madVR - high quality video renderer (GPU assisted)

[LoRd_MuldeR]

@HitomiKun and LoRd_MuldeR,

can you please try whether this build improves the situation for you?

http://madshi.net/madVR82e.zip

Nope, still no difference :o

Though I noticed that the window icon is NOT lost, when I first right-click to open the context menu (so it will exit "exclusive" mode, but remains fullscreen) and then uncheck "full screen".

If I exit from fullscreen via Ctrl+Enter, as I do usually, the window icon is lost reproducible.

Also note that the minimize/maximize/close buttons are lost only, if MPC-HC was maximized before the fullscreen mode is entered. Otherwise they seem to remain...

[madshi]

FSE to windowed works a lot better than before for me (both ways) on JRiver. Thanks for that!
Good to hear.

Btw I've noticed (not specific to this version) that the FSE seek bar doesn't work on JRiver. Instead, moving the mouse to the bottom of the screen makes madVR go into windowed mode and display the same overlay as when you move the mouse to the top of the screen. So I've disabled the seek bar in madVR settings and got used to it. Still, I liked having that tiny seekbar in FSE on MPC.
There's probably a way to configure this in J.River MC somewhere, but I don't know where/how.

Nope, still no difference :o

Though I noticed that the window icon is NOT lost, when I first right-click to open the context menu (so it will exit "exclusive" mode, but remains fullscreen) and then uncheck "full screen".

If I exit from fullscreen via Ctrl+Enter, as I do usually, the window icon is lost reproducible.

Also note that the minimize/maximize/close buttons are lost only, if MPC-HC was maximized before the fullscreen mode is entered. Otherwise they seem to remain...
Oh well, I spent half the day trying to fix that. I had the problem for a while here, too. And I thought I had fixed it, at least currently the problem doesn't seem to show on my HTPC (win7 x64). I have Aero turned off, though, maybe that makes a difference? I don't really know what else to do. The problem is that Direct3D requires a top level window handle, and that's the MPC-HC main window. madVR is only a child window and as such isn't good enough for D3D exclusive mode. Unfortunately D3D11 likes to play around with window styles, and so the buttons sometimes get lost. I don't really have any control over that. I've been trying to manually restore the buttons and it sometimes works, but seemingly not always. Don't know right now what else to do... :(

[ryrynz]

Freezing fixed with 0.81 and MPC 4128. Cheers.

[LoRd_MuldeR]

Oh well, I spent half the day trying to fix that. I had the problem for a while here, too. And I thought I had fixed it, at least currently the problem doesn't seem to show on my HTPC (win7 x64). I have Aero turned off, though, maybe that makes a difference? I don't really know what else to do. The problem is that Direct3D requires a top level window handle, and that's the MPC-HC main window. madVR is only a child window and as such isn't good enough for D3D exclusive mode. Unfortunately D3D11 likes to play around with window styles, and so the buttons sometimes get lost. I don't really have any control over that. I've been trying to manually restore the buttons and it sometimes works, but seemingly not always. Don't know right now what else to do... :(

If you don't already do that, I would think of calling GetWindowLong() to "backup" the old styles before initializing "exclusive" mode and then restore them with SetWindowLong() after "exclusive" mode has exited completely. Another idea that comes to my mind: Create your own "dummy" top-level window for "exclusive" mode rather then abusing MPC-HC's window for that. Don't know if that idea is feasible though...

Probably you know that, but SetWindowPos() has to be called after SetWindowLong() to make the changes take effect

[madshi]

madVR v0.82.2 released

http://madshi.net/madVR.zip

* fixed: SoftCubic was stuck on "softness: 50" (introduced in v0.82)
* fixed: in certain situations there was a new freeze (introduced in v0.82)
* fixed: PotPlayer crashed sometimes when turning audio processing on/off
* fixed: option "if in doubt activate deinterlacing" didn't work properly

[madshi]

If you don't already do that, I would think of calling GetWindowLong() to "backup" the old styles before initializing "exclusive" mode and then restore them with SetWindowLong() after "exclusive" mode has exited.
Yeah, I'm already doing something very similar.

