View Full Version : madVR - high quality video renderer (GPU assisted)
Thunderbolt8
3rd November 2014, 13:51
Samsungs Cinema Smooth 96hz mode works better than madVR smooth motion at 60hz. There is a bit of sharpness loss with madVRs implementation because it blends frames. Cinema Smooth displays the video at 4:4 cadence.does that basically mean that the PQ (sharpness) of the smooth motion option can still be improved?
huhn
3rd November 2014, 14:13
does that basically mean that the PQ (sharpness) of the smooth motion option can still be improved?
playing it at even higher refresh rates could make it sharper and harder to spot.
smooth motion exist to make a file that can not be playback at there native refresh rate or a multiply of it to look smooth at the "wrong" refresh rate.
this option in samsung Tv is something totally different. it kind of fix a bug in the TV.
StinDaWg
4th November 2014, 11:35
smooth motion exist to make a file that can not be playback at there native refresh rate or a multiply of it to look smooth at the "wrong" refresh rate.
this option in samsung Tv is something totally different. it kind of ix a bug in the TV.
What? Playing 24p content at 4:4 cadence (96hz) isn't a bug, it's the proper output. Eliminates judder without messing with the picture by interpolating or blending frames.
huhn
4th November 2014, 11:55
What? Playing 24p content at 4:4 cadence (96hz) isn't a bug, it's the proper output. Eliminates judder without messing with the picture by interpolating or blending frames.
no playing 24p counted without judder isn't a bug.
but you say the Tv creates judder without this option so this option fix this issue. I add the missing F.
James Freeman
4th November 2014, 13:01
Youtube now supports 48 & 60 fps.
There is a cool program that lets you view youtube videos in MPC-HC/BE using madVR called SVPtube.
Since there is no (non-chinese) monitor that supports 48Hz, nor 96Hz nor 120 divides by 48, nor there is a 240 monitor yet, Smooth Motion will have to do for 48fps.
Youtube acceptance of HFR is a huge step toward HFR Cinema & TV and HFR in general.
Sooner than later we will begin to see movie trailers in HFR in youtube then in Cinemas then the scale will tilt and the old 24fps is a goner.
Amen.
Cheers
huhn
4th November 2014, 13:08
there are 144 Hz monitors.
and the program "livestreamer" supports tons of live streaming and video clip pages.
pirlouy
4th November 2014, 13:56
If 24Hz had to die, I'd prefer it would be for 60Hz, not 48Hz. Like you said, 60Hz is often the common refresh rate.
But I prefer 24p to 60i. (De)interlacing is boring.
StinDaWg
4th November 2014, 14:04
no playing 24p counted without judder isn't a bug.
but you say the Tv creates judder without this option so this option fix this issue. I add the missing F.
OK, since you seem to be unable to understand...
Cinema Smooth on=96hz refresh rate
Cinema Smooth off=60hz refresh rate
The tv doesn't "create" judder. 24p played back at 60 hz on any tv is going to have judder. That's what madVR smooth motion is for. If you select Cinema Smooth and play back at 96hz, then you don't need smooth motion. Panasonic plasmas have the same option.
huhn
4th November 2014, 14:44
so I said something different from you?
expect that when the settings is disabled and 23p is the input signal it will display it with 60 hz so creating judder.
so I will never understand why there is even a switch for this and not simply forced...
I got it after your first post really!
if you still think I don't get it feel free to make a new thread and I will answer there.
Cinema Smooth on=96hz refresh rate judder free 23p/24p playback
Cinema Smooth off=60hz refresh rate playing 23p/24p at 60 hz
got it. I guess.
Procrastinating
6th November 2014, 02:48
Looks like at least one AMD developer (http://developer.amd.com/community/blog/2014/11/04/what-do-you-need/) is looking to improve things, maybe someone can let him know about the current interop problems?
MysteryX
8th November 2014, 08:26
Today I tried playing videos with a projector and no image would come out at all with madVR. By disabling madVR, it would play though.
So... madVR doesn't work with projectors.
Qaq
8th November 2014, 10:11
So... madVR doesn't work with projectors.
