Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion.

Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules.

 

Go Back   Doom9's Forum > Hardware & Software > Software players

Reply
 
Thread Tools Search this Thread Display Modes
Old 12th May 2015, 06:04   #29721  |  Link
huhn
Registered User
 
Join Date: Oct 2012
Posts: 7,903
Quote:
Originally Posted by agustin9 View Post
The thing is, I forgot to mention, that i have the display switcher off on both screens... I'm using Windows 8.1...
yeah display mode switcher is not needed to trigger this current bug. but i wouldn't be shocked if it's not needed for the old bug too.
i may load a old madVR version and give it a try.
and it doesn't seems to matter if it is windows 8, 8.1 in your case or 10 in my case.
huhn is offline   Reply With Quote
Old 12th May 2015, 06:15   #29722  |  Link
James Freeman
Registered User
 
Join Date: Sep 2013
Posts: 919
Quote:
Originally Posted by potomac View Post
i tested 10 bit on my cheap lg ips235 and i can see the improvement in the gray scale, but my 5850 doesn't work well in dx11 mode
This sentence contradicts itself in one go.
The LG IPS235V most definitely doesn't support 10bit input.

I beginning to think that madshi should include a simple test in madVR to make things clearer.

Please refer to this thread for more info and a simpler test:
http://forum.doom9.org/showthread.php?t=172128
__________________
System: i7 3770K, GTX660, Win7 64bit, Panasonic ST60, Dell U2410.
James Freeman is offline   Reply With Quote
Old 12th May 2015, 06:19   #29723  |  Link
cvrkuth
Registered User
 
Join Date: Apr 2011
Posts: 54
Quote:
Originally Posted by cvrkuth View Post
With Direct3D 11 presentation path enabled I have display switching issue again, now in madVR32 too (There was issue in the 32 bit version, v0.86.10. Madshi then do workaround and issue was gone).



Not any more
Display modes switch problem presents only if FSE enabled. When FSE is disabled Madvr switch display modes correctly
__________________
Intel Core i7-4790 CPU @ 3.60GHz, RAM 32 GB Dual-Channel DDR3 @ 665MHz (9-9-9-24),
Panasonic TX-P42G20E, NVIDIA GeForce GTX 970, Win 10 Pro x64,
PotPlayer 1.7.16291 64-bit, madVR v0.92.17
cvrkuth is offline   Reply With Quote
Old 12th May 2015, 06:23   #29724  |  Link
Hprd
Registered User
 
Join Date: Mar 2011
Posts: 60
Finesharp vs superres. I only did double scaling with nnedi3 (as quadrupling with superres doesn't work anyways for me). With double scaling (nnedi3 for chroma as well with superres there on defaults) finesharp with slightly tweaked settings (all defaults except mode set to 3, and linear light turned on) gives a (subjectively) noticably better result than superres (nnedi3 default settings), and lower rendering times too (perhaps more important in some ways obviously). I honestly didn't think upscaling could be this sharp, detailed and clean (for virtually any content I've thrown at it).

Last edited by Hprd; 12th May 2015 at 06:34.
Hprd is offline   Reply With Quote
Old 12th May 2015, 06:23   #29725  |  Link
huhn
Registered User
 
Join Date: Oct 2012
Posts: 7,903
Quote:
Originally Posted by James Freeman View Post
This sentence contradicts itself in one go.
The LG IPS235V most definitely doesn't support 10bit input.

I beginning to think that madshi should include a simple test in madVR to make things clearer.

Please refer to this thread for more info and a simpler test:
http://forum.doom9.org/showthread.php?t=172128
even a cheap dispaly can easily support 10 bit or even 12 bit input.
the interesting part is what the display does with it to display it on it's 6 bit or 8 bit panel.

on AMD and the newest nvidia beta driver for windows 10 you have full control of the output bit deep. of cause this option give you only possibilities the screen can take as an input.
huhn is offline   Reply With Quote
Old 12th May 2015, 06:29   #29726  |  Link
bcec
Registered User
 
Join Date: Nov 2014
Posts: 81
Quote:
Originally Posted by bcec View Post
I seem to be getting 10-bits only when my display is at 60Hz. When at 23Hz, I don't seem to be getting the 10-bit output.

I am using the following pattern in mpc-hc x64:
http://forum.doom9.org/showpost.php?...8&postcount=11

win7x64, gtx970 connected to 10-bit TV via HDMI 2.0.

At 60Hz dithering disabled, I see a smooth pattern.
At 23Hz dithering disabled, I see the same banding 8-bits and up.

I am not sure who is at fault, madvr, nvidia or TV? Any ideas? Does anyone successfully produce 10-bit output in 23Hz?
Proper 10-bit output seems to be only happening when at 60Hz with win7x64 + nvidia combination. Any other refresh rate, even though madvr reports 10-bits, only 8-bits is displayed (as evidenced by the test pattern). Are others seeing this also?
bcec is offline   Reply With Quote
Old 12th May 2015, 06:34   #29727  |  Link
omarank
Registered User
 
Join Date: Nov 2011
Posts: 187
Quote:
Originally Posted by madshi View Post
It's working fine here. However, with active 10bit output (in D3D11 FSE mode) error diffusion is disabled automatically and ordered dithering is used instead. The reason for that is that at 10bit, the quality of the dithering should play no role for image quality, anymore. This only applies to 10bit FSE mode, though. In windowed mode (which is always max 8bit) error diffusion is still used.
Well, you invested so much time in developing and fine tuning the error diffusion algorithms. They are a premium feature of madVR, but you have disabled them for another premium feature – 10 bit output. I completely agree that there would be hardly any benefit in using Error diffusion dithering over ordered dithering for 10 bit output, but I believe it doesn’t harm either. Also, HDR displays will be available sooner or later, and error diffusion dithering at 10 bit would be beneficial again.
omarank is offline   Reply With Quote
Old 12th May 2015, 08:12   #29728  |  Link
ryrynz
Registered User
 
ryrynz's Avatar
 
Join Date: Mar 2009
Posts: 3,646
Hey madshi could you allow for the bitdepth shortcut key to switch to 9 and 10 bit please? If I use the shortcut key it'll switch down from 10 to 9 and 8 bit
but it doesn't start back at 10 bit.

It would also be kinda cool for a shortcut to switch between two user specified bitdepths.. also could you add a shortcut to cycle dithering methods? Cheers.

Quote:
Originally Posted by omarank View Post
I completely agree that there would be hardly any benefit in using Error diffusion dithering over ordered dithering for 10 bit output, but I believe it doesn’t harm either. Also, HDR displays will be available sooner or later, and error diffusion dithering at 10 bit would be beneficial again.
I like how error diffusion looks on 8 bit over ordered dithering, I too would like to see this enabled for 10 bit so I can see for myself how it looks on my TV.
Being a plasma madVR's dithering patterns seem to affect it more than LCD's, It would certainly help in that instance having the keyboard shortcut I suggested above for comparisons.

Last edited by ryrynz; 12th May 2015 at 08:43.
ryrynz is offline   Reply With Quote
Old 12th May 2015, 09:13   #29729  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
Quote:
Originally Posted by Anima123 View Post
According to the GPU usage, I am afraid dithering fall back to ordered dithering all the time even with 8-bit output. Could you please double check?
I checked yesterday and error diffusion worked just fine for me. Except in 10bit FSE mode, where I intentionally disabled it. If you find the error diffusion doesn't work for you in other situations, as well, I'd need a way to reproduce that. Maybe you can find a specific set of circumstances in which it works and in which it doesn't? Try setting the display bitdepth to 2 bits. Then you can easily see the difference in the dither pattern when switching between different dither options.

Quote:
Originally Posted by aufkrawall View Post
I found out that my problem with NNEDI3 + FineSharp afterwards is caused by image quadrupling.
Doubling 64 + FS works fine, but additional quadroupling 64 produces corruption and NV driver failure.
Anyone confirm? Be sure that quadrupling is actually active.
Quote:
Originally Posted by edcrfv94 View Post
if use NNEDI3 double & quadruple image with superRes Player will stop only can be close
Oooops, yes, there's a bug still in v0.88.4, will be fixed in the next build.

Quote:
Originally Posted by aufkrawall View Post
Another issue: LumaSharpen doesn't seem to be active if strength isn't 1 or higher.
Right! Thanks for letting me know. Will be fixed in the next build.

Quote:
Originally Posted by XMonarchY View Post
madVR 0.88.4 is not faster for me in regards to Image Doubling and Quadrupling than madVR 0.88.2. I changed Image Upscaling from Jinc 3 tap + AR to Lanczos 3 tap + AR per your advice, but that did not help at all. Doubling + Quadrupling, both @ NNEDI3 16 neurons + NNEDI3 16 neurons for Chroma Upscaling result in SEVERE frame drops... That is WITH and WITHOUT SuperRes Chroma Upscaling filter and SuperRes Image Refinement... Using D3D11 or not using it also made no difference. Could it be just nVidia drivers? I use the latest 350.12 nVidia drivers, LAV Filters 0.65 (LAV Video uses CUVID), MPC-HC 1.7.8, ReClock
Ok, can you please retry with v0.88.5? If that still isn't working for you, I'll need some more tests:

1) With v0.87.21, what happens if you disable Chroma doubling and Chroma quadrupling (right side of the image doubling page)? Do you then get similar results to v0.88.5?