Another idea that comes to my mind: Create your own "dummy" top-level window for "exclusive" mode rather then abusing MPC-HC's window for that. Don't know if that idea is feasible though...
That is problematic because of multiple reasons. One of them being that madVR automatically needs to leave FSE mode when the user opens a settings dialog or something. The MPC-HC settings dialogs usually have the MPC-HC main window as their parent, so they appear in front of the MPC-HC window. If I make my own top level window (which is also topmost) the MPC-HC settings windows would be behind it, and madVR would not leave FSE mode. Another problem is that MPC-HC has to draw a context menu if the user right clicks on the madVR window. Not sure if I can make that work if the window is my own and not MPC-HCs. Next problem: MPC-HC manages the madVR rendering window, setting up the parent etc through DirectShow interfaces. If I create my own top-level window, I'd have to work against what MPC-HC is telling the DirectShow interfaces. And these are just a couple of problems that come to my mind right now. I'm sure there are more...

[HitomiKun]

I'm not quite sure how to install pure .ax files.
I tried regsvr32 and just moving them inside my madVR folder and starting install.bat, both didn't work.

[HitomiKun]

I just tested the newest build and on fullscreen leave I get a constant black screen. Audio is still running in the background though but I can't do anything and I can't see anything. Alt+F4 does not work and the only way to get out of the black screen is by pressing Ctrl+Alt+Del and opening TaskManager.

I also get a REALLY strange feeling that the frame res has been cropped unintentionally. I gonna retest this with older versions and approve this.
Edit: sorry was just my imagination LOL I've been away doing stuff and now I'm back testing all these builds, btw I just got a error messagebox from madVR but it disappeared too fast for me to view.

[sneaker_ger]

Still freezing for me on ctrl+e in FSE mode with MPC-HC 4128 and madVR 0.82.2:
http://www.mediafire.com/?9bxnwa4bh2fsvvh (madVR log and trace with the tool, the "break" key combo didn't work)

[LoRd_MuldeR]

I noticed another issue with MPC-HC r4128 and MadVR v0.82.2:

If I do "File" -> "Close" while a video is still playing, then all controls are disabled as if the video had been closed. But the video continues to play! If I then open another video, MPC-HC will stop responding.

No idea if that is a MadVR or MPC-HC problem. At least I can say that it doesn't happen with other renderers...

[HitomiKun]

Okay madshi I managed to send the report through the error messagebox that appeared with the new build on fullscreen leave. Check your email ;)
The errorbox only appeared in one video though, all the other videos I've been testing that had 1080p res just freeze with an blackscreen that I can't leave. (I can only leave with crtl+alt+del ofc)

[Andy o]

I noticed another issue with MPC-HC r4128 and MadVR v0.82.2:

If I do "File" -> "Close" while a video is still playing, then all controls are disabled as if the video had been closed. But the video continues to play! If I then open another video, MPC-HC will stop responding.


Same here, it doesn't happen with 0.82d.

[Plutotype]

Freezing fixed with 0.81 and MPC 4128. Cheers.

Confirmed here also. MPC-HC R4128 with 0.81 working fine. Starting playback with the MPC-HC option fullscreen turned ON or from XBMC is working flawlessly, the madVR display changer executes the freq change correctly.

Will test 0.82.2 now.

[sneaker_ger]

Oh, and I'm also having the following problems (MPC-HC 4128 + madVR 0.82.2):

In window mode:
Pressing ctrl+c makes mpc-hc seemingly close the file and stop playing (resize to "empty" mpc-hc, all buttons but "play" get grayed out"), but playing actually continues.

Still in window mode:
ctrl+e make the playback stop, but MPC-HC does not freeze and can be closed by pressing "X".

Still in window mode:
Trying to switch to another file while playing, results in the same problem as ctrl+e

[HitomiKun]

Okay madshi I've sent you 3 reports through the error message box. the last report was really tricky so don't wonder about the report details on the bottom that say "asd", this is a new crash that happens on MPC-HC close now. Really strange stuff.

