTPC, I'm actually preferring it's more natural/balanced appearance at least in combination with my Avisynth profile during viewing.

Had to add some adaptive sharpen to balance things out but the overall result is quite nice, I'm now using very high for my anime 480 profile and medium for my 720.

Madshi, would you consider releasing anything in between high and very high in future, there appears to be rather big jump between them in performance and appearance.
I'm also interested in looking at possibly any alternative builds of the very high setting, not sure if there are any tangents that can be looked at though.

[Warner306]

Im upscaling BDs to 4K with madvr and want to use a small amount of sharpening since I blow the image up to a large 145" scope dia screen.
Where is the best place to use sharpening, under Processing > Image enhancements or in Upscaling refinement?

Also is it advisable to use AR, does it add any softness or artefacts?
If so where is the best place to add it, Image enhancements or upscaling refinement?
As there are a number of sections where AR is, do you add it more than once or once only from one sectioned enhancement ?

Use image enhancements on 1080p content and upscaling refinement on content that needs to be upscaled.

[Neo-XP]

TBH all I'm seeing here is sharpness difference here, not anything to do with quality. I'm interested to see if NGU soft/AA can rival super-xbr 100-125 though.


original (extracted by MPC-BE) :

https://s30.postimg.org/z2epnpkfh/original.png (https://postimg.org/image/z2epnpkfh/)

super-xbr 100 / 125 / 150 :

https://s21.postimg.org/vhr0bt5ub/super_xbr100.png (https://postimg.org/image/vhr0bt5ub/) https://s21.postimg.org/3vo8r4mhf/super_xbr125.png (https://postimg.org/image/3vo8r4mhf/) https://s21.postimg.org/5c4cn0es3/super_xbr150.png (https://postimg.org/image/5c4cn0es3/)

NGU AA low / med / high / veryhigh :

https://s21.postimg.org/ejrzpywgj/NGU_low.png (https://postimg.org/image/ejrzpywgj/) https://s21.postimg.org/g03i83zdf/NGU_med.png (https://postimg.org/image/g03i83zdf/) https://s21.postimg.org/wxxxg7jdf/NGU_high.png (https://postimg.org/image/wxxxg7jdf/) https://s21.postimg.org/8y5klwvrn/NGU_veryhigh.png (https://postimg.org/image/8y5klwvrn/)

Nnedi 16 / 32 / 64 / 128 / 256 :

https://s21.postimg.org/jz0pqxo0j/Nnedi16.png (https://postimg.org/image/jz0pqxo0j/) https://s21.postimg.org/svbhuvemr/Nnedi32.png (https://postimg.org/image/svbhuvemr/) https://s21.postimg.org/cycpy5m8j/Nnedi64.png (https://postimg.org/image/cycpy5m8j/) https://s21.postimg.org/bkl32umz7/Nnedi128.png (https://postimg.org/image/bkl32umz7/) https://s21.postimg.org/cbdt8mpcj/Nnedi256.png (https://postimg.org/image/cbdt8mpcj/)

super-xbr is not very good unfortunately compared to NGU AA and Nnedi.
Nnedi seems to be more precise (but not necessarily sharper) and hides artifacts better (for instance the artifacts under the black letters) than NGU AA. It also looks cleaner overall.

[nevcairiel]

Using an RGB image as an "original" when the source is clearly 4:2:0 seems to be a bad way to test chroma upsampling.

[steakhutzeee]

Hi, what about if i want to learn how to set properly Madvr? I'm on an Intel i5 4590, MSI R9 270X 2GB and 8GB RAM.

[Neo-XP]

Using an RGB image as an "original" when the source is clearly 4:2:0 seems to be a bad way to test chroma upsampling.

The comparison was not done with the original (extracted by MPC-BE) image of course, I just don't know how to extract properly the source image for you to test if necessary, or if it is even possible, but whatever :D

I guess that a better test would be to have an original RGB image, encode it to 4:2:0, and then compare the result after chroma upscaling with madVR with the original. The closest to the original would win.

Anyway, I would use NGU AA in next versions for chroma upscaling if NNEDI3 is removed. It would also be nice to have more options for chroma doubling (Lanczos3 AR or Jinc AR).

This is what I've found to be the best for chroma doubling : Bicubic60 AR < Spline64 AR < Spline36 AR < Lanczos4 AR < Lanczos3 AR < Jinc AR

[70MM]

For Chroma Upscaling is it better to use NEEDI3 at 16 or Super Xbr at 150?

[steakhutzeee]

What about i want to learn how to properly configure Madvr last version? I'm on Intel i5 4590; an MSI R9 270X 2GB and 8GB ram.

[70MM]

Use image enhancements on 1080p content and upscaling refinement on content that needs to be upscaled.

So in the area of Upscaling Refinement is has absolutely no effect on 1080p content?

[Asmodian]

So in the area of Upscaling Refinement is has absolutely no effect on 1080p content?

Since you are watching at 4K they would because you are also upscaling 1080p. Based on Warner306's comment you should use Upscaling Refinement for everything but 4K content.

I much prefer sharpening with Upscaling Refinement as opposed to Image Enhancement but the effect is not as strong with the same settings.

[Warner306]

So in the area of Upscaling Refinement is has absolutely no effect on 1080p content?