2) What are your rendering times with: a) v0.87.21 with your original settings. b) v0.87.21 with disabled Chroma doubling/quadrupling. c) v0.88.5?

Please always keep the new features turned off for now, as not to complicate the performance comparison to v0.87.21.

Quote:
Originally Posted by XMonarchY View Post
I still want to figure these out in addition to figuring out how to resolve Image Quadrupling issue:
1. What about Direct3D 11? Does it improve quality or performance? When I enable it with or without the "present a frame for every Vsync" I get presentation glitches, but no delayed or dropped frames. What are the presentation glitches? Should presentation glitches be ignored if no frames are delayed and dropped?
2. Is there any comparison or reports from people regarding the difference between SuperRes for Chroma Upscaling filter VS. SuperRes for Image Refinement.
3. Does 10bit selection for "the native display bitdepth is" work for nVidia Maxwell owners who use 8bit+FRC monitors via DisplayPort? My Eizo Foris FG2421 is an 8bit+FRC panel, but nVidia lacks an option to enable 10bit depth in its settings. Should I select 10bit anyway?
I can't answer questions specific to your GPU or display. Just try out what looks best to your eyes and use that. If you get presentation glitches in D3D11 not only when seeking and pausing/playing, but also in the middle of straight playback, and if you don't get any in D3D9 then it might make sense to stick with D3D9. Although I would guess that in FSE mode D3D11 should show no glitches during straight playback. If it's only in windowed mode then you need to answer yourself whether it bothers you or not. If you don't want to use native 10bit output then there's no big loss by sticking to D3D9.

It is too early to discuss SuperRes and all the other new features. I've explicitly asked users to concentrate their tests on the new debanding feature for now.

Quote:
Originally Posted by bcec View Post
I seem to be getting 10-bits only when my display is at 60Hz. When at 23Hz, I don't seem to be getting the 10-bit output.
That is possible, I guess. It could be the GPU or the display, I can't say. There's most probably nothing I can do about it.

Quote:
Originally Posted by Anime Viewer View Post
Experimenting with the image enhancement and upscaling refinement settings it seems FineSharp and SuperRes provide similar image quality improvements to each other. The two together provide too many artifacts, so it seems (at least on my system) that its best to choose one or the other. SuperRes seems to provide slightly higher render times and faster present times while FineSharp has faster render times with higher present times. Given the 480 video upscaled to 1080 resolution I was testing there was only a small swing either way (SuperRes 15ms render 9ms present, FineSharp 8ms render, 12ms present). What are opinions on which is the better to keep low(er), or allow to drift high(er).

Edit: Tested 720p video upscaled to 1080p also. Finesharp was ~11ms on both render and present while SuperRes was 20ms render and 8ms present. Based on that I'll probably go with Finesharp for the moment.
Quote:
Originally Posted by Hprd View Post
Finesharp vs superres. I only did double scaling with nnedi3 (as quadrupling with superres doesn't work anyways for me). With double scaling (nnedi3 for chroma as well with superres there on defaults) finesharp with slightly tweaked settings (all defaults except mode set to 3, and linear light turned on) gives a (subjectively) noticably better result than superres (nnedi3 default settings), and lower rendering times too (perhaps more important in some ways obviously). I honestly didn't think upscaling could be this sharp, detailed and clean (for virtually any content I've thrown at it).
Thanks for the feedback, guys. But I do want to clear up debanding first, before moving to the other new options. FWIW, Shiandow is still working on maybe improving his debanding algorithm further. So let's give him a couple of days, and then retest, and then decide whether we keep his algorithm in madVR or not. And no, it will not stay as an option. It will either replace the current "high" preset, or it will go.

Quote:
Originally Posted by potomac View Post
madshi thanks for this huge work
i tested 10 bit on my cheap lg ips235 and i can see the improvement in the gray scale but my 5850 doesn't work well in dx11 mode
In which way does it not work well?