[madshi]

madVR v0.82.3 released

http://madshi.net/madVR.zip

* fixed: Ctrl+C / File Close didn't work in MPC-HC (introduced in v0.82)
* fixed: Ctrl+E in MPC-HC didn't work (introduced in v0.82)
* fixed: finalization crash due to stupid bug (introduced in v0.82)

[sneaker_ger]

Thx.
All my reported problems seem to be fixed.

[Budtz]

Seems 0.82.3 has fixed my issues as well. the other 82 versions were very unstable to me. so thx for the good work:)

[xiulet]

With which madVR version? I'm currently working on stability. It's quite possible that the freeze you're experiencing has a different cause. Please post a freeze report in the madVR thread, then I can say for sure.


i donwloaded madvr 0.82.3, but mpc-hc is showing that is the 0.82 version.

exclusive mode change only works when i pause de video and when i do it from right menu playing video.
i discover that this problem happens if USE D3D11 FOR PRESENTATION is activated, when not, exclusive change works correctly.

But in the past( few days ago) i was using this option with no problem.

thanks my friend.

[Plutotype]

0.82.3 working ok ( playback start with display changer ). Tried different situations - 60 to 23, 23 to 24 etc.

Im so happy that this has been resolved and that madshi has put his effort especially on the MPC-HC forum and instructed Aleksoid to fix things in the player code. Better than any tech support on the world.:thanks:

[Budtz]

Madvr seems faster and more stable now when switching to the next file in the folder while in exlusive mode. This has been unstable in the past with some versions being better then others. Could this be because of the changes made in 0.82?

[FlashGordon]

I have a quick question. When I play fully progressive NTSC DVDs with IVTC enabled, there is a discrepancy between the reported cadences when I use different decoders. For example, if i use libmpeg2 from ffdshow, madvr reports 2:2 cadence, and using LAV Video it reports 3:2. Why is that? Also, since I use Reclock and output at 24Hz, shouldn't it report 1:1? Just confused is all

[HitomiKun]

All my problems are gone now with the new build. Good job madshi and awesome work! Thank you ;)

[ajp_anton]

i donwloaded madvr 0.82.3, but mpc-hc is showing that is the 0.82 version.Same here. Maybe you just didn't include the last number, but it would be nice to know it actually is the newest version.
Also, could you add the version number somewhere in the settings editor too when accessed via the taskbar icon? =)

[Pat357]

Madshi,

I just send another bug report from v0.82.3 : I still get freezes/crashes when going to the next file in a directory.

It doesn't happen every time (as it did with v0.81), but the problem is still there.

Madvr 0.82.3 / MPC-HC 4128 / NV GTX-570 / latest WHQL drivers .295.73 / i7 -970

[madshi]

@Everyone, glad to hear that we're getting nearer to stability.

exclusive mode change only works when i pause de video and when i do it from right menu playing video.
i discover that this problem happens if USE D3D11 FOR PRESENTATION is activated, when not, exclusive change works correctly.
Can you explain in more detail how it does not work right now? What happens exactly? Does your computer explode? Or maybe something else? Please describe as detailed as possible. If you have a freeze -> please create a freeze report and send it to me. If you have a crash -> please send the crash report to me. If it's neither a freeze nor a crash, maybe a debug log could help.

Madvr seems faster and more stable now when switching to the next file in the folder while in exlusive mode. This has been unstable in the past with some versions being better then others. Could this be because of the changes made in 0.82?
Yes. The primary purpose of v0.82 was to increase long term stability. The changes were very deep, though, that's why there were new short term problems. The speed increase is a side effect of some bugfixes and minor improvements I did on the way.

I have a quick question. When I play fully progressive NTSC DVDs with IVTC enabled, there is a discrepancy between the reported cadences when I use different decoders. For example, if i use libmpeg2 from ffdshow, madvr reports 2:2 cadence, and using LAV Video it reports 3:2. Why is that? Also, since I use Reclock and output at 24Hz, shouldn't it report 1:1? Just confused is all
Please make sure you use an up-to-date ffdshow build. There was a bug in ffdshow not too long ago which stopped it from reporting specific flags properly. haruhiko_yamagata has fixed it recently. Also make sure you configure ffdshow correctly: Turn off "Decoder options -> Soft telecine".