Yes, that's correct. But only if you are outputting at 1080p. 1080p -> 2160p involves upscaling, so you'll want to use upscaling refinement.

[Asmodian]

Im upscaling BDs to 4K with madvr :devil:

[70MM]

Yes, that's correct. But only if you are outputting at 1080p. 1080p -> 2160p involves upscaling, so you'll want to use upscaling refinement.
Ok I understand, I am upscaling 1080p >2160p so Upscaling refinement is actually working for me.

What I want to know then is, should I use Image enhancements or Upscaling refinement if I want to add a bit of extra sharpness?

[Oguignant]

Hi. Can someone explain to me how "disable GPU gamma ramps" option works? I have 2 separate calibrations. One for madvr (3d lut file) and one for windows (matrix icc v4 file). If I enable the "disable GPU gamma ramps" option, does madvr ignore the windows calibration?

[Neo-XP]

Is there any way to use NGU Sharp (med) for chroma doubling with NGU Sharp (high) for luma doubling ?
When I select "very high" for chroma doubling, it doesn't let me when "high" is selected for luma doubling.

I try to do this :

artifact removal : reduce banding artifacts medium & high
chroma upscaling : NGU Anti-Alias (high)
image downscaling : Bicubic150 + AR [relaxed]
luma doubling : NGU Sharp (high) 2x supersampling
chroma doubling : NGU Sharp (med) 2x supersampling
image upscaling : Jinc + AR
upscaling refinement : add grain 3
dithering : Ordered + colored noise + change dither

I don't think that NGU Sharp (med) for chroma doubling is a bad choice considering the rest of the configuration.
The limit for chroma doubling should be equal or at maximum one step below the luma doubling IMO, not two...

@madshi ? any chance to change this ?

[ryrynz]

What I want to know then is, should I use Image enhancements or Upscaling refinement if I want to add a bit of extra sharpness?

I honestly don't understand why you're even asking this. Just tick boxes and watch.

[valdeski]

I remember some people saying image enhancements are better done before any upscaling. I don't know if thats really true.

[70MM]

I honestly don't understand why you're even asking this. Just tick boxes and watch.

Hey I have done that many times with hours and hours of experimentation. I am new with madvr and have less experience than you and others knowing and understanding all the settings here. If I dont ask the experts how can I learn or one day try to help new comers like me.???

Im sorry you are so upset with me asking a question!

[70MM]

I remember some people saying image enhancements are better done before any upscaling. I don't know if thats really true.
Thank you valdeski, hopefully some others who have the knowledge and dont mind sharing the knowledge will answer this for us.

[Warner306]

Ok I understand, I am upscaling 1080p >2160p so Upscaling refinement is actually working for me.

What I want to know then is, should I use Image enhancements or Upscaling refinement if I want to add a bit of extra sharpness?

I'd use NGU Sharp, as it doesn't require added sharpening. If I wanted to add sharpening, I'd use SuperRes under upscaling refinement. But it isn't necessary with NGU Sharp.

[YGPMOLE]

if the PC freezes you should find out what freezes event viewer is a good start for example.

Assuming that the Seasonic 650W Gold PSU is enough powerfull (the PC doesn't shut down), and due to the fact that the freezing happens when "very High" is selected instead of "High", I can reasonably say that is a GPU related issue: or temperature or consumption. The fan speed/noise points to overheating.

your GPU is not supposed to over heat and it is pretty hard to over heat a RX 4X0 card.

So it drives back to what I read about the RX 480 problem of to much current consumption over PCI for some old Mobo.

if it is the GPU/GPU driver try to contact AMD per phone that usally works some how.

AMD said that they solved the problem starting from the 17.1 Crimson Driver.

and i never said madVR is freezing?

I was only confirming what you said in the below quotation, not accusing you...

it is very unlikely that it is madVR that is freezing.

[YGPMOLE]

Try running SVP in CPU mode and see if it makes a difference.

SVP it's already without GPU acceleration, thanks for suggesting it.

I think that SVP + madVR NGU HIGH is the max that I can use for low res (480/576) -> 1080, but picture quality is amazing and some artifacts depends only from the frame rate conversion. so I really thank Madshi for his works!!!

[raymondjpg]

Using latest versions of madVR 89.19 and Emby for WMC (Windows 7 Professional). With MPC-HC 1.7.10 after refresh rate switching by madVR (windowed overlay mode) then video hangs and does not play. Video will play after closing then re-starting MPC.

Refresh rate switching when using MPC-HC 1.7.9 works with video playing fine after refresh rate switching.

Has anyone else seen this, better still have a solution for it? I have also posted this in the MPC-HC forum as I suspect it is more of an issue with the latest version of MPC-HC than madVR.
Hi. I reported this issue with the release of MPC-HC 1.7.10 about a year ago, and it has recurred with the recent release of MPC-HC 1.7.11 (using the latest release of madVR 0.91.7).

The issue is not apparent when using the Enhanced Video Renderer (custom presenter), and MPC-HC's internal Fullscreen options to switch video refresh rates.

It thus looks like the issue is limited to when madVR is used as the renderer in MPC-HC.

Does anyone have a solution for this?

[ryrynz]

super-xbr is not very good unfortunately compared to NGU AA and Nnedi.