Thats interesting. Let alone madVR autor uses a projector himself.
pirlouy
8th November 2014, 11:01
:D
dat deduction...
James Freeman
8th November 2014, 11:46
:D
dat deduction...
Life is simple, don't over complicate it. :)
MysteryX
10th November 2014, 02:48
Thats interesting. Let alone madVR autor uses a projector himself.
Then I guess various projectors have slight technical differences. It was a high-end 1080p JVC projector in a small theater room.
huhn
10th November 2014, 03:49
Then I guess various projectors have slight technical differences. It was a high-end 1080p JVC projector in a small theater room.
madVR is creating RGB picture and is giving them to the GPU. the end device doesn't matter. it's the job of the GPU to get the picture to the projector not madVR.
there was a miss configuration some where.
we can only guess. the used hardware of the PC, used connection and driver/OS would be a start.
jmonier
10th November 2014, 15:36
I believe that Madshi actually uses a JVC projector. I, myself, am currently using a JVC RS-4910 and I have used a RS-50 and RS-1X all with madVR successfully so I don't think that your problem is the projector (and, of course, Huhn is absolutely right in that the projector cannot affect madVR operation). If you would give ALL the details about your problem, it's likely that someone here can point you in the right direction.
To start with, look under "devices" in madVR settings and see if your projector (and AVR, if one is in the chain) is listed, and report on everything shown for them. Is there anything else in the chain between the video card and projector?
Qaq
10th November 2014, 18:00
I believe that Madshi actually uses a JVC projector.JVC X3 projector, to be exact. http://forum.doom9.org/showthread.php?p=1632523&highlight=jvc#post1632523
burfadel
11th November 2014, 08:43
I still have problems with the madVR progress bar in full screen exclusive mode (I do NOT mean the standard MPC-HC progress bar). Before anyone says that there is no progress bar in fullscreen exclusive mode, there is! It's got a dark grey top thick transparent line, a light grey line where a blue progress indicator goes, and another dark grey transparent line at the bottom. If you haven't seen it then you have the same issue.
Now, I have two screens at 1920x1080, my main computer monitor and my Samsung TV that I play MPC-HC through. Straight out, the indicator bar does NOT show in 1920x1080 mode on the TV, but does show in some other lesser modes (including one where the scaling is off so not suitable either). If you play around with the settings quite annoying so, you may get it to work. At times, it simply never gets working in 1920x1080.
If you delete the TV from device manager, disable and re-enable the display driver adaptor, play around with the settings some more you may be lucky to get the progress/seek bar back!
I should also point out it is NOT an overscan/underscan issue. The seek bar is simply not there, it's not off the bottom of the screen.
I am pretty sure it's an error with madVR not MPC-HC, but in any case it looks like there's a bug somewhere. I'm guessing either I'm the only one experiencing it, or other people who experience it either have no idea that there is meant to be a progress indicator in fullscreen exclusive mode or don't care.
I guess I'll just have to keep mucking around with it if I install Windows again at some point. What upset it this time was a display driver update.
MysteryX
14th November 2014, 07:36
madVR is creating RGB picture and is giving them to the GPU. the end device doesn't matter. it's the job of the GPU to get the picture to the projector not madVR.
there was a miss configuration some where.
we can only guess. the used hardware of the PC, used connection and driver/OS would be a start.
I use Windows 7 x64 laptop with a Radeon HD7660M. The projector was plugged via HDMI in 1080p mode. HDMI output to TV works fine.
When the output is the projector, MPC-HC remains black when madVR is activated, and plays the video when deactivating it.
What else do you need to know?
jmonier
14th November 2014, 13:59
I use Windows 7 x64 laptop with a Radeon HD7660M. The projector was plugged via HDMI in 1080p mode. HDMI output to TV works fine.
When the output is the projector, MPC-HC remains black when madVR is activated, and plays the video when deactivating it.
What else do you need to know?
You might look at my post above and answer the questions posed there. What you've said here doesn't really tell us anything of significance that we didn't already know.