Your DVD was most probably encoded as 24p with soft-telecine flags (some of the frames carry a "repeat first field" flag). madVR's IVTC algorithm interprets this field as if there were an additional field encoded in the video stream. That's why it's 3:2 and not 2:2. Hardware DVD player devices actually duplicate one of the fields when the "repeat first field" flag is set, before sending the data to the display, when the output mode is set to 60i.

It's not 1:1 because the numbers indicate the number of fields per frame. A frame always must have at least 2 fields, otherwise it would be video mode (instead of film mode) and would require a video mode deinterlacing algorithm like e.g. YADIF. Native interlaced video is practically 1:1. IVTC doesn't work for that.

I just send another bug report from v0.82.3 : I still get freezes/crashes when going to the next file in a directory.
Got them and fixed all of them. However, in one crash report's comment you wrote:

"The 0.82.3 version did'nt switch to the propriate freq : the display remained on 60 Hz, even when playing 23.976 stuff.
I have 48Hz, 50Hz and 60Hz configured."

I need a debug log from v0.82.3 which shows how it doesn't switch to 48Hz although it should. Please also let me know which display mode changer settings you're using exactly. Thx.

[703]

Any chance of making a settings screen in MadVR where I can set the default matrix co-efficient and color space (primaries) for SD videos - instead of leaving it for best guess? Or some kind of changable default behaviour for best guess.

There seems to be an increasing number of SD resolution videos out there who uses bt.709 for matrix and primaries, because they are off HD sources and left as-is, without following the industry convention.

[madshi]

Any chance of making a settings screen in MadVR where I can set the default matrix co-efficient and color space (primaries) for SD videos - instead of leaving it for best guess? Or some kind of changable default behaviour for best guess.
I guess I could add an option to force specific matrix and primaries for *all* videos. I could add this without touching the settings dialog by simply allowing you to "store" forced matrix and primary settings through shortcut keys + [F2].

Making such a setting available only for SD content, but not for HD content, would be much more complicated, though. For that I'd have to add options to the settings dialog, which IMHO would make things much harder to understand. And most people wouldn't use/need this functionality. So I don't really like that idea.

There seems to be an increasing number of SD resolution videos out there who uses bt.709 for matrix and primaries, because they are off HD sources and left as-is, without following the industry convention.
And I guess these don't have the proper matrix/primaries stored anywhere? How do you know then which the proper matrix/primaries are?

Let's say there were such an option as you're asking for: It would result in that you'd get incorrect matrix/primaries for all original MPEG2 PAL and NTSC DVDs. Does that really make sense!?

[DragonQ]

The Blu-Ray spec doesn't allow 25p, it only allows 50i, as far as I know. Ok, I guess you can encode 25p with telecine flags. Anyway, topic closed, I bowed to the majority vote. So no need to discuss this any further.
FWIW, 25p can be done on BD using 25 PsF (progressively segmented frames), which is essentially 25p inside a 25i stream. That's the way films and dramas have been shown on TV for years.

Players *should* detect whatever flag signifies this and output 25p but those that don't will simply output as 25i and you'll end up with it being deinterlaced to 50p with every frame shown twice. Probably no disernable difference in quality between the two methods with a good deinterlacer though.

[madshi]

FWIW, 25p can be done on BD using 25 PsF (progressively segmented frames), which is essentially 25p inside a 25i stream. That's the way films and dramas have been shown on TV for years.
Some PAL TV stations broadcast that way, others send true 50i even for movies. E.g. when SkyHD Germany switched their hardware decoders 2 years (?) ago, their broadcast encoding changed to truely interlaced encoding.

Players *should* detect whatever flag signifies this and output 25p
The problem is that flags like these are not always reliable. NTSC DVDs and USA broadcasts often switch between soft- and hard-telecine. The same could happen to a Blu-Ray encoding or to a PAL broadcast. E.g. when SkyHD Germany still broadcasted with their old hardware (a couple years back) they usually sent 25 frames per second with interlaced flags, but once in a while in the middle of the movie the encoder switched to fully interlaced encoding for a few seconds. It took me a while to get this fully working in eac3to, so I know this for a fact. Also you can never know, some documentaries may have natively interlaced sections (50i) and natively progressive sections (25p). Not sure how this would be encoded for a Blu-Ray.

