- January 19, 2007
- 791
- 741
- Home Country
-
England
Ahhhhhhh, I get it sorry Kotik 
I should have read further on in that thread I linked...
I should have read further on in that thread I linked...
[WiP] madVR support in MP1 (5 Viewers)
- Thread starter Sebastiii
- Start date
No worries.
Btw all evening today not a single DXVA Interface error, good job @Sebastiii !!!!
I threw at my setup everything i had SD, HD, progressive, interlaced, not a single error, at least not today.
I will keep testing it.
P.S. Testing was done on v113 with 114 Core on top of it.
Btw all evening today not a single DXVA Interface error, good job @Sebastiii !!!!
I threw at my setup everything i had SD, HD, progressive, interlaced, not a single error, at least not today.
I will keep testing it.
P.S. Testing was done on v113 with 114 Core on top of it.
- Thread starter
- Moderator
- #1,483
3D on Nvidia didn't work well on my side but even with MP, on MPC-HC too, can you try with MPC-HC ?
The madVR processing failed is not perfect but it is better now ?
- Thread starter
- Moderator
- #1,484
It seems but not sure @100% :
When Tvservice crash : P1: TVService.exe P2: 1.15.0.0 P3: 5774016c P4: gboxplugin.dll
MP get stuck.
So when it will freeze for the next time, could you try to kill tvservice and restart it and wait that MP unfreeze ?
- Thread starter
- Moderator
- #1,485
That's a good news
ofc we can run sometimes into the issue but it should be rarely (hoping)- January 31, 2008
- 541
- 493
- Home Country
-
Spain
@Sebastiii
No crash. No freeze.
Just madvr process failed. May be you get a different error from before. After installation I did some test with tvservice. That's not related.
A dxva process fail never brings a crash or freeze. Simply the error and lost deinterlace.
I suggest to look between 21:39 to 21:40.
In NVIDIA LAV Cuda zap goes fine. In LAV copy back also. In native, like previously on Intel hd4000 dxva process failed. That's the log collection info may provide.
Will you mind to check it back?
Last edited:
- January 31, 2008
- 541
- 493
- Home Country
-
Spain
Thanks. I'm aware 3D MVC MKV.
So, I have 2 things. Zap and 3D.
My suggestion was to change LAV Vide decoder to DXVA2 (native) and repeat the zap test.
Regarding 3D MVC videos can you try to play some 3D MVC videos with LAV, this time set to copy-back?
Thanks
A
Last edited:
- Thread starter
- Moderator
- #1,488
- January 31, 2008
- 541
- 493
- Home Country
-
Spain
@Sebastiii
Regarding Nvidia 3D MVC investigation I have found the following (very interesting).
MPC-HC never failed on 3D mvc video start regardless of LAV DXVA mode selected. Never failed but I discovered one estrange thing: moving the mouse (pointer presented on the screen) destroyed 3D image. Video continued in 2D. As soon as the mouse pointer is removed from the screen (because of inactivity) 3D mode returned properly. Funny isn't it.
Going back to MP, I found that almost 100% of the times (almost...) Madvr detects 3D video DX11... bla, bla, properly but presents video in 2D.
Here comes the nice thing. moving the mouse, a quick blink in 3D and... after disappearing mouse pointer, 3D back to screen normal.
So 3D works in Nvidia but its affected by something (layers?).
So IMO there is something wrong in the way info is presented.
A deepest check of how the things go when 3D video is played in MP I found:
3D video properly start (3D mode active on TV first).
Very briefly I can see Madvr status in 3D (less 1/4 second).
Then blank screen (madvr status disapears) for 2 seconds, then video appears BUT in 2D.
Finally I need to move mouse on screen and, voila!
Normal 3D.
Tried on 114, and 109 Pre1.15 with LAV final 68.1 and git25
EDIT-------
I forgot 2 thigs.
We have no 100% success just because if I play a 3D video after another 3D video, its highly probable to fail 3D mode (already described with intel HD4000 few days a go).