My own testing shows different results, while I am a fan of NNEDI3 for chroma it's performance impact for it isn't what I'm prepared to pay. Also I've seen super-xbr come off sharper in situations than NNEDI3 which I prefer the look of as well, so I've drawn my own conclusions with this after many hours and hundreds of screenshots of inspection, I know exactly what I'm getting.

[Ditto666]

Hey guys, so I'm a little confused about how switching on some of the functions behave, in particular upscaling for 4K, anti-ringing for SSIM, and 200% super sampling for NNEDI3.

To begin with, I should mention that my source is a 4K video being played back on a 1080p display. If it matters, it comes from a UHD source which contains HDR merged/converted using pixel shader math.


So for the first issue, I'm confused why changing any of the upscaling options as any effect AT ALL when the only possible configuration I could obviously have of the video is downscaling.

For the second issue, how in the world would the anti-ringing filter when applied to anything (in this case only to the SSIM algorithm) result in the image being sharper? In this case, it's a significant difference too.

Lastly, never mind that it shouldn't do anything anyway given the mentioned setup, why does super sampling completely screw up and otherwise also have the opposite effect of blurring the image if using the NNEDI3 algorithm?


Thanks in advance to anyone who may have a clue XD

PS- I've had this question in mind for a while and was wondering if there was any real answer to it. With pretty much any scaling or other related options (excluding any image enhancements or sharpness refinement of course), do any of them actually add sharpness (as in contrast that's added between line edges) to the picture or in each case does it simply bring out more/less of the original image in its original resolution closer to if it were like a vector object? If it's the latter, would it be accurate to say that the sharper the image becomes, the better/more accurate it is, where it then just becomes a matter of lowering the sharpness on your display? I get that there can be parallel differences depending on the algorithm where increases in sharpness may be in different areas, but that aside... Obviously if I max everything out, the image appears to be overly sharp but, again, if the previous statement is accurate, this I guess would simply be because the picture was overcompensated for prior to the optimized/more accurate rendering provided by MadVR.

[Neo-XP]

My own testing shows different results, while I am a fan of NNEDI3 for chroma it's performance impact for it isn't what I'm prepared to pay. Also I've seen super-xbr come off sharper in situations than NNEDI3 which I prefer the look of as well, so I've drawn my own conclusions with this after many hours and hundreds of screenshots of inspection, I know exactly what I'm getting.

Did you test this with anime content ? NNEDI3 looks more natural for me when testing on films.

[webs0r]

So for the first issue, I'm confused why changing any of the upscaling options as any effect AT ALL when the only possible configuration I could obviously have of the video is downscaling.

Chroma upscaling will always apply (typically the chroma channel is supplied at a lower resolution than luma). madVR scales that first to match the luma.
Image upscaling will only apply in this case if you've asked it to always supersample (I assume) and that's because you've told it to do so (hence the option called "always - supersampling").


For the second issue, how in the world would the anti-ringing filter when applied to anything (in this case only to the SSIM algorithm) result in the image being sharper? In this case, it's a significant difference too.

I don't see that happen in my case.


Lastly, never mind that it shouldn't do anything anyway given the mentioned setup, why does super sampling completely screw up and otherwise also have the opposite effect of blurring the image if using the NNEDI3 algorithm?[/B]

If you are supersampling 2k vertical to 4k, then down to 1k... I would seriously question the need to further upscale it before downscaling.
I believe what you are seeing is the interaction of how NNEDI "fills in the gaps" so to speak, then the downscaling algo interacting with NNEDI after that. The additional gap fill in seems to give it a better "anti-aliased" look, which you may percieve as a decrease in sharpness. But there is no detail loss. And I wouldn't call it blurring the image. I expect in your test you are looking at scene with high contrast edges only, or a test pattern?

With pretty much any scaling or other related options (excluding any image enhancements or sharpness refinement of course), do any of them actually add sharpness (as in contrast that's added between line edges) to the picture or in each case does it simply bring out more/less of the original image in its original resolution closer to if it were like a vector object?
The whole point of the different algos for upscaling, is to try to interpolate, if this was the original image size, what was there in those missing pixels? i.e. your vector analogy.
The algos result in different levels of perceived sharpness.
There's no way to tell what it really should be (i.e. accuracy), unless you are doing a comparison with a "groundtruth", that is downscaled, then upscaled. Then it introduces a dependency on the downscaling algo anyway, so its great for your controlled test but on other sources, who knows??
Downscaling might be a different affair but I don't know too much about this stuff.
Sharpness is not accuracy itself.

I think a reason why madVR users appreciate the choices is that everyone can choose an option that suits their taste in sharpess, detail retention (or interpolation) and real-time performance, whilst considering the input quality of their sources.

[Ditto666]

webs0r,

For upscaling, I wasn't talking about the chroma channel.

For the anti-ringing filter, were you trying it out with a 4k video?

Regarding NNEDI, no, it has the opposite effect on it vs. any other algorithm, and it actually does appear as a loss in detail/accuracy. Also, I mentioned that I was playing/testing this with a native 4K video, and it wasn't a test pattern; it was a demo nature video used by Samsung to showcase their Quantum Dot displays. I was watching it on a 1080p display though.

With respect to what upscaling is intended for, well, that's what I already knew and I guess the further logic of what you said makes sense. I guess I was wondering what someone really knowledgeable would make of what I further said. My main intention in the end would really just be to retain as much detail as possible without LOSING any sharpness. I would imagine there's a somewhat objective way of finding which options together give you that best.

