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 7th May 2016, 08:53   #37781  |  Link
QBhd
QB the Slayer
 
QBhd's Avatar
 
Join Date: Feb 2011
Location: Toronto
Posts: 697
Quote:
Originally Posted by huhn View Post
but image doubling will have an effect on image quality unlike nnedi3 chroma at higher neurons.
Exactly, spare those Chroma Neurons and spend them on NNEDI3 Image Doubling. I only bump my Chroma high when there is nothing left to give on Image. When I set up profiles, I always start with Image Doubling and see how high it can go, then if there is room to spare do I add to Chroma. I am sure Betroz could easily drop to 32 Chroma and get some NNEDI3 goodness out of Image Dougling

Quote:
Originally Posted by Betroz View Post
I know, but super-xbr anti-bloat 25% is good for image doubling too, and the fans on my 970 card doesn't spin up as much as they do with NNEDI3. A good compromise I think. Chroma Upscaling 128 is nice to have with 1080p movies that run 1:1 (no image scaling).
If the fans are not spinning up for Chroma 128, I highly doubt that it's even being used... Can you post a Screen shot of the OSD with that particular setup?

QB
__________________

Last edited by QBhd; 7th May 2016 at 08:56.
QBhd is offline   Reply With Quote
Old 7th May 2016, 09:48   #37782  |  Link
Betroz
Is this for real?
 
Betroz's Avatar
 
Join Date: Mar 2016
Location: Norway
Posts: 168
Quote:
Originally Posted by QBhd View Post
If the fans are not spinning up for Chroma 128, I highly doubt that it's even being used... Can you post a Screen shot of the OSD with that particular setup?
QB
It is, but Chroma 128 doesn't tax my card as hard as Image doubling, at least not when super-xbr is used for doubling/quad. I do have tried diffferent settings...
__________________
My HTPC : i9 10900K | nVidia RTX 4070 Super | TV : Samsung 75Q9FN QLED
Betroz is offline   Reply With Quote
Old 7th May 2016, 14:05   #37783  |  Link
Sunset1982
Registered User
 
Join Date: Sep 2014
Posts: 280
just read the new presentation of the new geforce cards. 1070 as strong as 980ti and 1080 stronger than titan... seems we will get a good jump in performance. Now lets see what amd got and then we'll see how much benefits madvr will get from this more powerfull gpu's when upsaling 1080p to 4k on my machine.
__________________
Intel i5 6600, 16 GB DDR4, AMD Vega RX56 8 GB, Windows 10 x64, Kodi DS Player 17.6, MadVR (x64), LAV Filters (x64), XySubfilter .746 (x64)
LG 4K OLED (65C8D), Denon X-4200 AVR, Dali Zensor 5.1 Set
Sunset1982 is offline   Reply With Quote
Old 7th May 2016, 14:55   #37784  |  Link
wanezhiling
Registered User
 
Join Date: Apr 2011
Posts: 1,184
1070 is even stronger than titan x.
wanezhiling is offline   Reply With Quote
Old 7th May 2016, 15:02   #37785  |  Link
Uoppi
Registered User
 
Join Date: Oct 2015
Posts: 99
I currently use Super XBR 75 for doubling with some profiles where I can't afford NNEDI3. I'm considering whether I should try switching to Super XBR anti-bloat 25 because many seem to like it.

Any pointers as to the strengths/weaknesses of each, i.e. with what material would I best see their benefits or drawbacks? I've understood they should look pretty close.
Uoppi is offline   Reply With Quote
Old 7th May 2016, 15:34   #37786  |  Link
BluesFanUK
Registered User
 
Join Date: Sep 2015
Posts: 60
Quote:
Originally Posted by Sunset1982 View Post
just read the new presentation of the new geforce cards. 1070 as strong as 980ti and 1080 stronger than titan... seems we will get a good jump in performance. Now lets see what amd got and then we'll see how much benefits madvr will get from this more powerfull gpu's when upsaling 1080p to 4k on my machine.
Neither card is a substantial enough upgrade over a 980Ti to get more performance out of MadVR. Nvidia have provided no real proof other than one's apparently the equivalent of a Titan X and the other can out perform 980SLI.

The real performance cards are due out next year. The next Ti should allow some serious high settings for pushing 1080p@60 on to a 4K screen.

Speaking of which, i'm actually glad I downgraded my monitor back to 1080 (the 4K one I had was too hazy for my liking). I can use pretty much any setting I like now with my 980Ti and I can barely notice any difference. 4K is a gimmick. Beneficial to gamers perhaps, but you'd need a good 60-70" screen to appreciate the difference for video watching.