Basically that means that a Blu-Ray player which wants to be robust against such problems might opt not to rely on flags and instead always output 50i instead of 25p for Blu-Rays. Or alternatively run the stream through a deinterlacer and then output 25p/50p.

BTW, only few TVs understand 1080p25 in the first place, so in the end the more usual practice will be to output either 50i or 50p.

[DragonQ]

Some PAL TV stations broadcast that way, others send true 50i even for movies. E.g. when SkyHD Germany switched their hardware decoders 2 years (?) ago, their broadcast encoding changed to truely interlaced encoding.

Well they all* send "true" 25i for everything but the source is still progressive so it's 25 PsF, that's my point. Whether they flag it as such is a different matter but like I said, with a good TV deinterlacer, it shouldn't make much difference to the end product.

*The only exceptions I know of are the BBC HD channels that can transmit true 1080p25 and even switch mid-programme when necessary (e.g. the intro to Match of the Day is broadcast as 1080p25, then the stream switches to 1080i25 for the programme itself).

Basically that means that a Blu-Ray player which wants to be robust against such problems might opt not to rely on flags and instead always output 50i instead of 25p for Blu-Rays. Or alternatively run the stream through a deinterlacer and then output 25p/50p.

BTW, only few TVs understand 1080p25 in the first place, so in the end the more usual practice will be to output either 50i or 50p.
I completely agree, flags are notoriously unreliable. I'd imagine most players simply output 1080i25 - I doubt players ever output 1080p50 as it's not in the BD specification and it's also not required for "HD Ready" certified TVs. Not sure about how many HDTVs accept 25p but I know the two in my house do. However, they're probably internally frame doubling to 1080p50.

[kerman]

The auto-frequency adjusting function is totally automatic? does it get the fps info by the source filter and autoadjust refresh output to it?

[Pat357]

Got them and fixed all of them. However, in one crash report's comment you wrote:

"The 0.82.3 version did'nt switch to the propriate freq : the display remained on 60 Hz, even when playing 23.976 stuff.
I have 48Hz, 50Hz and 60Hz configured."

I need a debug log from v0.82.3 which shows how it doesn't switch to 48Hz although it should. Please also let me know which display mode changer settings you're using exactly. Thx.
I use the MadVR build in display-mode switcher to adapt the frequence from display to the content displayed.
No other tools are involved : the MPC-HC disp. changer is disabled (it never worked for me).
Yesterday the MadVR display changer stopped working and I clearly saw some red text displayed in the upper left corner :
something like "D3Ddevice initialization failed". (don't remember exact words).

I can still change the display using the Slysoft's "SetDisplayFrequency.exe" however.

I went through a couple of files with different framerates
Here is the Madvr v0.82.3 log :
http://www.mediafire.com/?o5y5hlwuwk9w3db

[madshi]

madVR v0.82.4 released

http://madshi.net/madVR.zip

* fixed: 2 more finalization crashes
* fixed: display mode changer stopped working after a 2nd video file was loaded
* fixed: filter properties dialog didn't show the full madVR version number
Any more (big) problems left, guys?

[vivan]

Me again with notebook with nVidia optimus and freezes.

I have acer 3830TG which has integrated (SB) and nVidia GPU. With iGPU everything is ok, except 1080p 60fps video, it's just too slow for that. nVidia GPU is fast enough, but... After ~30 seconds video freezes. Also in exclusive mode it has a lot of dropped frames (so video looks like 5 fps), though madVR OSD says that everything is ok.
Logs: http://www.mediafire.com/?gatoxb3llozedek
"1_madVR - log" is in windowed mode. When video freezed - "1_madVR - freeze report (1)".
"2_madVR - crash report" is in exclusive mode. When video freezed, madVR crashed. So "2_madVR - crash report".
Hope new logs could help to sort out the problem.