[dioxholster]

are my setting okay?:

https://i.imgur.com/atx8WRX.png

[leeperry]

Wasn't Deringing producing some artifacts with NGU? And ppl recommended to turn it off with NGU?

Not as far as I know? Anyway I keep it on at all times, not going to turn if off EVER.

[webs0r]

webs0r,

For upscaling, I wasn't talking about the chroma channel.

For the anti-ringing filter, were you trying it out with a 4k video?

Regarding NNEDI, no, it has the opposite affect on it vs. any other algorithm, and it actually does appear as a loss in detail/accuracy. Also, I mentioned that I was playing/testing this with a native 4K video, and it wasn't a test pattern; it was a demo nature video used by Samsung to showcase their Quantum Dot displays. I was watching it on a 1080p display though.

With respect to what upscaling is intended for, well, that's what I already knew and I guess the further logic of what you said makes sense. I guess I was wondering what someone really knowledgeable would make of what I further said. My main intention in the end would really just be to retain as much detail as possible without LOSING any sharpness. I would imagine there's a somewhat objective way of finding which options together give you that best.

Re: upscaling; i covered both chroma and luma. You were telling it to always upscale luma; correct? And then asked why it was upscaling when it didn't need to?

On the anti-ringing - yes I was testing with 4k video. Hard to spot, but when I noticed changed parts, AR on was definitely reducing "apparent" sharpness/haloing.

On the NNEDI I'm confused about what you are asking? The opposite effect of what?
If I understand the way you put it, you are upscaling 2160 vertical lines to ~4000 via NNEDI, then downscaling to 1080 with SSIM 2D (no AR) - at least that was my test . And you're saying it causes loss of fine details.
It's possible of course, but I only saw what I saw in my quick test, which was finer shading between aliased edges.
Probably need screen shots.

Tbh I found all of these super hard to spot! There really are diminishing returns.

[Ditto666]

Re: upscaling; i covered both chroma and luma. You were telling it to always upscale luma; correct? And then asked why it was upscaling when it didn't need to?

On the anti-ringing - yes I was testing with 4k video. Hard to spot, but when I noticed changed parts, AR on was definitely reducing "apparent" sharpness/haloing.

On the NNEDI I'm confused about what you are asking? The opposite effect of what?
If I understand the way you put it, you are upscaling 2160 vertical lines to ~4000 via NNEDI, then downscaling to 1080 with SSIM 2D (no AR) - at least that was my test . And you're saying it causes loss of fine details.
It's possible of course, but I only saw what I saw in my quick test, which was finer shading between aliased edges.
Probably need screen shots.

Tbh I found all of these super hard to spot! There really are diminishing returns.

Ok, well the point isn't to test what I said to find contradictory results. I was asking effectively the devs if they knew why what I mentioned would be happening. Here's the video I'm using btw: http://4kmedia.org/ultra-hd-hdr-samsung-4k-demo-wonderland/

Also, how did I set it to always upscale? What's the option that does that? That could explain it. All I mentioned was that I have a 4K video being played back on a 1080p display and was confused why the non-chroma upscaling options were having any effect (and I guess you answered why chroma may be affected, though, damn how much lower of a resolution could it be at?).

By the opposite effect, I meant that turning it on does the inverse of what it appears to do on anything else. You're only confused cause you're not getting the same results. I mentioned what happens; it definitely is; no need to question me on it. I can't imagine how you'd know the reasons why this would be the case that it's happening and only the devs would know. Hopefully it doesn't just get buried or is otherwise perceived as already having been answered... In any case, you can try it with the video I was actually playing.

[leeperry]

On the anti-ringing - yes I was testing with 4k video. Hard to spot, but when I noticed changed parts, AR on was definitely reducing "apparent" sharpness/haloing.
I haven't followed the whole story but AR is not a free-lunch feature IME, as much as I like it on SR2 LL I far prefer having it disabled on SSIM 2D 100% LL downscaling. I tried to switch both but still preferred having it on SR alone. What matters is finding the best looking combination to your eyes, there's hardly any one-size-fits-all feature that doesn't come with side effects in mVR so it's all about compromises. Of course I got the "artifacts removal" deringer forced so this might explain :sly:

[ryrynz]

Just changed my chroma upscaler from Super-xbr to NGU AA High. Very clean and gives NNEDI3 64+ a run for it's money which were typically pretty pointless basically because of the rendering load and small differences over 32 neurons.

NGU AA Very High isn't worth it for the rendering time increase based on the comparison I've done (over twice the rendering time on my 960) It seems to have a different profile to the other NGU versions resulting in a fairly minor change & not necessarily improved changes. madshi perhaps you could work something better suited for chroma as a very high setting?

[enphenate]

My MPC-HC setup has become extremely unstable after upgrading from 0.91.5 to 0.91.7.

Whenever I start a new video and double click it to make it fullscreen, the video glitches out, freezes and becomes unresponsive. The audio will keep playing, then cuts out. If any windows prompt (UAC) opens up, my entire screen goes black and the mouse cursor just flashes really quickly. Can't do anything at this point, the computer is not responsive to any input, the mouse moves around but can't control+alt+delete or do anything. Only a hard shutdown allows me to use the computer again. This only happens when I attempt to make a video full screen. The video is stable when its not in fullscreen. A couple of times this would trigger the MadVR bug report app, which I used to submit a report.

