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 December 2016, 12:46   #41521  |  Link
ryrynz
Registered User
 
ryrynz's Avatar
 
Join Date: Mar 2009
Posts: 3,090
Quote:
Originally Posted by Werewolfy View Post
I'm interested to see more screenshots. It's a great example to show that chroma does count.

Could you post screenshots of this scene with Jinc or NNEDI3 32/64 neurons just to see the results? I didn't expect that super-xbr was so good actually, I aslo find it better than NGU very high in this example.
Thanks. This isn't just limited to red scenes either, but this scene really does look to be a good test for comparisons of chroma scalers and is a full quality rip of the blu-ray.

As you'd expect based on what I've said previously NNEDI3 32 neurons and 64 neurons are almost identical, 64 is a waste of resources IMO :P
I think NNEDI3 has the edge here but it's slight.

Jinc AR vs NNEDI3 64 neurons.
NNEDI3 32 neurons vs NNEDI3 64 neurons.
ryrynz is offline   Reply With Quote
Old 15th December 2016, 13:22   #41522  |  Link
Patrik G
Registered User
 
Join Date: Jan 2015
Location: Sweden
Posts: 58
Quote:
Originally Posted by ryrynz View Post
Can post more scenes if anyone would like to see them.

Check it out.
Thats useless as it also depends on the display you use.
And they also performs different in motion.
Static screenshots and the image you get when the movie is running is not the same.
Plasma tvs for example has much higher motion resolution than lcds or oleds and so they will show a sharper image.

You are not maximizing madvrs upscaling algorithms on an LCD or OLED lol
300 lines is sub DVD resolution in motion.
Plasma tvs can show all 1080 lines from Blu Ray content

Last edited by Patrik G; 15th December 2016 at 13:28.
Patrik G is offline   Reply With Quote
Old 15th December 2016, 13:22   #41523  |  Link
Werewolfy
Registered User
 
Join Date: Feb 2013
Posts: 135
Quote:
Originally Posted by ryrynz View Post
Thanks. This isn't just limited to red scenes either, but this scene really does look to be a good test for comparisons of chroma scalers and is a full quality rip of the blu-ray.

As you'd expect based on what I've said previously NNEDI3 32 neurons and 64 neurons are almost identical, 64 is a waste of resources IMO :P
I think NNEDI3 has the edge here but it's slight.

Jinc AR vs NNEDI3 64 neurons.
NNEDI3 32 neurons vs NNEDI3 64 neurons.
Thanks! I think too that NNEDI3 64 neurons is a waste of resources for chroma, I just wanted to check.

Super-xbr 150 is the sharpest between all these options but I find that Jinc is still a great performer when you consider that it doesn't need much resources. NGU low could be a great deal too.

Quote:
Originally Posted by Neo-XP View Post
Yes, be careful with Bilateral, it can give you the best, but also the worst results depending on the situation.

Super-xbr is much more polyvalent, with values of 100-150.
I use NGU-Med at the moment (or NGU-High on new versions), because it is overall sharper and cleaner... and it is cold now where I live

Example with BilateralSharp / Bicubic60 AR / super-xbr150 / NGU-Low / NGU-Med / NGU-High :



I zoom for you in the right spot :



Now I let you decide which one to use.
But in this scene, I find NGU high a lot better than Super-xbr 150, even NGU low is better. The pearls are more defined with NGU.
__________________
Windows 8.1 x64 - Intel Core i5-4670K (4.2 GHz) - 8 GB DDR3 - MSI Geforce GTX 1080 8 GB - Sony KD-55A1

Last edited by Werewolfy; 15th December 2016 at 13:27.
Werewolfy is offline   Reply With Quote
Old 15th December 2016, 14:21   #41524  |  Link
SpoCk0nd0pe
Registered User
 
Join Date: Jun 2015
Posts: 25
In the screens by Neo-XP I actually think BilateralSharp looks best. Compare the red on the neck. All other algorithms seem to bleed the red into the surrounding skin colour. I'd be interested in how reconstruction soft performed in that image.
SpoCk0nd0pe is offline   Reply With Quote
Old 15th December 2016, 18:59   #41525  |  Link
Asmodian
Registered User
 
Join Date: Feb 2002
Location: San Jose, California
Posts: 3,439
Quote:
Originally Posted by Oguignant View Post
Render times are very bad. I do not understand why.
This gpu is the fastest I found.
Those look fine actually, I was hoping NNEDI3 128 would be faster for you but that isn't too far off. I have a Titan X (Pascal), OC 2050 MHz.
__________________
madVR options explained
Asmodian is online now   Reply With Quote
Old 15th December 2016, 19:01   #41526  |  Link
AngelGraves13
Registered User
 
Join Date: Dec 2010
Posts: 239
Quote:
Originally Posted by SpoCk0nd0pe View Post
In the screens by Neo-XP I actually think BilateralSharp looks best. Compare the red on the neck. All other algorithms seem to bleed the red into the surrounding skin colour. I'd be interested in how reconstruction soft performed in that image.
Look again...the reds are bleeding together in Bilateral and there is no separation between the bright and dark red beads.
AngelGraves13 is offline   Reply With Quote
Old 15th December 2016, 19:45   #41527  |  Link
SpoCk0nd0pe
Registered User
 