[madshi]

I have acer 3830TG which has integrated (SB) and nVidia GPU. With iGPU everything is ok, except 1080p 60fps video, it's just too slow for that. nVidia GPU is fast enough, but... After ~30 seconds video freezes. Also in exclusive mode it has a lot of dropped frames (so video looks like 5 fps), though madVR OSD says that everything is ok.
Logs: http://www.mediafire.com/?gatoxb3llozedek
"1_madVR - log" is in windowed mode. When video freezed - "1_madVR - freeze report (1)".
"2_madVR - crash report" is in exclusive mode. When video freezed, madVR crashed. So "2_madVR - crash report".
Hope new logs could help to sort out the problem.
The freeze report and debug log don't show me anything that looks wrong to me, except that suddenly out of the blue after about 30 seconds one the Direct3D APIs stops working, which has worked just fine the past 30 seconds. I can't see any obvious reason for that. Of course there's always a chance that there's a bug in madVR somewhere that causes this, but looking at that log I rather think the problem is with the GPU driver.

Interestingly, the crash happens in the "same" API that fails in the freeze. Well, it's not really the same API. The freeze occurs in D3D9, while the crash happens in D3D11. But in both cases the problem occurs when trying to copy a rendered frame into the D3D backbuffer. I don't know what's going on there, but I do think it's a GPU driver problem. The crash happens in an NVidia dll, not directly in madVR.

If exclusive mode doesn't perform well for you, you could different settings. E.g. try different flush settings.

But above all: Try different GPU drivers!

[TheShadowRunner]

Hi madshi, I could recreate the cosmetic oddity I experience with builds after 0.80.
It's only cosmetic and madVR works otherwise perfectly fine, not a big bug by any means.

In effect, after playing a couple videos, instead of seeing the "Exclusive" pop up on the top-left corner of the screen indicating that madVR is going in FSE, the whole screen is just going black for like 0.2 second when madVR goes into FSE. The "Exclusive" pop-up never shows up. After the black screen, playback continues just fine.

It seems the "Exclusive" pop-up is just replaced by a black frame.

Here is a log that I tried to make as short as possible, but sadly when I try to make it shorter the issue doesn't happen!
In this log, the bug happens the very last time madVR goes into exclusive mode. Earlier, the "Exclusive" pop-up shows up just fine.
http://videoff7.free.fr/madVR_TSR_cosmeticFSE.zip
Can you see anything strange?
See you,

TSR

[madshi]

@TheShadowRunner, the "Exclusive" pop-up is only supposed to show when you switch from one mode to the other. However, if a display mode change is involved, too, madVR might instead of just switching release the old Direct3D stuff and re-create everything. In that case the pop-up might not appear. Bla bla, I think what I'm saying is too complicated. Anyway. The key point is: Depending on whether madVR just puts Direct3D into a different mode, or whether it totally re-creates the Direct3D device, the pop-up might appear or not. That might not make too much sense, though, so maybe I should redesign that. I'll put that on my to do list. But it's pretty near the bottom there... ;)

[TheShadowRunner]

Thanks for your super fast reply.
But the issue isn't that the Exclusive pop-up does not show up. It is that it is replaced by a (almost subliminal so it's fast) black screen instead.
If it's expected, no worries ;)
See you,

TSR

[vivan]

I've tried avaivable WHQL drivers before (280.26, 285.62, 295.73). I'll try playing with settings...
nVidia optimus has such scheme: http://media.bestofmicro.com/nvidia-optimus-mobile,T-A-238222-13.png. After GPU is ready with rendering - it copies frame to iGPU. Long way...

P.s. maybe this is the reason why nVidia prohibit launching "mpc-hc.exe" with discrete GPU starting from 285.62. That's why it's named "mpc-hc2.exe" on my machine.

UPD: unticking "use a separate device for presentation" fixed freezes, and even exclusive mode (however if after 30 seconds of exclusive mode switch back to windowed - video starts flashing with black frame (looks like every 5th frame is plain black))

[madshi]

Thanks for your super fast reply.
But the issue isn't that the Exclusive pop-up does not show up. It is that it is replaced by a (almost subliminal so it's fast) black screen instead.
If it's expected, no worries ;)

I'm not sure it's expected. Did you not have a black screen (with or without the "Exclusive" pop-up on top) with older madVR versions? Is a display mode changed involved, maybe? Changing display modes usually results in the display having to re-synchronize. Older madVR versions didn't change display modes when going into exclusive mode. The newer version do that if you configure the display mode changer accordingly.