I noticed there are a ton of new settings in this build, may have to redo some of my settings, but I cant imagine what would be causing this since ive never had this issue before.

I tried upgrading MPC-HC to 1.7.11, LAV to 0.69.0 and to the latest Nvidia drivers and nothing has fixed this issue.

[ryrynz]

I tried upgrading MPC-HC to 1.7.11, LAV to 0.69.0 and to the latest Nvidia drivers and nothing has fixed this issue.

The smart thing to do would be to reset madVR to defaults, if only after updating only madVR you have issues.. or alternatively switching back to .5 to see if that resolves it, why you didn't just start there I don't know.

[jkauff]

My MPC-HC setup has become extremely unstable after upgrading from 0.91.5 to 0.91.7.
I tried upgrading MPC-HC to 1.7.11, LAV to 0.69.0 and to the latest Nvidia drivers and nothing has fixed this issue.
I'm running the same combination with no problems, so it's almost certainly a madVR setup problem.

ryrynz's suggestion is the best way to go, but a slightly less drastic step is running the madVR uninstall.bat, then running install.bat. That's occasionally worked for me, but in the past when I've run into setup issues I've almost always had to go back to default settings to fix it.

If you use profiles, copy your profile script text to the clipboard and paste it into a Notepad file for reference when you're rebuilding. There are so many new options in madVR that didn't exist before, having to re-do your profiles is probably a good thing, but definitely save your scripts!

[galneon]

My MPC-HC setup has become extremely unstable after upgrading from 0.91.5 to 0.91.7.

Whenever I start a new video and double click it to make it fullscreen, the video glitches out, freezes and becomes unresponsive. The audio will keep playing, then cuts out. If any windows prompt (UAC) opens up, my entire screen goes black and the mouse cursor just flashes really quickly. Can't do anything at this point, the computer is not responsive to any input, the mouse moves around but can't control+alt+delete or do anything. Only a hard shutdown allows me to use the computer again. This only happens when I attempt to make a video full screen. The video is stable when its not in fullscreen. A couple of times this would trigger the MadVR bug report app, which I used to submit a report.

I noticed there are a ton of new settings in this build, may have to redo some of my settings, but I cant imagine what would be causing this since ive never had this issue before.

I tried upgrading MPC-HC to 1.7.11, LAV to 0.69.0 and to the latest Nvidia drivers and nothing has fixed this issue.

I mentioned this here a few days ago. I only noticed it upon updating to MPC-HC 1.7.11, but if it happened upon going from madVR 0.91.5 to 0.91.7 for you that is probably the case for me as well and I just didn't notice.

It only happens with fullscreen exclusive mode. Are you in Windows 10 as well? Windows 10 takes longer to switch to FSE than Windows 7 (and IIRC Windows 8) which could have something to do with it.

[clsid]

If you disable desktop dimming in the UAC settings, then it won't interfere with video playback anymore.

Also try this setting:
MPC-HC options > playback > Output > reinitialize when changing display

[ryrynz]

but definitely save your scripts!

I typed this but didn't end up posting it, I'm sure everyone already knows you can copy your settings.dat and keep it for reference/backup.
This is certainly what I did to avoid issues like this when upgrading to .7.

I copied it to another computer installed .1 on it and copy and pasted everything across to the HTPC via a text file, it took some time but I ended up with a clean settings file as a result.

As I said earlier, changes this big in config I feel should probably come with a suggestion for a fresh setup in the release notes, hopefully I don't need to do this again.

[galneon]

clsid, I have UAC disabled entirely (using built-in Administrator account) and the reinitialize option is greyed out when madVR is selected.

[raymondjpg]

If you disable desktop dimming in the UAC settings, then it won't interfere with video playback anymore.

Also try this setting:
MPC-HC options > playback > Output > reinitialize when changing display
Do you have any idea why MPC-HC version 1.7.11 hangs after a refresh rate change? Stop/restart or pause/restart starts play OK.

This behaviour was first apparent in MPC-HC version 1.7.10. It is not apparent in version 1.7.9.

btw MPC-HC options > playback > Output > reinitialize when changing display is greyed out when using madVR as the video renderer.

[raymondjpg]

I mentioned this here a few days ago. I only noticed it upon updating to MPC-HC 1.7.11, but if it happened upon going from madVR 0.91.5 to 0.91.7 for you that is probably the case for me as well and I just didn't notice.

It only happens with fullscreen exclusive mode. Are you in Windows 10 as well? Windows 10 takes longer to switch to FSE than Windows 7 (and IIRC Windows 8) which could have something to do with it.
The issue I reported about a year ago, video hangs with MPC-HC 1.7.10 after a refresh rate change, happened with both fullscreen exclusive mode and also with windowed overlay mode. Video plays OK after stop/restart or pause/restart.

The issue has re-occurred with MPC-HC 1.7.11. It was not apparent with MPC-HC 1.7.9.

N.B. This is with the x86 version of MPC-HC running in Win 7 Professional, both in x86 and x64 O/S installations.

[glc650]

Hi,

I use madVR with mpc-hc and have mpc-hc's AR set to default (DAR) which from my understanding means the source's AR should be maintained. But when watching a 21:9 film madVR's AR (1920/944 = 2.033) is well off of the soruce's AR (1920/808 = 2.376). Shouldn't it at least be closer to the source's AR if not match it?