On another note, IMO, NNEDI3 is a complete waste of resources and I think Super-XBR looks miles better. I had all the settings ramped up on one of my profiles for 720p using NNEDI3, GPU was at about 99% with no dropped frames and no real improvement on screen. Changed to Super-XBR, less resource heavy and sharpens my images enough to look 1080. Guess it really is down to personal taste.
BluesFanUK is offline   Reply With Quote
Old 7th May 2016, 15:38   #37787  |  Link
nevcairiel
Registered Developer
 
Join Date: Mar 2010
Location: Hamburg/Germany
Posts: 10,344
Quote:
Originally Posted by BluesFanUK View Post
Neither card is a substantial enough upgrade over a 980Ti to get more performance out of MadVR. Nvidia have provided no real proof other than one's apparently the equivalent of a Titan X and the other can out perform 980SLI.
The current estimates (from their presentation information and the core/clock ratio) go around 15-20% over a TitanX/980Ti at stock, with easy 20% OC headroom, which personally I find substantial. Only benchmarks will reveal the real information either way however, and anything else is just speculation, both positive and negative.
__________________
LAV Filters - open source ffmpeg based media splitter and decoders
nevcairiel is offline   Reply With Quote
Old 7th May 2016, 16:01   #37788  |  Link
Ver Greeneyes
Registered User
 
Join Date: May 2012
Posts: 447
Quote:
Originally Posted by wanezhiling View Post
1070 is even stronger than titan x.
They said that at the end, but I think that was taking their VR viewport trick into account (since that adds another ~40% performance in VR). Still, the 1070 won't exactly be a pushover either.
Ver Greeneyes is offline   Reply With Quote
Old 7th May 2016, 16:16   #37789  |  Link
Stereodude
Registered User
 
Join Date: Dec 2002
Location: Region 0
Posts: 1,436
Quote:
Originally Posted by nevcairiel View Post
The current estimates (from their presentation information and the core/clock ratio) go around 15-20% over a TitanX/980Ti at stock, with easy 20% OC headroom, which personally I find substantial. Only benchmarks will reveal the real information either way however, and anything else is just speculation, both positive and negative.
But who knows if the new architecture will actually be faster or not for madVR. Nvidia pretty much had AMD's number in the last gen when it came to gaming performance, but AMD's cards offered better madVR performance. A Nvidia card that was effectively the equal in gaming of a given AMD model had considerably higher rendering times with the same madVR options.
Stereodude is offline   Reply With Quote
Old 7th May 2016, 16:33   #37790  |  Link
nevcairiel
Registered Developer
 
Join Date: Mar 2010
Location: Hamburg/Germany
Posts: 10,344
Quote:
Originally Posted by Stereodude View Post
But who knows if the new architecture will actually be faster or not for madVR. Nvidia pretty much had AMD's number in the last gen when it came to gaming performance, but AMD's cards offered better madVR performance. A Nvidia card that was effectively the equal in gaming of a given AMD model had considerably higher rendering times with the same madVR options.
The Pascal architecture doubles the parallelism by making SM units smaller and doubling their count (64 vs 128 cores per SM, compared to Maxwell, Kepler had 192 cores per SM for comparison), which should help any tasks that are incredibly parallel, which image scaling should fall under. But in the end we won't know until we know.
__________________
LAV Filters - open source ffmpeg based media splitter and decoders

Last edited by nevcairiel; 7th May 2016 at 17:02.
nevcairiel is offline   Reply With Quote
Old 7th May 2016, 18:27   #37791  |  Link
har3inger
Registered User
 
Join Date: Feb 2014
Posts: 139
BluesFanUK, try out bilateral chroma with superchromares 2. I personally prefer it to all other algorithms (including nnedi3), and it costs way less than nnedi3.
har3inger is offline   Reply With Quote
Old 7th May 2016, 19:54   #37792  |  Link
sat4all
Registered User
 
Join Date: Apr 2015
Posts: 62
fellas, what's your debanding settings for bluray's remux, is it better to keep it off?
sat4all is offline   Reply With Quote
Old 7th May 2016, 20:34   #37793  |  Link
Stereodude
Registered User
 
Join Date: Dec 2002
Location: Region 0
Posts: 1,436
Quote:
Originally Posted by nevcairiel View Post
The Pascal architecture doubles the parallelism by making SM units smaller and doubling their count (64 vs 128 cores per SM, compared to Maxwell, Kepler had 192 cores per SM for comparison), which should help any tasks that are incredibly parallel, which image scaling should fall under. But in the end we won't know until we know.
I'm hopeful they're better this generation. I will probably replace the GTX 650 in my desktop PC with a GTX 1070. Getting big jump in madVR performance would be appreciated since the GTX 650 basically can't do anything beyond bicubic driving my 2550x1440 display.

