View Full Version : Media Player .NET (MPDN) - D3D HQ GPU Video Renderer [v2.49.0/v1.31.0 27 Dec 2018]
Zachs
12th August 2015, 00:12
Aye, this was a Dolby Digital stream. LAV shows bitstreaming under the info tab, also the receiver states Dolby Digital also when the script is inactive.
BTW, do you remember seeing MPDN Audio Transform Filter in the filter list when you were bitstreaming?
ryrynz
12th August 2015, 00:52
It's there.
Zachs
13th August 2015, 13:47
AudioScript development is progressing well and the framework now utilizes OpenCL. It currently only uses your CPU via OpenCL but there's no reason why it can't use any OpenCL device - I just haven't got around to making a screen to let you select an OpenCL device. I imagine the best combination would be to use a discrete GPU for video rendering and integrated GPU for audio processing.
I've added a simple audio gain (amplification) script on GitHub that demonstrates how easy it is to write OpenCL code with MPDN's AudioScript framework.
It's still experimental at this stage so I urge everyone to give it a try and report any bugs here. It's also missing a "Chain" script that allows you to stack AudioScripts. I'm looking for someone to help me write one so if you're interested, feel free to let me know. It's basically just a subset of the Render Chain script and shouldn't be hard to implement.
Anyway without further ado, the latest MPDN test build is here: http://mpdn.zachsaw.com/Test%20Builds/3310/
You can use the updated MPDN extensions from GitHub here: https://github.com/zachsaw/MPDN_Extensions/tree/5579d52fcff5060f26b882637a335a77964ab7b0
Zachs
13th August 2015, 13:48
@ryrynz can you test this latest version to see if bitstreaming works now?
jaju123
13th August 2015, 19:14
MPDN crashes when pausing for longer than 30 seconds or so with AMD 15.7.1 drivers in Windows 10 :L I have 2x r9 290s.
aufkrawall
13th August 2015, 20:54
Can't confirm with 980. Did you try a different render path, e.g. D3D11 instead of 9?
jaju123
13th August 2015, 22:08
Can't confirm with 980. Did you try a different render path, e.g. D3D11 instead of 9?
If you're referring to me, this issue has only occurred after AMD released their W10 driver branch (15.200). On 15.6 drivers the problem doesn't occur (14.xxx branch).
Zachs
13th August 2015, 22:58
Well then you know the answer to that one...
Zachs
14th August 2015, 07:15
Hi guys,
Just committed a new audio script - Reclock. It does what reclock does and if you want to try it out, grab the MPDN test build and latest sources on GitHub.
It's compatible with DirectSound and Sanear Audio Renderers (Sanear will give you best audio quality when used with MPDN Reclock). However, MPC Audio Renderer isn't compatible.
Cheers!
http://i.imgur.com/iiexIzb.png
ryrynz
14th August 2015, 10:21
@ryrynz can you test this latest version to see if bitstreaming works now?
It works. No transform filter is showing now.
I got this the last time and forgot to mention it, I thought it might be related to the source code extensions sharing the same config as the installer version but after deleting the config it's still happening.
http://i.imgur.com/ofDQ9BD.jpg
Double clicking on it exposes the buttons when the window goes fullscreen.
Zachs
14th August 2015, 10:22
@belphemur can you have a look at the issue?
Belphemur
14th August 2015, 11:00
I dont really know how to reproduce this ... Especially the size of the Form is fixed.
I'm changing the way the button are anchored into the window.
@ryrynz: Did you modify in any way the UI of Windows ?
ryrynz
14th August 2015, 11:28
I dont really know how to reproduce this ... Especially the size of the Form is fixed.
I'm changing the way the button are anchored into the window.
@ryrynz: Did you modify in any way the UI of Windows ?
Oh yeah I should mention I have the UI size set to 150% which tends show issues like cut off forms and such..
BTW Zach I still have the same issue with ffdshow/Avisynth I reported earlier even with 2.39.3
Zachs
14th August 2015, 12:32
The problem you reported has been fixed in the test build. It'll be released soon.
ryrynz
14th August 2015, 12:47
The problem you reported has been fixed in the test build. It'll be released soon.
Oh yeah, somehow I thought that made it into a released build. Knew I should've checked that..
Belphemur
14th August 2015, 12:55
Oh yeah I should mention I have the UI size set to 150% which tends show issues like cut off forms and such..
BTW Zach I still have the same issue with ffdshow/Avisynth I reported earlier even with 2.39.3
I made a big commit that changes the way the SimpleUpdate works.