Thanks,

->g.

[Q-the-STORM]

Hi,

I use madVR with mpc-hc and have mpc-hc's AR set to default (DAR) which from my understanding means the source's AR should be maintained. But when watching a 21:9 film madVR's AR (1920/944 = 2.033) is well off of the soruce's AR (1920/808 = 2.376). Shouldn't it at least be closer to the source's AR if not match it?

Thanks,

->g.

post mediainfo of the file and don't attach images, upload them on a image hoster and post the link... It takes ages to get attachments approved....

[glc650]

post mediainfo of the file and don't attach images, upload them on a image hoster and post the link... It takes ages to get attachments approved....General
Unique ID : 247413461189807995988510018151845753021 (0xBA22196F54321E11B980ECADEE9470BD)
Complete name : D:\Assassin's Creed.mkv
Format : Matroska
Format version : Version 4 / Version 2
File size : 11.8 GiB
Duration : 1 h 55 min
Overall bit rate : 14.7 Mb/s
Movie name : Assassin's Creed
Encoded date : UTC 2017-03-14 05:13:06
Writing application : mkvmerge v9.9.0 ('Pick Up') 64bit
Writing library : libebml v1.3.4 + libmatroska v1.4.5
Cover : Yes
Attachments : cover_land.jpg / small_cover.jpg / small_cover_land.jpg / cover.jpg

Video
ID : 1
Format : AVC
Format/Info : Advanced Video Codec
Format profile : High@L4.1
Format settings, CABAC : Yes
Format settings, ReFrames : 5 frames
Codec ID : V_MPEG4/ISO/AVC
Duration : 1 h 55 min
Nominal bit rate : 9 329 kb/s
Width : 1 920 pixels
Height : 808 pixels
Display aspect ratio : 2.40:1
Frame rate mode : Constant
Frame rate : 23.976 FPS
Color space : YUV
Chroma subsampling : 4:2:0
Bit depth : 8 bits
Scan type : Progressive
Bits/(Pixel*Frame) : 0.251
Title : Assassin's Creed
Writing library : x264 core 148 r2744 b97ae06
Encoding settings : cabac=1 / ref=5 / deblock=1:0:0 / analyse=0x3:0x133 / me=umh / subme=7 / psy=1 / psy_rd=1.00:0.00 / mixed_ref=1 / me_range=16 / chroma_me=1 / trellis=1 / 8x8dct=1 / cqm=0 / deadzone=21,11 / fast_pskip=0 / chroma_qp_offset=-2 / threads=25 / lookahead_threads=4 / sliced_threads=0 / nr=0 / decimate=1 / interlaced=0 / bluray_compat=0 / constrained_intra=0 / bframes=3 / b_pyramid=2 / b_adapt=1 / b_bias=0 / direct=1 / weightb=1 / open_gop=0 / weightp=2 / keyint=250 / keyint_min=23 / scenecut=40 / intra_refresh=0 / rc_lookahead=40 / rc=2pass / mbtree=1 / bitrate=9329 / ratetol=1.0 / qcomp=0.60 / qpmin=0 / qpmax=69 / qpstep=4 / cplxblur=20.0 / qblur=0.5 / ip_ratio=1.40 / aq=1:1.00
Language : English
Default : Yes
Forced : No

Audio
ID : 2
Format : DTS
Format/Info : Digital Theater Systems
Format profile : MA / Core
Mode : 16
Format settings, Endianness : Big
Codec ID : A_DTS
Duration : 1 h 55 min
Bit rate mode : Variable / Constant
Bit rate : Unknown / 1 509 kb/s
Channel(s) : 8 channels / 6 channels
Channel positions : Front: L C R, Side: L R, Back: L R, LFE / Front: L C R, Side: L R, LFE
Sampling rate : 48.0 kHz
Frame rate : 93.750 FPS (512 spf)
Bit depth : 24 bits
Compression mode : Lossless / Lossy
Title : Assassin's Creed
Language : English
Default : Yes
Forced : No

Text #1
ID : 3
Format : UTF-8
Codec ID : S_TEXT/UTF8
Codec ID/Info : UTF-8 Plain Text
Title : FORCED-SRT
Language : English
Default : Yes
Forced : Yes

Text #2
ID : 4
Format : UTF-8
Codec ID : S_TEXT/UTF8
Codec ID/Info : UTF-8 Plain Text
Title : English-SRT
Language : English
Default : No
Forced : No

Text #3
ID : 5
Format : UTF-8
Codec ID : S_TEXT/UTF8
Codec ID/Info : UTF-8 Plain Text
Title : English-SDH-SRT
Language : English
Default : No
Forced : No