Quote:
Originally Posted by omarank View Post
Well, you invested so much time in developing and fine tuning the error diffusion algorithms. They are a premium feature of madVR, but you have disabled them for another premium feature – 10 bit output. I completely agree that there would be hardly any benefit in using Error diffusion dithering over ordered dithering for 10 bit output, but I believe it doesn’t harm either. Also, HDR displays will be available sooner or later, and error diffusion dithering at 10 bit would be beneficial again.
Quote:
Originally Posted by ryrynz View Post
I like how error diffusion looks on 8 bit over ordered dithering, I too would like to see this enabled for 10 bit so I can see for myself how it looks on my TV.
Being a plasma madVR's dithering patterns seem to affect it more than LCD's, It would certainly help in that instance having the keyboard shortcut I suggested above for comparisons.
Of course error diffusion doesn't harm quality, it does harm performance, though. Maybe I'll add a trade quality option to disable error diffusion in 10bit FSE mode? And allow it to be enabled if the trade quality option is not activated?

Quote:
Originally Posted by ryrynz View Post
Hey madshi could you allow for the bitdepth shortcut key to switch to 9 and 10 bit please? If I use the shortcut key it'll switch down from 10 to 9 and 8 bit
but it doesn't start back at 10 bit.
Oh well, this is not as easy as it might sound: The switch between true 10bit output and true 8bit output requires me to recreate the whole D3D11 devices etc. Currently when you use the shortcut keys to switch from 10bit down to lower bitdepths, madVR will still output native 10bit, just rounded/dithered to a lower bitdepth. That's not a really true comparison because the GPU and/or the display might behave differently between native 10bit output and native 8bit output. But switching between those on the fly isn't so easy.

Quote:
Originally Posted by ryrynz View Post
also could you add a shortcut to cycle dithering methods? Cheers.
I guess that should be possible. But the two extra options ("colored noise" etc) will not be cycled, just the algorithms themselves?

Quote:
Originally Posted by pirlouy View Post
With x64 build of 0.87.21 (and not x86), even if I disable "switch to matching display mode", madVR automatically changes refresh rate. Instead of 60Hz (define in Windows before launching movie), it automatically switches to 59Hz when I enter Fullscreen.
Quote:
Originally Posted by agustin9 View Post
I'm having a small problem using D11. My refresh rate changes from 59.99 in windowed to 59.93 in FSE, which makes the transition somehow slow.
This happens on my TV, and not on my monitor. I tried with the TV as main screen and it happens the exact same way. With D9 no problem.
Quote:
Originally Posted by huhn View Post
i have the same problem could be this bug but for d3d11:
http://bugs.madshi.net/view.php?id=90
FWIW, this is *not* a bug in madVR, but a bug in either D3D or in the OS itself. I partially succeeded in hacking the OS to fix this bug with D3D9, which may or may not work for D3D11. Maybe I'll find a way to fix this for D3D11, too. Just want it to be understood that it's *not* a bug in madVR.
madshi is offline   Reply With Quote
Old 12th May 2015, 09:14   #29730  |  Link
tyto
Registered User
 
Join Date: May 2014
Posts: 9
with D3D11 for any settings I can use I get 3-5 rendered frames. Doesnt really affects me much, 0 glitches, 0 dropped frames. I tried changing the "max prerendender" from nvidia panel but nothing changes.
That said, should I set triplebuffering from nvidia panel?
tyto is offline   Reply With Quote
Old 12th May 2015, 09:14   #29731  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
madVR v0.88.5 released

http://madshi.net/madVR.zip

Code:
* fixed: image quadrupling was buggy, sometimes crashed
* fixed: LumaSharpen strength was truncated (0.99 -> 0.00)
Seems like the most important bugs are fixed now. At least I hope so...
madshi is offline   Reply With Quote
Old 12th May 2015, 09:21   #29732  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
For those of you with refresh rate switching problems when using D3D11, does this test build fix things or change anything?

http://madshi.net/madVR885refreshRate.zip
madshi is offline   Reply With Quote
Old 12th May 2015, 09:43   #29733  |  Link
detmek
Registered User
 
Join Date: Aug 2009
Posts: 463
Thanks for the new features Madshi. But I have a small problem with 0.88 series.
I am testing all 0.88 versions and I have a problem with D3D11. When I switch from windowed mode to full screen or vice versa picture flickers for a few seconds (black frame/regular frame/black frame/regular ect). Also, it seams like frames are show out of order.
No problem with MPDN in D3D11 mode.

My PC spec:
Intel Pentium G3220 IGPU
4GB RAM
Windows 8.1 x64 updated
Driver: 10.18.14.4170 (latest)
MPC-HC/MPC-BE x64 latest nightly

Last edited by detmek; 12th May 2015 at 09:47.
detmek is offline   Reply With Quote
Old 12th May 2015, 09:45   #29734  |  Link
ryrynz
Registered User
 
ryrynz's Avatar
 
Join Date: Mar 2009
Posts: 3,646
madshi, are you seeing a ~90MB difference in GPU memory allocation when in 10 bit selecting ordered dithering vs error diffusion? (which shouldn't be the case considering error diffusion is disabled in 10 bit)

Quote:
Originally Posted by madshi View Post
Of course error diffusion doesn't harm quality, it does harm performance, though. Maybe I'll add a trade quality option to disable error diffusion in 10bit FSE mode? And allow it to be enabled if the trade quality option is not activated?
Sounds good.

Quote:
Originally Posted by madshi View Post
I guess that should be possible. But the two extra options ("colored noise" etc) will not be cycled, just the algorithms themselves?
Yeah that's fine.

Last edited by ryrynz; 12th May 2015 at 09:48.
ryrynz is offline   Reply With Quote
Old 12th May 2015, 10:14   #29735  |  Link
omarank
Registered User
 
Join Date: Nov 2011
Posts: 187
Quote:
Originally Posted by madshi View Post
Maybe I'll add a trade quality option to disable error diffusion in 10bit FSE mode? And allow it to be enabled if the trade quality option is not activated?
Yes, this will be ideal. Thanks!
omarank is offline   Reply With Quote
Old 12th May 2015, 10:55   #29736  |  Link
pirlouy
_
 
Join Date: May 2008
Location: France
Posts: 692
Quote:
Originally Posted by madshi View Post
FWIW, this is *not* a bug in madVR, but a bug in either D3D or in the OS itself. I partially succeeded in hacking the OS to fix this bug with D3D9, which may or may not work for D3D11. Maybe I'll find a way to fix this for D3D11, too. Just want it to be understood that it's *not* a bug in madVR.
Ok I totally believe you on the fact it is a Windows/driver bug. But on 0.87.21, I though there were not D3D11 yet.
What is strange is that this behavior is present in x64 build, not in x86.

I'll do tests with 0.88 later.
pirlouy is offline   Reply With Quote
Old 12th May 2015, 11:08   #29737  |  Link
cvrkuth
Registered User
 
Join Date: Apr 2011
Posts: 54
Quote:
Originally Posted by madshi View Post
For those of you with refresh rate switching problems when using D3D11, does this test build fix things or change anything?

http://madshi.net/madVR885refreshRate.zip
No success here
__________________
Intel Core i7-4790 CPU @ 3.60GHz, RAM 32 GB Dual-Channel DDR3 @ 665MHz (9-9-9-24),
Panasonic TX-P42G20E, NVIDIA GeForce GTX 970, Win 10 Pro x64,
PotPlayer 1.7.16291 64-bit, madVR v0.92.17
cvrkuth is offline   Reply With Quote
Old 12th May 2015, 12:49   #29738  |  Link
cca
Anime Otaku
 
Join Date: Oct 2002
Location: Somewhere in Cyberspace...
Posts: 437
Quote:
Originally Posted by madshi View Post
For those of you with refresh rate switching problems when using D3D11, does this test build fix things or change anything?

http://madshi.net/madVR885refreshRate.zip
Tried it, no difference, instead of 24Hz it still switches to 23.97.
__________________
AMD FX8350 on Gigabyte GA-970A-D3 / 8192 MB DDR3-1600 SDRAM / AMD R9 285 with Catalyst 1.5.9.1/ Asus Xonar D2X / Windows 10 pro 64bit
cca is offline   Reply With Quote
Old 12th May 2015, 12:55   #29739  |  Link
Sm3n
Registered User
 
Join Date: Jul 2012
Posts: 94
thx for all the updates
Sm3n is offline   Reply With Quote
Old 12th May 2015, 14:13   #29740  |  Link
huhn
Registered User
 
Join Date: Oct 2012
Posts: 7,903
Quote:
Originally Posted by madshi View Post
For those of you with refresh rate switching problems when using D3D11, does this test build fix things or change anything?

http://madshi.net/madVR885refreshRate.zip
doesn't work for me 60p switches to 59p
huhn is offline   Reply With Quote
Reply

Tags
direct compute, dithering, error diffusion, madvr, ngu, nnedi3, quality, renderer, scaling, uhd upscaling, upsampling

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +1. The time now is 19:00.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2024, vBulletin Solutions Inc.