EDIT: Here is a guide to using madVR:
http://yabb.jriver.com/interact/index.php?topic=80253.0
Ignore the section that mentions Jriver and be aware that not everything in this guide is the best way to go. While there is probably a benefit to using madVR without changing any settings, the real benefit comes with understanding what it does and setting it up for best performance in your specific situation.
huhn
14th November 2014, 17:41
I use Windows 7 x64 laptop with a Radeon HD7660M. The projector was plugged via HDMI in 1080p mode. HDMI output to TV works fine.
When the output is the projector, MPC-HC remains black when madVR is activated, and plays the video when deactivating it.
What else do you need to know?
maybe the same issue as this one?
http://bugs.madshi.net/view.php?id=233
just a guess the information you have giving us is way to limited.
jkauff
17th November 2014, 09:08
I'm now running MPC-HC with madVR on my brand-new HP Stream 7 tablet I bought from the Microsoft Store for $99. This little thing manages to run a full version of 8.1 in 1GB RAM. Works fine and looks good, so far with default madVR settings, so next I'm going to see how much I can push the built-in Atom graphics. Anyway, it's cool to have it working on a 7" tablet.
MysteryX
18th November 2014, 22:05
You might look at my post above and answer the questions posed there. What you've said here doesn't really tell us anything of significance that we didn't already know.
EDIT: Here is a guide to using madVR:
http://yabb.jriver.com/interact/index.php?topic=80253.0
Ignore the section that mentions Jriver and be aware that not everything in this guide is the best way to go. While there is probably a benefit to using madVR without changing any settings, the real benefit comes with understanding what it does and setting it up for best performance in your specific situation.
I hadn't gone into madVR configuration to configure the device as a projector, so perhaps that's what caused the problem.
huhn
18th November 2014, 23:41
I hadn't gone into madVR configuration to configure the device as a projector, so perhaps that's what caused the problem.
that's defiantly not the problem.
hashanga
20th November 2014, 03:24
@madshi
Hi,
This is a purely academic question (for my B.Sc. thesis).
Is it even remotely possible to do motion interpolation as in SVP by extracting motion information directly from h.264?
Or do you know any problem that I can work on for my research? (for which I am required to create an open-source software). Something that is useful. Maybe something that you don't have time to work with.
Thank you.
PS:Sorry for posting this here. Maybe I'm supposed to create a thread. (moderators may move this).
James Freeman
22nd November 2014, 09:39
Last night I had a dream that madshi surprises us and posts version 0.88 with nice improvements to play with.
Hope my dream is meaningful...
Anime Viewer
22nd November 2014, 16:18
Last night I had a dream that madshi surprises us and posts version 0.88 with nice improvements to play with.
Hope my dream is meaningful...
Based on posts madshi has made over the past three months I wouldn't expect a new version of madVR any time soon, and if one were to come out it likely wouldn't have any ground breaking huge improvements. Based on his past three months of posts he's been looking at/working on: shaders, XYsubfilter bug fix(es), and possibly recoding how/when certain scaling, passes, or effects take place in the order of operations. The third subject is the only one that would be directly connected to madVR and require a version update while the other two are independent and likely wouldn't require any changes to madVR.
Shiandow
22nd November 2014, 16:58
Based on posts madshi has made over the past three months I wouldn't expect a new version of madVR any time soon, and if one were to come out it likely wouldn't have any ground breaking huge improvements. Based on his past three months of posts he's been looking at/working on: shaders, XYsubfilter bug fix(es), and possibly recoding how/when certain scaling, passes, or effects take place in the order of operations. The third subject is the only one that would be directly connected to madVR and require a version update while the other two are independent and likely wouldn't require any changes to madVR.
Well, what I understood from madshi's ideas for more advanced control over shaders was pretty groundbreaking (although MPDN got there first :p). It would allow you to write add all kinds of processing and even new scaling algorithms. It didn't sound like it was close to completion, though.
pirlouy
22nd November 2014, 19:47
Last update 7 months ago. Man !
1) I hope next version will contain a whole movie with lots of special effects and a good scenario. If not, I rate it 1 star everywhere it is possible to star it.