Can you check if you still have the issues with the buttons ? I changed the way they should be anchored to the window. I also made the window bigger, I hope it'll resolve your issue.
ryrynz
14th August 2015, 12:59
I made a big commit that changes the way the SimpleUpdate works.
Can you check if you still have the issues with the buttons ? I changed the way they should be anchored to the window. I also made the window bigger, I hope it'll resolve your issue.
Yeah saw it, I'll check it out now and update the post.
ryrynz
14th August 2015, 13:01
Nah, the form is a bit bigger but there's still no buttons.
With that update when the window is maximized I can't see the buttons at all.
Might just make a issue for it, no biggie, just revert that change and deal with it later.
Zachs
14th August 2015, 13:09
Can you try with my latest commit instead?
ryrynz
14th August 2015, 13:16
Can you try with my latest commit instead?
Oh yeah.. that nailed it.
ryrynz
14th August 2015, 13:17
Can you try with my latest commit instead?
Pressing cancel though made MPDN stop working though. Forget too.
Will wait for Belphemur's commit.
Zachs
14th August 2015, 13:21
Ah yes I'll get that fixed.
Edit: done.
ryrynz
14th August 2015, 13:26
Ah yes I'll get that fixed.
Edit: done.
I hope you're not pissing him off by doing that.. :D
Zachs
14th August 2015, 13:28
Oops my bad.
aufkrawall
14th August 2015, 15:41
Maybe default dithering strength of 2.0 is a bit high?
With this picture, a strength of 1.5 seems to be sufficient (8 bit display):
http://forum.doom9.org/showthread.php?t=172128
It would be neat if we could choose a lower bitdepth output than 8 to try out dithering settings.
Anima123
14th August 2015, 22:47
Feature request:
It would be nice if MPDN included some kind of video decimation, which is especially useful if one wants to enjoy good scaling algorithm with limited GPU power for high fps videos.
Zachs
14th August 2015, 23:35
It already includes some kind of fps decimation. It's decimated to your display refresh rate. Do you mean decimating the fps to a lower rate than refresh?
Zachs
14th August 2015, 23:35
Maybe default dithering strength of 2.0 is a bit high?
With this picture, a strength of 1.5 seems to be sufficient (8 bit display):
http://forum.doom9.org/showthread.php?t=172128
It would be neat if we could choose a lower bitdepth output than 8 to try out dithering settings.
Hmm Shiandow hasn't been around for some time now... Hope he's just taking a break.
Anima123
14th August 2015, 23:59
It already includes some kind of fps decimation. It's decimated to your display refresh rate. Do you mean decimating the fps to a lower rate than refresh?
Yes. For example, there's 60 fps videos that my nVidia 880M cannot handle using NNEDI3 for image doubling.
It's not like fluid motion feature which blend frames together to forge intermediate frames that do not need scaling at all, hence much more less GPU-hungry compared to scaling each frames.
Zachs
15th August 2015, 05:17
I think you might find that using NEDI or Super-xBR might be better than decimating frame rate...
Shiandow
15th August 2015, 09:53
Hmm Shiandow hasn't been around for some time now... Hope he's just taking a break.
I'm on vacation with somewhat unreliable internet access. I'll be back in a week or so.
Zachs
15th August 2015, 10:30
Ah good to know! Enjoy your vacation mate!
Zachs
15th August 2015, 12:33
Hi everyone,
MPDN 2.40.0 and MPDN Extensions 1.15.0 have been released.
The new audio script feature is now available and there's now a Reclock feature that works automatically (when video frame rate is within 3% of display refresh rate) - i.e. it does what SlySoft's Reclock does with the help of Sanear Audio Renderer (which is now included with MPDN installer). MPDN Reclock also works with DirectSound. It continually adjusts the clock so it matches the display rate so after a minute or two into the playback, you'll find your 'video frame rate (ref clock)' to be very close to your display refresh rate (just a touch under to give refresh rate measurement some error margin). Reclock is one of the simplest audio scripts you can create with MPDN. Other examples (e.g. Gain using OpenCL) are also available in the source version (in the Examples folder).
Belphemur's subtitle drag-and-drop feature is now also available.
Have fun!
Cheers.
Anima123
16th August 2015, 00:00
Thank you for bringing us the re-clock into MPDN. Actually I tried it as soon as I can, and found it not working as well with Optimus as with discrete video card systems.