Menu
00:00:00.000 : en:00:00:00.000
00:06:04.155 : en:00:06:04.155
00:09:32.697 : en:00:09:32.697
00:12:47.517 : en:00:12:47.517
00:16:47.715 : en:00:16:47.715
00:21:07.433 : en:00:21:07.433
00:26:56.448 : en:00:26:56.448
00:31:55.080 : en:00:31:55.080
00:38:06.284 : en:00:38:06.284
00:41:43.167 : en:00:41:43.167
00:46:33.040 : en:00:46:33.040
00:50:10.257 : en:00:50:10.257
00:53:04.390 : en:00:53:04.390
00:56:17.624 : en:00:56:17.624
00:59:26.480 : en:00:59:26.480
01:02:32.582 : en:01:02:32.582
01:06:37.035 : en:01:06:37.035
01:13:08.759 : en:01:13:08.759
01:18:45.888 : en:01:18:45.888
01:23:01.101 : en:01:23:01.101
01:30:17.996 : en:01:30:17.996
01:34:18.194 : en:01:34:18.194
01:37:45.193 : en:01:37:45.193
01:40:51.003 : en:01:40:51.003


https://1drv.ms/i/s!AnsGKXR_EKR0gSBorQAwdRG2DWoV

[Q-the-STORM]

You are concerned about this line:

draw 0,0,1920,808 -> 0,136,1920,944

well you don't have to be. It's still displaying 1920x808, this just means instead of having the image on top with a big black bar on the bottom, madVR moves the image to the middle to have 2 smaller black bars on top and bottom...

in this case, the video moves 136 pixels down, so the first row of pixels moves from 0 to 136 and the last row of pixels from 808 to 944... 944-136=808... any other player/renderer does this, madVR just tells you that it does it...

[dioxholster]

I'd use NGU Sharp, as it doesn't require added sharpening. If I wanted to add sharpening, I'd use SuperRes under upscaling refinement. But it isn't necessary with NGU Sharp.

Does that mean NGU sharp under image upscaling/doubling or under chroma upscaling?

[madshi]

Sharing setting between 3 different NGU sharpness levels is definitely reasonable, but as you've said NGU AA has different core.
Yes, but once NGU Soft is available, you'll see that it has a somewhat similar sharpness level to NGU AA, with some other pros and cons. So it's still quite useful to be able to compare e.g. NGU Soft to NGU AA, with all other config being identical.

It would be great to have NGU AA 4x. But what about scale ratio above 4x (sextupling, octupling etc.)? Like NGU AA 2x > NGU Sharp 4x < Bicubic 150?
3x and 6x might be difficult to achieve in good quality. But direct 8x is an option I might look into at some point. It should be very fast, so it should be suitable for e.g. DVD -> 4K upscales.

Also I would like to ask several theoretical questions, if you allow:
1. Is madVR completely spatial (except dither patterns)?
2. In madVR all work is done via GPU shaders, but most avisynth filters, for example, are CPU based. Are you knowingly avoid to use CPU heavy algos (by design), or modern multicore CPUs just less fit in real-time image processing?
1. Not sure exactly what you mean?
2. GPUs are simply *THAT* much better at doing image processing. Using even the fastest CPU would only produce a fraction of the framerate a budget GPU can do. Using both CPU + GPU might help achieving a slightly higher overall framerate, but developing algos on CPU and GPU are totally different, so it would come with a gigantic development cost for me. In other words: Not going to happen.

madshi, a very humble suggestion from me: I think we should take a page from Hydrogenaudio and set up blind tests for scalers comparisons.
Sure, you're welcome to do that! :p Really, it's a good idea, but doing such a test costs a lot of time. And *anybody* can do it, doesn't have to be me. So I won't do it, because madVR development only I can do.

i only want to say thanks for NGU with AA filter
now my 500M looks perfect!
no more artificial sharpness just a smooth clean picture that still is sharp!

i used NGU low before now i use NGU with VHQ AA
damn i looks good!

and alot better than the old NNEDI3 128 i used before NGU

GPU usage slight higher but who cares :)

so for Kuro users
NGU with AA filter gives the most accurate picture on these already sharp 1080P tvs.
Glad to hear you like it! FWIW, scientifically, NGU Sharp produces the most "accurate" picture. But I suppose NGU AA produces are very smooth image, which might look better to your eyes, although it's somewhat soft...

Here is a clip of a poor quality VHS encode:

https://www.dropbox.com/s/r23lsv9w1ycxpfn/Test%20Clip.zip?dl=0

It's a tough call, this shows how good NGU Anti-Alias Very High is vs NNEDI3 256, cleans up more of the aliasing. I think I just like the slightly noisier image NNEDI3 outputs with this video, have a look and see what you think.
Hmmmm... I don't see all that much of a difference with this video, to be honest. But I'm not sure exactly where to look.

No, sharpness is not hiding the pixels. The original pixels must be distinct from each other. Aliasing is good in this case. I don't want the pixels to be blended into each other, unless on a very small scale, such as in the algorithms I linked, which only blend a maximum of 1 screen pixel column or row on upscale, keeping the upscaled pixels intact and looking evenly-sized even at non-integer upscale values. Seeing the individual pixels helps to allow visual analysis of the game state, especially frame by frame. I'll just stick with this version of madVR then.
Or maybe use MPC-HC's "EVR custom" for these game videos, which has a nearest neighbor option.

3) NGU-AA quad low is faster than NNEDI3 quad 16Neurons, quality difference is nearly negligible but in favor for NNEDI3.
NGU-AA quad medium is equal speed to NNEDI3 quad 16Neurons, quality difference is nearly negligible but in favor for NGU-AA.
i get occasional frame drops on either NGU-AA quad medium or NNEDI3 quad 16Neurons, not enough GPU performance?
GPU hardware = HD7950 @ 950Mhz.
K, thanks.

is there any reason for disabling chroma selection when NGU-AA is set to low?
Yes, it doesn't make sense for your to select a higher quality chroma algo when using NGU-AA low. Instead you should switch to NGU-AA med.

also, how do i know whether quadrupling is working or not?
Direct quadrupling? Or quadrupling by doubling twice? Both are shown in the OSD. E.g. for direct quadrupling (which is only available for NGU Sharp at this point in time) you'll see something like "NGU Sharp (low) 4x".