2) In general it's in this time area that we insult madshi for not open-sourcing his work. Selfish guy. Noob. :D
Thunderbolt8
23rd November 2014, 12:23
I hope next version will contain a whole movie with lots of special effects and a good scenario. If not, I rate it 1 star everywhere it is possible to star it.:rolleyes:
...although I have to say it would indeed be nice to see a new version around christmas :D
cyberbeing
23rd November 2014, 23:19
It will likely be longer than that. I predict it won't be until March that he focuses heavily on madVR again.
A conservative timeline for madshi would probably look something like:
Considering Delphi XE7 was released only a couple months ago, his first order of business is likely pushing out updates for his paid software. (December-January)
Afterwords, since he's a responsible developer comes a month or two of dedicated support and bug-fixes for paid software updates. (January-February)
Once the need for support on his new releases dies down, then it's break time. (February-March)
Lastly comes madVR, where he'll begin eliminating bugs reported on his bug tracker in rapid fashion. (March-April)
New madVR release (???-???)
An aggressive timeline for madshi would probably look something like:
Prepared in advance for Delphi XE7, and has almost finished a new stable release of his paid software. (December)
Since he had a beta period, only a week or two of dedicated support and bug-fixes for paid software updates. (January)
No break time, he's tough.
Most bugs were already mostly fixed months ago, only a week or so needed to eliminate the rest of the critical bugs on his bug tracker. (January-February)
New madVR release (???-???)
So yeah, be patient and either way we'll likely see a slew of madVR updates in time for spring. ;)
foxyshadis
24th November 2014, 07:29
Last update 7 months ago. Man !
1) I hope next version will contain a whole movie with lots of special effects and a good scenario. If not, I rate it 1 star everywhere it is possible to star it.
2) In general it's in this time area that we insult madshi for not open-sourcing his work. Selfish guy. Noob. :D
That wasn't actually funny, it was just really uncool. That's one of the ways you can really piss off and demotivate a developer.
pirlouy
24th November 2014, 10:03
Well. I didn't think someone could take it seriously. I was wrong.
My apologies.
I never asked for an update. I have no feature request. I'm not an advanced user who needs special features.
ryrynz
24th November 2014, 10:59
demotivate a developer.
Can't imagine Madshi would be affected by pirlouy's teasing.
Zachs
24th November 2014, 11:20
Hi madshi, maybe I haven't been clear but I am not trying to compete with madvr in anyway, just offering an alternative way of video playback. It doesn't even have a lot of the things madvr has. I wouldn't call it a product too, it's more just a pet project :)
dansrfe
24th November 2014, 17:18
Just curious, would a multi-monitor setup with different refresh rates cause frames to be dropped in overlay vs exclusive mode?
e.g. primary monitor and composition rate is at 96Hz, secondary is at 60Hz. Playing in overlay mode on the secondary shows dropped frames while exclusive mode on secondary shows none.
Razoola
24th November 2014, 19:03
Just curious, would a multi-monitor setup with different refresh rates cause frames to be dropped in overlay vs exclusive mode?
e.g. primary monitor and composition rate is at 96Hz, secondary is at 60Hz. Playing in overlay mode on the secondary shows dropped frames while exclusive mode on secondary shows none.
That is normal, I see the same here since forever which is why I only ever use exclusive mode on a secondary display.
huhn
24th November 2014, 20:01
Just curious, would a multi-monitor setup with different refresh rates cause frames to be dropped in overlay vs exclusive mode?
e.g. primary monitor and composition rate is at 96Hz, secondary is at 60Hz. Playing in overlay mode on the secondary shows dropped frames while exclusive mode on secondary shows none.
this shouldn't be any problme with windows 8 or newer.
i was using a gaming PC with a 60 hz and a 120 hz screen for some time with windows 7 and got huge problem with the composition rate from aero.
it is a nvidia GPU so i could disbale aero and still tearing free playback. but this is not working with 2 screen too so i installed the window 10 preview and all issue where gone.
my trick be for was windows 7 overlay and no desktop composition but this sometimes makes problem with games.
or with other words windows 7 is the problem.
leeperry
26th November 2014, 20:03
Looks like at least one AMD developer (http://developer.amd.com/community/blog/2014/11/04/what-do-you-need/) is looking to improve things, maybe someone can let him know about the current interop problems?