You must wait for MP to release 3D mode.
In MP-HC the release is only 1.5 secs. In MP the release is 6 secs. can we reduce this release time? Another discovery is that if video played without waiting for 3D mode release, it will be played in 2D (with madvr 3D active) and it can be recovered by manual activation of 3D in the TV. so the fact is that probably in this situation (wait time) 3D is lost because MP or Madvr sends a 2D mode to TV from previous video.
In Exclusive mode, mouse movement doesn't affect a 3D and almost always 3D start properly.
With MadVR "enable windowed overlay", 3D activation in TV and MadVr is very reliable (no 100%) but 3D will be presented in 2D untill mouse movement.
D3D11 fullscreen windowed (3D) when no box click in Madvr General settings its the most failure friendly, even mouse movement will make madvr status in 3D but not the video.
Thanks again
A
Regarding Nvidia 3D MVC investigation I have found the following (very interesting).
MPC-HC never failed on 3D mvc video start regardless of LAV DXVA mode selected. Never failed but I discovered one estrange thing: moving the mouse (pointer presented on the screen) destroyed 3D image. Video continued in 2D. As soon as the mouse pointer is removed from the screen (because of inactivity) 3D mode returned properly. Funny isn't it.
Going back to MP, I found that almost 100% of the times (almost...) Madvr detects 3D video DX11... bla, bla, properly but presents video in 2D.
Here comes the nice thing. moving the mouse, a quick blink in 3D and... after disappearing mouse pointer, 3D back to screen normal.
So 3D works in Nvidia but its affected by something (layers?).
So IMO there is something wrong in the way info is presented.
A deepest check of how the things go when 3D video is played in MP I found:
3D video properly start (3D mode active on TV first).
Very briefly I can see Madvr status in 3D (less 1/4 second).
Then blank screen (madvr status disapears) for 2 seconds, then video appears BUT in 2D.
Finally I need to move mouse on screen and, voila!
Normal 3D.
Tried on 114, and 109 Pre1.15 with LAV final 68.1 and git25
EDIT-------
I forgot 2 thigs.
We have no 100% success just because if I play a 3D video after another 3D video, its highly probable to fail 3D mode (already described with intel HD4000 few days a go).
You must wait for MP to release 3D mode.
In MP-HC the release is only 1.5 secs. In MP the release is 6 secs. can we reduce this release time? Another discovery is that if video played without waiting for 3D mode release, it will be played in 2D (with madvr 3D active) and it can be recovered by manual activation of 3D in the TV. so the fact is that probably in this situation (wait time) 3D is lost because MP or Madvr sends a 2D mode to TV from previous video.
In Exclusive mode, mouse movement doesn't affect a 3D and almost always 3D start properly.
With MadVR "enable windowed overlay", 3D activation in TV and MadVr is very reliable (no 100%) but 3D will be presented in 2D untill mouse movement.
D3D11 fullscreen windowed (3D) when no box click in Madvr General settings its the most failure friendly, even mouse movement will make madvr status in 3D but not the video.
Thanks again
A
Last edited:
tried 114 patch on clients, tv evr seem to work and no mad errors till now .... hooray seb.
but if evr is used for tv - should the mepo rate switch work?
@alexdepalma thanks for all that 3d tests ya did i try some on next weekend btw. by chance i think imma saw your bug about always shift the image to the top or bottom on bug tracker: http://bugs.madshi.net/view.php?id=394 i was looking for some else http://bugs.madshi.net/view.php?id=410 we may should keep an eye on that. ^^
cheers guys
but if evr is used for tv - should the mepo rate switch work?
@alexdepalma thanks for all that 3d tests ya did i try some on next weekend btw. by chance i think imma saw your bug about always shift the image to the top or bottom on bug tracker: http://bugs.madshi.net/view.php?id=394 i was looking for some else http://bugs.madshi.net/view.php?id=410 we may should keep an eye on that. ^^
cheers guys
Last edited: