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 15th October 2013, 18:46   #20401  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
Quote:
Originally Posted by Skankee View Post
Unfortunately not. I tested linear light with low and high debanding, no difference visible to me.
Ok, good news, that saves rendering time and options.

Quote:
Originally Posted by Skankee View Post
But i found out that i also get these brighter areas when i use dithering (it is disabled on my previous screenshots because of better blacks and a reduced filesize).

So maybe my problem is related to dither and not to debanding.

With dithering (and disabled debanding) there is noticeable less banding in my video-sample, so maybe it is not so good for medium dithering-tests anymore, i don't know.

But there are still 2 questions left:
-can i disable dithering when i use low debanding (because both do some debanding),
-are my "dark areas" in my test-images a bug because of missing dithering or is dithering doing something "strange" and create too bright areas?
You should *NEVER* *EVER* disable dithering, especially not when doing anything image quality related. Doing so will introduce banding and quantization errors. If those areas are darker with dithering disabled then that's an error caused by disabled dithering. Dithering is not doing anything strange, it just gets rid of quantization errors on the cost of increased noise floor.

Just think about it yourself: If you have a large image area which after YCbCr -> RGB conversion has an 8bit brightness of e.g. 80.4, then without dithering every pixel in that image area will be rounded down to 80.0. So the average brightness of the whole image area will be 80.0. If you enable dithering (as you should!) some pixels will be drawn as 80.0 and some with 81.0, resulting in an average brightness of 80.4. So only by using dithering you'll get the correct image brightness.

Dithering does not deband anything, it just avoids quantization errors. And quantization errors can *cause* banding artifacts. So enabling dithering doesn't repair/improve anything. It's the other way round: Disabling dithering can introduce severe image artifacts.

It doesn't matter whether you enable debanding or not, you should never disable dithering.
madshi is offline   Reply With Quote
Old 16th October 2013, 05:09   #20402  |  Link
andybkma
Registered User
 
Join Date: Sep 2006
Posts: 212
Quote:
Originally Posted by madshi View Post

You should *NEVER* *EVER* disable dithering, especially not when doing anything image quality related. Doing so will introduce banding and quantization errors.
madshi, quick question since you are on the subject of dithering. I use ffdshow raw video processor as a post processor only with avisynth script "LimitedSharpenfaster" after decoding with LAV before using mVR as the renderer. In the ffdshow raw settings under the RGB Conversion tab there are two options; one for "High Quality YV12 to RGB Conversion" and the other for "Dithering". Should these be checked or unchecked? I am unsure as I thought that mVR does its own conversion and dithering (?) I am only using the avisynth script with the raw ffdshow, nothing else. Thanks much
andybkma is offline   Reply With Quote
Old 16th October 2013, 06:02   #20403  |  Link
ryrynz
Registered User
 
ryrynz's Avatar
 
Join Date: Mar 2009
Posts: 3,646
Quote:
Originally Posted by andybkma View Post
In the ffdshow raw settings under the RGB Conversion tab there are two options; one for "High Quality YV12 to RGB Conversion" and the other for "Dithering". Should these be checked or unchecked?
Checking these won't change a thing unless you're outputting RGB, which you shouldn't be, doing so will actually decrease quality.
If you ever needed to output RGB then most certainly you'd want these ticked, without them enabled the image quality is quite bad.

Check with madVR (Ctrl-J fourth line) it should show YV12 or NV12, personally I stick with YV12.
I don't believe MadVR benefits any from working with NV12, so it's a waste of resources converting to it.

Last edited by ryrynz; 16th October 2013 at 06:34.
ryrynz is offline   Reply With Quote
Old 16th October 2013, 06:48   #20404  |  Link
THX-UltraII
Registered User
 
Join Date: Aug 2008
Location: the Netherlands
Posts: 851
Quote:
Originally Posted by THX-UltraII View Post
I m not talking about the problem FULL and LIMITED. This already works out-of-the-box with the latest Intel driver (you can simply select AUTO, FULL or LIMITED in the Intel Control Panel and it works perfect).

I am talking about huhn saying that Intel cards output ycbcr and not rgb and that this is supposed to affect PQ in a bad way when using madVR.
can you come back on this subject huhn?
THX-UltraII is offline   Reply With Quote
Old 16th October 2013, 10:34   #20405  |  Link
Francois76l
Registered User
 
Join Date: Oct 2013
Location: France - Paris
Posts: 21
Smooth motion

Hello Madshi!

First, thanks for your amazing work along the years!

I've one question concerning the smooth motion option under madVr v0.86.11.
On my computer (i7 3770k @4.6Ghz + 660GTX Ti + 8Go ram) The picture quality is bad with this option (@60Hz). There is no difference in the picture fluidity and sometimes I can see some frame doubling. It's really strange.

I'm using Windows 7 Pro N (special version without Windows Media Player and Media Center) Do you think my problem can be link?

Thanks again for your help

++
Francois76l is offline   Reply With Quote
Old 16th October 2013, 17:59   #20406  |  Link
huhn
Registered User
 
Join Date: Oct 2012
Posts: 7,903
Quote:
Originally Posted by THX-UltraII View Post
can you come back on this subject huhn?
ok i try it.

i said output YCbCr is not the ideal way because you asked about a way to output this because you heard you ~beamer?~ got problems with rgb right?

so i talked about about a way to output YCbCr with madvr because i heard it's possible (i still don't know it's true).

so madvr outputs 8 bit 0-255 rgb at default settings. if your gpu outputs rgb unlimited noting should happen.

so i don't know what your plan is right now.

are you going with your new amd card or not.

than there was something about rgb YCbCr output in intel.

you can just switch to both in the driver (never heard that this is not working)

but outputing rgb 0-255 was a or is a problem with intel there is no option for this (something about "it"or normal content but this does nothing...) and it's a well known problem.

don't mix that up with video output 16-235 or 0-255.

even if you are going to use limited rgb you still have to use a 16-235 to 0-255 (madvr does this y default) because the desktop is 0-255 and both together needs to be limited again. not the best way to do this but yeah there are plenty of people in this forum with much better english and much better knowledge of this topic then me.

so what's your problem in the detail?
huhn is offline   Reply With Quote
Old 16th October 2013, 23:52   #20407  |  Link
nautilus7
Registered User
 
nautilus7's Avatar
 
Join Date: Jan 2006
Location: Athens, Greece
Posts: 1,518
Hi, nice to see the development is going on! Nice things are coming all the time!

I need your help selecting a good yet not expensive videocard for use with madvr. I need cuda so I am looking for an nvidia. Is a gtx 650 Ti Boost a good one for madvr? Do you recomend another one at this range of money?
nautilus7 is offline   Reply With Quote
Old 17th October 2013, 00:31   #20408  |  Link
huhn
Registered User
 
Join Date: Oct 2012
Posts: 7,903
why cuda dxva is much faster and near if not everything what works with cuda works with dxva too and this leaves more breathing room for madvr.


660 ti 1344 cuda cores 185 €
650 ti 768 cuda cores 140 €

i think the 650 ti is very bad for it's money.

you may look at the new amd gpu

edit: or have a look at an richland apu

Last edited by huhn; 17th October 2013 at 00:34.
huhn is offline   Reply With Quote
Old 17th October 2013, 16:46   #20409  |  Link
detmek
Registered User
 
Join Date: Aug 2009
Posts: 463
What do you mean, dxva is much faster? DXVA and CUVID use the same ASIC. There may be some differences but "much faster" is simply not true. Also, madVR does not use ASIC but shaders (main GPU) so there is no competition for same resources between madVR and decoding.
detmek is offline   Reply With Quote
Old 17th October 2013, 16:53   #20410  |  Link
shimaflarex
Registered User
 
Join Date: Oct 2011
Posts: 41
Quote:
Originally Posted by detmek View Post
there is no competition for same resources between madVR and decoding
They both compete for VRAM (bandwidth and memory usage).

nautilus7 said he needed CUDA(not CUVID). He may want to use CUDA for other proposes besides video decoding/encoding.
shimaflarex is offline   Reply With Quote
Old 17th October 2013, 17:17   #20411  |  Link
baii
Registered User
 
Join Date: Dec 2011
Posts: 180
Not sure about decoding, but hw deinterlacing definitely use up some noticeable gpu power.
baii is offline   Reply With Quote
Old 17th October 2013, 17:18   #20412  |  Link
huhn
Registered User
 
Join Date: Oct 2012
Posts: 7,903
CUVID forces the gpu in max power state and is more like dxva copy back.

Quote:
nautilus7 said he needed CUDA(not CUVID). He may want to use CUDA for other proposes besides video decoding/encoding.
very good point
huhn is offline   Reply With Quote
Old 17th October 2013, 21:31   #20413  |  Link
e-t172
Registered User
 
Join Date: Jan 2008
Posts: 589
Quote:
Originally Posted by detmek View Post
What do you mean, dxva is much faster? DXVA and CUVID use the same ASIC. There may be some differences but "much faster" is simply not true.
When using CUVID and DXVA copy-back the decoded video passes back and forth between the GPU and the CPU (well, between video memory and main memory, to be precise). With DXVA native the video stays on the GPU from decoding to rendering, which makes it somewhat faster.
e-t172 is offline   Reply With Quote
Old 17th October 2013, 22:25   #20414  |  Link
SoulSmasher
Registered User
 
Join Date: Jan 2005
Posts: 26
Hi,

I've a problem, hope you can help me.

Ever since I've updated to Windows 8.1, when I go to exclusive mode with MPC-HC (using LAV filters), I get a red message saying "out of memory" in a red font in a black frame, but the video runs fine.

Here's a sample screenshot (sorry, the frame is not shown in screenshot, I don't know why).

http://imgur.com/YHlXJ4X

What could be the issue? And how do I fix this? I've already re-installed LAV filters, it's still same.

Edit: Weirdly, after re-installing (clean) the NVidia driver (although it was showing official driver is already installed at GeForce Experience panel), it's fixed.

Thanks again for the awesome renderer!
__________________
Spirit... It's the hero's strength, the mother's restlince and the poor man's armor. It cannot be broken and it cannot be taken away.
This I must believe
Drizzt

Last edited by SoulSmasher; 17th October 2013 at 22:32.
SoulSmasher is offline   Reply With Quote
Old 17th October 2013, 23:29   #20415  |  Link
detmek
Registered User
 
Join Date: Aug 2009
Posts: 463
Quote:
Originally Posted by e-t172 View Post
When using CUVID and DXVA copy-back the decoded video passes back and forth between the GPU and the CPU (well, between video memory and main memory, to be precise). With DXVA native the video stays on the GPU from decoding to rendering, which makes it somewhat faster.
Depends on GPU. On my Nvidia 9500GT (a bit old card) DXVA Native gives me highest rendering times with madVR, and DXVA Copy-back lowest. CUVID is in the middle if we compare rendering times.

On the other hand, I was talking about decoding speed on ASIC which is basically the same for any method. So, it is not "much faster", that is all I wanted to comment.

Also, sometimes it is better to put GPU in max power mode because some users reported frame drops because GPU goes to lower power mod and does not have power for madVR rendering job.
detmek is offline   Reply With Quote
Old 17th October 2013, 23:48   #20416  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
Quote:
Originally Posted by Francois76l View Post
I've one question concerning the smooth motion option under madVr v0.86.11.
On my computer (i7 3770k @4.6Ghz + 660GTX Ti + 8Go ram) The picture quality is bad with this option (@60Hz). There is no difference in the picture fluidity and sometimes I can see some frame doubling. It's really strange.
Are you using Windows mode, Overlay mode or Fullscreen Exclusive Mode? For some people, smooth motion only really reliably works in Fullscreen Exclusive Mode. So try whether that fixes the problem. If not, does the OSD report presentation glitches or dropped frames?

Quote:
Originally Posted by SoulSmasher View Post
Edit: Weirdly, after re-installing (clean) the NVidia driver (although it was showing official driver is already installed at GeForce Experience panel), it's fixed.
Yeah, weird...

-------------------------

I've found a quite promising new tweak for the debanding algorithm. So all ya debanding geeks out there, here's a new test build:

http://madshi.net/madVRdeband6.rar

The new feature is called "angleBoost" and you can control it via Ctrl+Alt+F9/F10 (decrease/increase) and Ctrl+Alt+Shift+F9/F10 (off/on). This feature checks whether the surrounding pixels all have the same gradient angle. If they do, the other parameters (avgDif, maxDif etc) are all multiplied with the "angleBoost" factor. So a factor of 1.0 means no change. A factor of 2.0 means double parameters.

It seems that this new feature might be able to separate details from gradients, so I hope we'll get a nice debanding improvement without losing any image detail. On the content I've tried so far an angleBoost of 2.0 seems to work quite nicely, although that might be a tad too high. Let me know what you think...

P.S: The angleBoost feature does consume a bit of additional rendering power. But from my first tests it seems to be more than worth it.
madshi is offline   Reply With Quote
Old 18th October 2013, 00:12   #20417  |  Link
nautilus7
Registered User
 
nautilus7's Avatar
 
Join Date: Jan 2006
Location: Athens, Greece
Posts: 1,518
Quote:
Originally Posted by huhn View Post
660 ti 1344 cuda cores 185 €
650 ti 768 cuda cores 140 €

i think the 650 ti is very bad for it's money.
Not exactly, my friend.

Code:
660 Ti      1344 cores ~235€
660          960 cores ~175€
650 Ti Boost 768 cores ~160€
650 Ti       768 cores ~125€
Quote:
Originally Posted by shimaflarex View Post
nautilus7 said he needed CUDA(not CUVID). He may want to use CUDA for other proposes besides video decoding/encoding.
Yes, exactly.

Any recomendation?

Last edited by nautilus7; 18th October 2013 at 00:19.
nautilus7 is offline   Reply With Quote
Old 18th October 2013, 02:58   #20418  |  Link
huhn
Registered User
 
Join Date: Oct 2012
Posts: 7,903
sry in germany you can get a 660 ti for 185

235 wow ...

in your case the 650 ti NON boost.

or think about an import
huhn is offline   Reply With Quote
Old 18th October 2013, 05:12   #20419  |  Link
6233638
Registered User
 
Join Date: Apr 2009
Posts: 1,019
Quote:
Originally Posted by madshi View Post
I've found a quite promising new tweak for the debanding algorithm. So all ya debanding geeks out there, here's a new test build:

http://madshi.net/madVRdeband6.rar

The new feature is called "angleBoost" and you can control it via Ctrl+Alt+F9/F10 (decrease/increase) and Ctrl+Alt+Shift+F9/F10 (off/on). This feature checks whether the surrounding pixels all have the same gradient angle. If they do, the other parameters (avgDif, maxDif etc) are all multiplied with the "angleBoost" factor. So a factor of 1.0 means no change. A factor of 2.0 means double parameters.

It seems that this new feature might be able to separate details from gradients, so I hope we'll get a nice debanding improvement without losing any image detail. On the content I've tried so far an angleBoost of 2.0 seems to work quite nicely, although that might be a tad too high. Let me know what you think...

P.S: The angleBoost feature does consume a bit of additional rendering power. But from my first tests it seems to be more than worth it.
I might not get a chance to do much testing until the weekend now, but I'll do what I can.

Can we get a build which lets us customize the keyboard shortcuts though? I find them increasingly difficult to use.

I'd much rather be using QA, WS, ED, RF etc.
6233638 is offline   Reply With Quote
Old 18th October 2013, 07:20   #20420  |  Link
AngelGraves13
Registered User
 
Join Date: Dec 2010
Posts: 255
Quote:
Originally Posted by madshi View Post
Are you using Windows mode, Overlay mode or Fullscreen Exclusive Mode? For some people, smooth motion only really reliably works in Fullscreen Exclusive Mode. So try whether that fixes the problem. If not, does the OSD report presentation glitches or dropped frames?


Yeah, weird...

-------------------------

I've found a quite promising new tweak for the debanding algorithm. So all ya debanding geeks out there, here's a new test build:

http://madshi.net/madVRdeband6.rar

The new feature is called "angleBoost" and you can control it via Ctrl+Alt+F9/F10 (decrease/increase) and Ctrl+Alt+Shift+F9/F10 (off/on). This feature checks whether the surrounding pixels all have the same gradient angle. If they do, the other parameters (avgDif, maxDif etc) are all multiplied with the "angleBoost" factor. So a factor of 1.0 means no change. A factor of 2.0 means double parameters.

It seems that this new feature might be able to separate details from gradients, so I hope we'll get a nice debanding improvement without losing any image detail. On the content I've tried so far an angleBoost of 2.0 seems to work quite nicely, although that might be a tad too high. Let me know what you think...

P.S: The angleBoost feature does consume a bit of additional rendering power. But from my first tests it seems to be more than worth it.
Setting view does not go away. If I enable it, it just floats forever on top of the picture.
AngelGraves13 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 09:12.


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