Quote:
Originally Posted by sat4all View Post
fellas, what's your debanding settings for bluray's remux, is it better to keep it off?
I use it on low, med during fades myself. I use that setting for all content.
Stereodude is offline   Reply With Quote
Old 7th May 2016, 22:56   #37794  |  Link
aufkrawall
Registered User
 
Join Date: Dec 2011
Posts: 1,812
Quote:
Originally Posted by sat4all View Post
fellas, what's your debanding settings for bluray's remux, is it better to keep it off?
BDs are 8 bit and often badly mastered, so there's a good reason to use deband filter at least at low setting if you don't mind very minimal detail loss (I doubt it can be noticed without concentratedly starring at image comparisons.).
When you use fade-detection, I recommend not to disable the performance/quality trade-off option for it, as you will need longer CPU and GPU queues else.
aufkrawall is offline   Reply With Quote
Old 7th May 2016, 23:09   #37795  |  Link
huhn
Registered User
 
Join Date: Oct 2012
Posts: 7,903
"banding" let's call it steps i guess could can be intentional like in the shinsekai yori ED.

so >I< i don't have it on by default.
huhn is offline   Reply With Quote
Old 8th May 2016, 02:39   #37796  |  Link
SpoCk0nd0pe
Registered User
 
Join Date: Jun 2015
Posts: 25
Quote:
Originally Posted by nevcairiel View Post
The Pascal architecture doubles the parallelism by making SM units smaller and doubling their count (64 vs 128 cores per SM, compared to Maxwell, Kepler had 192 cores per SM for comparison), which should help any tasks that are incredibly parallel, which image scaling should fall under. But in the end we won't know until we know.
Yes, it seems NVidia has been trying to fix their shader architecture starting with Maxwell. We have to see if Pascal can close the gap to AMD.
Is half precision enough for NNEDI3? If yes, Pascal should be very fast at that.
SpoCk0nd0pe is offline   Reply With Quote
Old 8th May 2016, 06:47   #37797  |  Link
CarlosCaco
Registered User
 
Join Date: Oct 2015
Location: Brasil, SP, São Paulo
Posts: 154
I find a strange but good behave on madvr
When i m downscaling a 1080 source i getting some great extra performance if set chroma to reconstruction soft, all other algorithms have greater performance hits when downscaling 1080 to 1366x768 even mitchel nevali is slower than reconstuction soft on this scenario, getting almost less 10 ms using reconstruction soft, the same thing if uncheck don t use linear light for dithering i get better render times that if that option is checked... Someone know why this?
__________________
Desktop, i5 2500, 8GB, N570 GTX TF III PE/OC
Asus X555LF, i7-5500U, 6GB Ram, Nvidia 930m/HD 5500
Windows 8.1 Pro x64
CarlosCaco is offline   Reply With Quote
Old 8th May 2016, 09:16   #37798  |  Link
huhn
Registered User
 
Join Date: Oct 2012
Posts: 7,903
could be GPU "powerstate".
huhn is offline   Reply With Quote
Old 8th May 2016, 11:56   #37799  |  Link
Uoppi
Registered User
 
Join Date: Oct 2015
Posts: 99
SSIM downscaling is not working when deinterlacing - it always reverts to Bicubic150. It says so in the OSD and also switching between 1D and 2D makes no difference in rendering times.

Looking a some older posts, it looks like at one point SSIM may not have been working at all after doubling (?) but it does work now apart from when deinterlacing (dxva) is going on. I'm just wondering if this behavior is to be expected or not.
Uoppi is offline   Reply With Quote
Old 8th May 2016, 12:51   #37800  |  Link
Piyoko
Registered User
 
Join Date: Sep 2005
Posts: 29
A question regarding HDR, why does it currently not work and how would it theoretically work in the future? If I have file encoded in HDR and I'm connected to an HDR TV where does the information that it's an HDR stream get stuck?

Let's assume HDR10 and not the more complex Dolby Vision. As I understand it all of the characteristics are globally defined just like a gamma curve so theoretically the GPU-makers could just provide an HDR output toggle in the driver settings and it would work correctly even without passing through any metadata from the player? (Though if you had a windowed player and enabled that mode the rest of your desktop would probably scorch your eyeballs.)
Piyoko 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 20:54.


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