@madshi: not sure whether it's classified information but did AMD finally show any interest in assisting you with mVR? I plan on scoring an MSI 750Ti during BF as I don't wanna be stuck with the 13.12 drivers until the end of days(and their TF4 version would supposedly be as quiet as it gets), but if AMD finally decided to unlock the NNEDI3 latencies currently taking place then a passive 7790 might suddenly become an option :cool:
ryrynz
27th November 2014, 01:08
If AMD finally decided to unlock the NNEDI3 latencies currently taking place then a passive 7790 might suddenly become an option :cool:
Don't bother waiting IMO, get the 750 TI it's boss. Overclock the memory too for a bit of a speed boost too barely changing the heat profile.
I can do 64 neurons with most stuff around 720 and 128 neurons under 480. And if I unlocked it's power constraints I could get a lot more, but I'm happy with how quiet it is and it's performance.
StinDaWg
27th November 2014, 05:19
750Ti can't do 720p60->1080p60 NNEDI3 can it?
ryrynz
27th November 2014, 07:05
750Ti can't do 720p60->1080p60 NNEDI3 can it?
Unlikely. And as I've said earlier it's almost pointless (barring text) to upscale 1280x720 using NNEDI, such a waste of processing for such a marginal improvement, but it'll depend on your content. I'd recommend enabling it and taking screenshots to see if you find the difference worth justifying the purchase of a graphics card hefty enough to do it.
wolfman2791
27th November 2014, 10:21
I have a very strange issue: Madvr is causing sound distortion when i put on my headphones. Before you call me crazy... i spent an hour isolating the problem. I put MPC-HC at default setting and the sound was fine. Then i tried the stuff that i use one by one (LAV, etc.). The sound distortion only happened when i switched the renderer to Madvr.
Anyone have a clue as to why this would occur? I'm baffled.
StinDaWg
27th November 2014, 10:33
Unlikely. And as I've said earlier it's almost pointless (barring text) to upscale 1280x720 using NNEDI, such a waste of processing for such a marginal improvement, but it'll depend on your content. I'd recommend enabling it and taking screenshots to see if you find the difference worth justifying the purchase of a graphics card hefty enough to do it.
Pointless in your opinion. It's completely worth it to me as I watch a lot of 720p60 sports.
HD7850 level AMD gpu can do it, which is cheaper than a 750ti.
huhn
27th November 2014, 10:40
gpus on load can put some power on the case that effects onboard and none "high end" sound cards(not that rare have a PC that does this too). in my experience it affect the front speaker connector more than the one at the back panel.
huhn
27th November 2014, 10:49
Pointless in your opinion. It's completely worth it to me as I watch a lot of 720p60 sports.
HD7850 level AMD gpu can do it, which is cheaper than a 750ti.
I can get a 750 ti for less than the 7850 and on top it is a lot cheaper in the end thanks to the fact it uses a lot less power.
a sharper image is not a better one the downscaler adds aliasing and ringing nothing I like to see.
wolfman2791
27th November 2014, 11:07
gpus on load can put some power on the case that effects onboard and none "high end" sound cards(not that rare have a PC that does this too). in my experience it affect the front speaker connector more than the one at the back panel.Ah. Thanks. Not knowing what's up was getting on my nerves lol.
huhn
27th November 2014, 13:15
Ah. Thanks. Not knowing what's up was getting on my nerves lol.
please double check this. don't forget i'm just guessing.
you could use EVR and let a GPU benchmark run in the back ground and see if this audio issue is there too.
StinDaWg
27th November 2014, 13:57
a sharper image is not a better one the downscaler adds aliasing and ringing nothing I like to see.
No, and no. There is less aliasing with NNEDI3+downscale than straight up Jinc. It's the primary reason I use it.
vBulletin® v3.8.11, Copyright ©2000-2026, vBulletin Solutions Inc.