With Optimus it has been noticed that the ref clock has to tune below the source fps to achieve 2s average closer to the source fps, and the actual refresh rate fluctuate quite a lot.
I do know that Optimus has it's flaws. However, since a lot of game laptops have it, it would be most appreciated if you guys can improve re-clock on the performance on Optimus systems.
Zachs
16th August 2015, 00:03
It was actually implemented on my Optimus laptop. Have a look at the screen shot a couple of pages back. Not sure why your Optimus behaves so differently.
Anima123
16th August 2015, 00:04
I'll post some screenshots when I can.
aufkrawall
16th August 2015, 03:33
With that Reclock feature, MPDN gets mroe interesting for movie watchers. I personally don't need any deinterlacing, but probably MPDN would be more suitable for movies if it had GPU deinterlacer?
Zachs
16th August 2015, 04:06
LAV can deinterlace using Quicksync and Cuvid. Only thing it is lacking is DXVA deinterlacing. At since stage I'll add that but I don't exactly need it personally too.
aufkrawall
16th August 2015, 04:57
Cool. Even cooler if it will be open source extension. :)
Anima123
16th August 2015, 05:04
It was actually implemented on my Optimus laptop. Have a look at the screen shot a couple of pages back. Not sure why your Optimus behaves so differently.
Just double checked, and nothing weird with Optimus when using the latest version. What I observed was possibly due to the fault of the usb audio driver of windows 2000.
Anima123
16th August 2015, 20:46
Bug report:
Wasn't the new display refresh-rate automatic adjustment only be applied when Reclock AudioScript is used? Or else it might cause audio dropout after sometime of playback?
Zachs
17th August 2015, 01:53
Display refresh-rate automatic adjustment? No. Reclock doesn't do and neither does MPDN's Reclock.
What exactly does ReClock do? (http://forum.doom9.org/showthread.php?t=155623)
Anima123
17th August 2015, 04:50
Display refresh-rate automatic adjustment? No. Reclock doesn't do and neither does MPDN's Reclock.
What exactly does ReClock do? (http://forum.doom9.org/showthread.php?t=155623)
Still, what if ReClock is not using, in which case should MPDN NOT speed up or down to match the screen display rate?
Zachs
17th August 2015, 04:51
If it's not in use, no it won't speed up or slow down.
Zachs
17th August 2015, 04:57
Confirmed 343.59 fixes OpenCL. Huhn you might wanna let people know about this on the madVR forum, also mention the device installation settings option to disable auto updating from Windows Update.
I just installed Windows 10 (fresh build) and it comes with v353.62 installed for my NVIDIA NVS 4200M. MPDN's OpenCL NNEDI3 works fine with it. What version of NVIDIA driver did you encounter a problem with?
ryrynz
17th August 2015, 05:07
what version of nvidia driver did you encounter a problem with?
355.60, watch out for Windows 10's broken disabling of its third party driver updating..
Zachs
17th August 2015, 05:13
355.60, watch out for Windows 10's broken disabling of its third party driver updating..
Well according to this page, it was 353.62 that first had the problem: https://forums.geforce.com/default/topic/860152/official-windows-10-353-62-game-ready-display-driver-feedback-thread-7-29-15-/?offset=109
Perhaps they haven't actually removed the extension - they just haven't implemented it for their WDDM 2.0 drivers (NVS 4200M is a Fermi GPU so it's not running WDDM 2.0). So only users with Kepler or newer GPUs are affected.
ryrynz
17th August 2015, 05:26
Well according to this page, it was 353.62 that first had the problem: https://forums.geforce.com/default/topic/860152/official-windows-10-353-62-game-ready-display-driver-feedback-thread-7-29-15-/?offset=109
Perhaps they haven't actually removed the extension - they just haven't implemented it for their WDDM 2.0 drivers (NVS 4200M is a Fermi GPU so it's not running WDDM 2.0). So only users with Kepler or newer GPUs are affected.
That's the thought yeah. I just straight up reverted to what worked and disabled those updates and everything's been sweet as ever since.
Zachs
17th August 2015, 05:41
That's the thought yeah. I just straight up reverted to what worked and disabled those updates and everything's been sweet as ever since.
Does DX12 work with the older driver?
aufkrawall
17th August 2015, 06:16
Yes. Probably they will also remove it for Fermi GPUs when they introduce WDDM 2.0 support for it.
From what I've been told, the extension will not come back.
We still don't know why they did remove it, as AMD and Intel didn't.
vBulletin® v3.8.11, Copyright ©2000-2026, vBulletin Solutions Inc.