Join Date: Jun 2015
Posts: 25
Quote:
Originally Posted by AngelGraves13 View Post
Look again...the reds are bleeding together in Bilateral and there is no separation between the bright and dark red beads.
Yes they are. But the red on the neck bleeds into the skin colour (the red part slightly to the left between the two black parts). It is a very noticeable artefact imho.
SpoCk0nd0pe is offline   Reply With Quote
Old 15th December 2016, 21:16   #41528  |  Link
AngelGraves13
Registered User
 
Join Date: Dec 2010
Posts: 239
Quote:
Originally Posted by SpoCk0nd0pe View Post
Yes they are. But the red on the neck bleeds into the skin colour (the red part slightly to the left between the two black parts). It is a very noticeable artefact imho.
I'm sure madshi will improve NGU when he adds the luma lookup into it.
AngelGraves13 is offline   Reply With Quote
Old 15th December 2016, 21:43   #41529  |  Link
Oguignant
Registered User
 
Oguignant's Avatar
 
Join Date: Nov 2016
Posts: 181
Quote:
Originally Posted by Asmodian View Post
Those look fine actually, I was hoping NNEDI3 128 would be faster for you but that isn't too far off. I have a Titan X (Pascal), OC 2050 MHz.
Titan x Pascal Is more powerfull. Significantly more render output processors 96 vs 64 and more cuda cores 3,072 vs 2,560...

Gtx 1080 + madvr with new movies, like guardians of the galaxy, NGU It does very well. The colors and details look great on my tv oled 4k tv...

Chroma upscaling NGU high
Image upscaling luma + chroma NGU very high
Crop black bars
Upscaling refinement enhance detail 1.5
Others option let madvr decide
Render time 26 ms
__________________
"To infinity, and beyond!"

Last edited by Oguignant; 15th December 2016 at 22:13.
Oguignant is offline   Reply With Quote
Old 15th December 2016, 22:00   #41530  |  Link
Telion
Registered User
 
Join Date: Sep 2011
Posts: 78
madshi, can you please make a small change in switching the debanding on/off by keyboard? I've requested this around a year ago and you've answered that you will take a look at it but it might take some time. So I'm just meekly reminding about it and begging you to implement this during your current work on making the UI a bit more streamlined:
Quote:
Originally Posted by Telion View Post
I usually use either a high debanding or no debanding during several files in a series. If I'd switch through GUI, it's all plain and simple - the radiobutton is always at "high" position and I can just toggle only the checkbox. Obviously I want to speed up things by using keyboard shortcuts, but the different story begins here. For example, I start from debanding at "high" but off and want to turn it on. So I press Ctlr-Alt-Shift-A twice (OSD shows off->high) and then hit F2 to save. Everything goes as expected. Afterwards, I want to switch it back off and repeat the same sequence, OSD showing high->off. But now it turns out that not only the checkbox was toggled - the radiobutton was also set to "low" value, so next time I'd need to press Ctlr-Alt-Shift-A four times, going all the way through off->low->medium->high. I humbly ask you to change this behaviour so that the radiobutton won't get reset from its initial value if the final result of keypresses is "off" at the moment of releasing all keys in the shortcut combination. It's not that I'm too bothered with two extra keypresses - I just want to get used to the same keypress sequence for toggling debanding on and off and pay no attention to its initial state or OSD.
Telion is offline   Reply With Quote
Old 15th December 2016, 22:16   #41531  |  Link
ryrynz
Registered User
 
ryrynz's Avatar
 
Join Date: Mar 2009
Posts: 3,090
Quote:
Originally Posted by Patrik G View Post
Thats useless as it also depends on the display you use.
And they also performs different in motion.
Static screenshots and the image you get when the movie is running is not the same.
Plasma tvs for example has much higher motion resolution than lcds or oleds and so they will show a sharper image.

You are not maximizing madvrs upscaling algorithms on an LCD or OLED lol
300 lines is sub DVD resolution in motion.
Plasma tvs can show all 1080 lines from Blu Ray content
Jesus Christ. By that reasoning we shouldn't be using screenshots at all.. Nothing matters because everyone's display is different. What a load of tripe!

The results are consistent across everyone's display regardless of whether you see a difference or not! The better the display and the better optimised it is the more obvious the changes will be.

Different displays don't invalidate screen shots at all. Might as well tell madshi to stop taking screenshots based on the above nonsense.

Last edited by ryrynz; 15th December 2016 at 23:17.
ryrynz is offline   Reply With Quote
Old 15th December 2016, 22:19   #41532  |  Link
Damie
Registered User
 
Join Date: Sep 2008
Posts: 21
HEVC 4K, colors off in fullscreen

Hello,

why are the colors blown off (especially blue) in fullscreen with the following HEVC 4K video & MadVR?

https://cloud.mail.ru/public/FsZX/4GqUibRKS (click the title to download the sample video)

I'm using Sony X850C as the display (10bit & HDR & BT.2020 compatible).

edit: It seems that DXVA2 image downscaling is disabled @ windowed mode, so somehow DXVA2 processing & MadVR fails at producing correct colors with this file.
I'm using the latest MadVR & Lav 0.69.0 & MPC-BE 1.5 x86.
The file plays fine using Windows Media Player using the default GPU decoding (GTX 960).