NVidia just released their latest drivers a day ago and they mentioned introduction of a bunch of new OpenCL CUDA stuff, SDK, etc. I just wonder if any of those new things could help madVR in some way...
Nope.

Anyway, why remove any algorithm? If it's about GUI space, you can just add "removed" algorithms to an inconspicuous dropdown list which can grow indefinitely and put them in another source file so you don't have to look at them. The beauty of madVR is that it gives users many choices for how they want to render their videos.
Atm there's only one settings dialog for both experts and noobs. Which means I need to try to find a balance between simple and flexible. Removing options makes things simpler. Maybe with v1.0 I could split the settings for noobs and experts, that would allow more options to be added for experts, only. But this is all not decided yet.

What scaling factor enables image doubling when set to "Let madVR decide?" 1.2x and 3x??
Depends on the algo:

super-xbr: 1.2x, 3.0x (but if you manually set 2.0x for doubling, madVR decides 4.0x for quadrupling)
NNEDI3: 1.5x, 3.0x (but if you manually set 2.0x for doubling, madVR decides 4.0x for quadrupling)
NGU AA: 1.2x, 2.4x (but if you manually set 1.5x or 2.0x for doubling, madVR decides 3.0x or 4.0x for quadrupling)
NGU Sharp double-twice: same as NGU AA
NGU Sharp direct quad: 1.2x, 2.00001x (because direct quad is very fast)

Also, what happens when luma quadrupling is set to "Let madVR decide?" Direct quadruple??
Yes.

4) NNEDI3 256-taps is better than NGU Anti-Alias Very High for chroma upscaling.
For my use, NNEDI3 64-taps (I can't do 128-taps with my graphic card) looks better than any NGU Anti-Alias.
However, NGU Anti-Alias is a very big improvement over NGU (Sharp) for chroma upscaling, but it is not quite as good as NNEDI3 yet.
Anyway, I use Jinc for chroma upscaling now, because the image is visually similar to NNEDI3 64-taps, with a lot less of resources needed (GPU and CPU).
I think the test image you're using is not a good one for seeing differences between NNEDI3 and NGU AA/Sharp. In theory it should be because red fonts on black background sound like a good chroma test, but I can barely see any difference. Here's an image where the different is much bigger:

http://madshi.net/chromaTestImage.y4m

Try doubling the image, to make the differences even easier to see.

Still think NNEDI3 is better than NGU AA for chroma? :p

Also, I am still not convinced with NGU Sharp High in v0.91.7 vs NGU-med in v0.91.1 when upscaling FHD to 4K.

Here is another comparison I have made :

Original : http://www.mediafire.com/file/bo44396go9301l0/original3.png
v0.91.1 : http://www.mediafire.com/file/z09geg7cov5exid/old3.png
v0.91.7 : http://www.mediafire.com/file/0069oc381y59pch/new3.png

You can clearly see the differences on the watch and the ring of the man in the left side, and in the eyes of the three.
I can see that the clock is "interpreted" slightly differently, but I can't really say which interpretation is "better". Anyway, as I said before, NGU Sharp will get some modifications soon. I've no idea if you will like the modifications, though, we'll see... :scared:

Yes, very impressive comparison. But you've used superres with NGU AA. So, SuperRes can improve NGA AA output? Or it's just a workaround to subpixel shift?
NGA AA is rather soft, so a little bit of SuperRes can help, But it also does bring aliasing back to some extent. I'm not sure if I like SuperRes on or off with this specific image. But with SuperRes on it's somehwat nearer to the very sharp Nearest Neighbor look (just without the heavy aliasing). So that's why I turned it on for this comparison.

Having a fullscreen exclusive mode issue after updating to MPC-HC 1.7.11. In the past, when I'd right click, there would be a pause and the context menu would come up, then after I was done with the menu, exclusive would re-engage. Now, the screen goes black for ~6 seconds and returns to fullscreen exclusive without displaying a context menu. I think what is happening is it's switching out, then immediately switching back into exclusive without the menu ever showing. Workaround is to only access the context menu from windowed mode. This is a little annoying since I've already experienced an increased delay in fullscreen exclusive switching when I went from Win7 to 10 (I know, that's MS's fault).
I'm not sure what has changed in MPC-HC to account for this different behaviour. I've seen somewhere else that there's an option in MPC-HC now to reset the d3d device in some situations? Maybe turning that off might help.

If the context menu doesn't appear, trying to right click in the middle of the screen, wait for half a second, then move the mouse a little to the left and top and right click again (blind). Maybe that helps as a workaround?

I suppose at some point it might make sense to somehow detect your right click and disable FSE mode for a couple of seconds, *before* showing the context menu. Right now it's the other way round: The context menu is displayed. Then madVR detects that there's a context menu which won't be visible in FSE mode, so madVR switches to windowed mode, but depending on OS, GPU drivers and media player, the switch to windowed mode may make the context menu close again...