[madshi]

P.s. maybe this is the reason why nVidia prohibit launching "mpc-hc.exe" with discrete GPU starting from 285.62. That's why it's named "mpc-hc2.exe" on my machine.
Seriously!? This is terrible!! Maybe you should get an AMD notebook next time... :p

[vivan]

Next time ivy bridge would be fast enough :P

Anyway, in windowed mode now eveything is ok, so thanks :)

[TheShadowRunner]

Madshi, no, with older madVR builds, I never had a black screen replace the "exclusive" pop-up, ever.

Since I "delay switch to exclusive mode by 3 seconds", I know it's not the resolution change itself that does the black screen.
I open the video -> a resolution switch occurs because framerate doesn't match frequency -> video is back -> a couple seconds later madVR switches to exclusive mode.

It's this last switch when the bug occurs, most often the "exclusive" pop-up shows up fine as usual.
Other times, it'll be replaced by a super fast black screen.

In both cases, after this, madVR is in Exclusive mode alright. It is only cosmetic.

This black screen really replaces the "Exclusive" pop-up itself, like if madVR was confused and didn't know where to "print" it.

It does happen after I switch frequency "behind" madVR's back though (ReClock's RunEvent) and cannot be reproduced with madVR's built-in changer as it's for cases such as 1280x720p60 -> 1920x1080p60, which cannot be replicated in madVR yet.

But again no biggie, it's only cosmetic and those last builds have been totally epic. madVR is now far faster and stable than it ever was ;)

[madshi]

Why does madVR not support 1280x720p60 -> 1920x1080p60 switching? It should, at least in theory.

[TheShadowRunner]

...??
How could madVR possibly know when to use 720p60 rather than 1080p60 or the other way round?
I need this kind of switching (resolution only, untouched frequency) based on media source height ^^;

[madshi]

madVR tries to find the best matching display mode. If you have both 720p60 and 1080p60 in the madVR display mode switch edit control, then for 720p content madVR should switch to 720p60 and for 1080p content madVR should switch to 1080p60. The refresh rate is considered most important by madVR, though. So if you have e.g. "720p60, 1080p24, 1080p60" in the madVR display mode changer edit control, then for e.g. a 720p24 trailer madVR would switch to 1080p24. But for 720p60 content it should switch to 720p60 and for 1080p60 content it should switch to 1080p60. At least that's how it's supposed to work.

[TheShadowRunner]

madshi, that's nuts, I never knew that :D
So madVR does take source height into account now?
With 720p60 and 1080p60 set, when media height is < or = to 720, it'd switch to 720p60 and for anything above > 720, it'd switch to 1080p60? (of course we're talking ntsc media only atm).
That's completely new to me, when was this implemented??

edit: indeed it works, and i've been fighting with ReClock's runevent to coexist with madVR for all this time when it was implemented in madVR in the first place :nuts: !!

[madshi]

So madVR does take source height into account now?
It always did, at least it was always supposed to. It simply tries to find the best match from the list of modes you supplied. However, for a 480p60 file both 720p60 and 1080p60 are considered equally good matches, so whichever of those 2 is listed first is selected. For > 720 content, 1080p60 is considered better than 720p60.

[mark0077]

I personally can't produce any crashes with v0.82.4, thanks!

Quick question if you have time: When using "delay playback until render queue is full", in my setup for example using a render queue of 24, is this feature designed to get to X-24 / 24 (with X being anything), or fully 23or24-24 / 24?

The reason I ask, some clips that stress my machine out a bit more than others, tend to take a little bit longer to fully fill the queues (naturally), ie to reach 24 / 24, and so playback will stutter for a couple of seconds, and often the queues will even drop back down to very low levels initially / drop frames until it gets a chance to catch up to 100% full and its fine from then on, ie no frame drops. Im just wondering is there any way to tweak this feature to cater for this, ie to wait a little bit longer until everything is 100% full at the start :) Not a problem either ways, this feature works great on 90% of my content.