Thanks for any help!

Last edited by Damie; 16th December 2016 at 00:57.
Damie is offline   Reply With Quote
Old 16th December 2016, 02:35   #41533  |  Link
AngelGraves13
Registered User
 
Join Date: Dec 2010
Posts: 239
Quote:
Originally Posted by Damie View Post
Hello,

why are the colors blown off (especially blue) in fullscreen with the following HEVC 4K video & MadVR?

https://cloud.mail.ru/public/FsZX/4GqUibRKS (click the title to download the sample video)

I'm using Sony X850C as the display (10bit & HDR & BT.2020 compatible).

edit: It seems that DXVA2 image downscaling is disabled @ windowed mode, so somehow DXVA2 processing & MadVR fails at producing correct colors with this file.
I'm using the latest MadVR & Lav 0.69.0 & MPC-BE 1.5 x86.
The file plays fine using Windows Media Player using the default GPU decoding (GTX 960).

Thanks for any help!
DXVA scaling should be avoided. Try using Jinc for everything and report back.
AngelGraves13 is offline   Reply With Quote
Old 16th December 2016, 09:37   #41534  |  Link
Oguignant
Registered User
 
Oguignant's Avatar
 
Join Date: Nov 2016
Posts: 181
trade quality for performance options

Quote:
Originally Posted by Asmodian View Post
Those look fine actually, I was hoping NNEDI3 128 would be faster for you but that isn't too far off. I have a Titan X (Pascal), OC 2050 MHz.
@Asmodian (or someone expert), if you are so kind, pls tell me what options you have enabled in "Trade quality for performance".

These are my options. I think I'm missing something...



thks!!!
__________________
"To infinity, and beyond!"
Oguignant is offline   Reply With Quote
Old 16th December 2016, 10:02   #41535  |  Link
fedpul
Registered User
 
Join Date: Feb 2014
Posts: 93
Quote:
Originally Posted by Oguignant View Post
@Asmodian (or someone expert), if you are so kind, pls tell me what options you have enabled in "Trade quality for performance".

These are my options. I think I'm missing something...



thks!!!
Hi, for maximum picture quality you should untick everything there.

Enviado desde mi MI 4LTE mediante Tapatalk
fedpul is offline   Reply With Quote
Old 16th December 2016, 10:05   #41536  |  Link
ryrynz
Registered User
 
ryrynz's Avatar
 
Join Date: Mar 2009
Posts: 3,090
Quote:
Originally Posted by Oguignant View Post
These are my options. I think I'm missing something...
Missing something??
Nothing you have ticked really changes much of anything, certainly nothing you'd notice unless your using debanding or DXVA even then it's questionable.
If you want full quality like I said only a few pages back, untick the lot that's what most here have done unless they have slow/dated hardware.
ryrynz is offline   Reply With Quote
Old 16th December 2016, 10:25   #41537  |  Link
Oguignant
Registered User
 
Oguignant's Avatar
 
Join Date: Nov 2016
Posts: 181
Quote:
Originally Posted by fedpul View Post
Hi, for maximum picture quality you should untick everything there.

Enviado desde mi MI 4LTE mediante Tapatalk
Thks, hablas espaņol? Disabling everything, the render time goes up by 3 or 4 ms (29 ms max total render time for 2k to 4k with NGU Very high), it's fine for me.
How do you use it? everything disabled too? My gtx 1080 g1 cries!

__________________
"To infinity, and beyond!"

Last edited by Oguignant; 16th December 2016 at 10:28.
Oguignant is offline   Reply With Quote
Old 16th December 2016, 11:27   #41538  |  Link
Oguignant
Registered User
 
Oguignant's Avatar
 
Join Date: Nov 2016
Posts: 181
About dithering

Quick questiin about dithering...

On 10-bit tv screens with full screen D3D11 exclusive mode is better without dithering? Or is it always better to use ordered dithering?
__________________
"To infinity, and beyond!"
Oguignant is offline   Reply With Quote
Old 16th December 2016, 11:35   #41539  |  Link
huhn
Registered User
 
Join Date: Oct 2012
Posts: 5,507
never disable dithering.
huhn is offline   Reply With Quote
Old 16th December 2016, 11:40   #41540  |  Link
Oguignant
Registered User
 
Oguignant's Avatar
 
Join Date: Nov 2016
Posts: 181
Quote:
Originally Posted by huhn View Post
never disable dithering.
For that reason I ask. I read in a forum that with tv 10-bit panel + d3d11 in fullscreen exclusive mode disable it. What happens if I use it disabled? TV breaks or something?
__________________
"To infinity, and beyond!"
Oguignant 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 21:33.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2019, vBulletin Solutions Inc.