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 10th July 2015, 03:06   #31621  |  Link
leeperry
Kid for Today
 
Join Date: Aug 2004
Posts: 3,477
So I just spent quite some time comparing .15 and .16:

-0.00 sharpness is less forgiving, I kinda liked slightly increasing it in order to hide compression artifacts but yeah OK unforgiving is good too and I'm currently sitting 80cm away from a 3500:1 32"....I guess 0.00 would be just fine from a 3 meters distance.

-I still far prefer HQ off in .15, enabling it utterly veils the picture to me. Major bottleneck at work, this is a definite no-go(using NEDI/J3AR/CR AR LL/monostatic ED2@8bit on mostly 720p content, of course the constantly dancing noise of dynamic dithering would more than likely completely hide the HQ veil and make it virtually impossible to spot - especially ED1).

-I'm well aware that my views towards that nasty HQ option would be nitpicking to anyone on this planet but then asking for impressions between HQ algos in .16 is way worse, good luck DBT'ing them for that matter lol. All algos add that same HQ veil to the picture IME and the colorspace option doesn't make any visible difference to me either.

All this said, I'm still extremely impressed by how natural PQ looks with 3 passes and 0.75 strength. I would eventually thoroughly enjoy a debug option to disable HQ in .17 but it actually sounds like NEDI might soon be tossed and SR become a silly checkbox without any knob, so I would also be totally cool with sticking with .15 and calling it a day. PQ is honestly and literally beyond all my expectations and I will happily purchase mVR should it go commercial one day


Last edited by leeperry; 10th July 2015 at 06:50.
leeperry is offline   Reply With Quote
Old 10th July 2015, 05:14   #31622  |  Link
AngelGraves13
Registered User
 
Join Date: Dec 2010
Posts: 243
madshi, can we please get super-xbr for image upscaling as well as chroma upscaling?

Currently, Jinc is the best choice for upscaling unless I use image doubling, which causes a slight delay opening and maximizing videos, and it doubles or quadruples and then scales down using Spline instead of just scaling to the proper size, so it's basically doing twice as much work for no reason.
AngelGraves13 is offline   Reply With Quote
Old 10th July 2015, 07:37   #31623  |  Link
Warner306
Registered User
 
Join Date: Dec 2014
Posts: 1,127
Quote:
Originally Posted by AngelGraves13 View Post
madshi, can we please get super-xbr for image upscaling as well as chroma upscaling?

Currently, Jinc is the best choice for upscaling unless I use image doubling, which causes a slight delay opening and maximizing videos, and it doubles or quadruples and then scales down using Spline instead of just scaling to the proper size, so it's basically doing twice as much work for no reason.
Unfortunately, super-xbr is a form of image doubling - it can only double the image resolution. This is why it is limited to chroma upscaling (which is also a form of image doubling) and luma/chroma doubling.

A new linear scaler would be nice, though. Jinc is a little boring after years of use.

Last edited by Warner306; 10th July 2015 at 07:40.
Warner306 is offline   Reply With Quote
Old 10th July 2015, 07:57   #31624  |  Link
AngelGraves13
Registered User
 
Join Date: Dec 2010
Posts: 243
Quote:
Originally Posted by Warner306 View Post
Unfortunately, super-xbr is a form of image doubling - it can only double the image resolution. This is why it is limited to chroma upscaling (which is also a form of image doubling) and luma/chroma doubling.

A new linear scaler would be nice, though. Jinc is a little boring after years of use.
I see. I wonder if there is also a super-xbrz based on the xbrz scaler.

Guess I'll stick to Jinc for now.
AngelGraves13 is offline   Reply With Quote
Old 10th July 2015, 09:39   #31625  |  Link
surgical
Dopax
 
Join Date: Dec 2009
Posts: 17
Greetings to all:
I use madVR with MPC-HC , LAV and ffdshow ; the latter, because occasionally use some avisynth script.
Anyway, I do it on a Nvidia GTX 570 and still have doubts about the correct color space should I use and need your wise counsel.
Currently, based on what I've learned here, I've my Nvidia in full RGB mode , LAV decoding with RGB levels "untouched as imput" and madVR in TV levels 16-235, as I'm connected via HDMI to a TV Sony KDL 32EX-500 (Bravia Engine 3) that, if I'm not mistaken, does'nt support 0-255 (although I'm not sure if this TV Works internally the signal in limited RGB or YCbCr)
It is correct my config ?
Thank you all in advance
surgical is offline   Reply With Quote
Old 10th July 2015, 10:46   #31626  |  Link
Asmodian
Registered User
 
Join Date: Feb 2002
Location: San Jose, California
Posts: 4,403
That is correct if your TV does not support full RGB.

If you want the desktop correct as well you would need to switch Nvidia to limited RGB and madVR to full. This does reduce the quality of madVR because the GPU does a full to limited conversion which damages the video compared to madVR outputting limited range and the GPU leaving it untouched.
__________________
madVR options explained
Asmodian is offline   Reply With Quote
Old 10th July 2015, 12:33   #31627  |  Link
Hyllian
Registered User
 
Hyllian's Avatar
 
Join Date: Jun 2014
Posts: 42
Quote:
Originally Posted by AngelGraves13 View Post
I see. I wonder if there is also a super-xbrz based on the xbrz scaler.

Guess I'll stick to Jinc for now.
xbr and xbrz are basically the same algorithm with slightly differencies in corner treatments. The major differencies between them are in the implementation side, as xbrz was a xbr reimplemented in C++ using paralell instructions for use in some emulators that only use CPU to upscale their framebuffers.

See this post to know more about the differencies.

And Super-xBR is almost totally different than xBR. It uses a slightly modified edge detection from xbr and interpolate pixels using known linear interpolators (sinc, cubic, bilinear, etc). The edge detection is just a direction guide. And it doesn't use the corner treatment from xbr, so I can't see how a hypothetical super-xbrz would differ from the existent super-xbr.

The way I figured out how to combine the edge detection with the linear interpolation of pixels without introducing artifacts only works when exactly doubling the resolution. The same way as NEDI works. For now, if I try to scale by a non power-of-two factor, something goes wrong. Maybe it's possible and I just don't know yet. It's something to research in the future.

Last edited by Hyllian; 10th July 2015 at 13:02.
Hyllian is offline   Reply With Quote
Old 10th July 2015, 13:26   #31628  |  Link
Mano
Registered User
 
Join Date: Jul 2008
Posts: 54
Anyone see any improvement with AMD Catalyst 15.7 compare to 13.12?
Mano is offline   Reply With Quote
Old 10th July 2015, 14:03   #31629  |  Link
Schwartz
Registered User
 
Join Date: Dec 2012
Posts: 66
I've never seen the appeal of using an upscaler and then slapping on a 'corrective' AR algorithm. Ideally, a scaler shouldn't introduce ringing in the first place. Maybe it's so popular because the red and green bars suggest that it's the best. Remember that these bars aren't an objective assessment. I'd be curious to see each scaler's wave diagram instead. If you want to try something new, there's plenty of other upscalers. I'm a big fan of Spline 3 for upscaling and Mitchell for downscaling. SoftCubic 70-80 is great for dealing with bad encodes and old SD content.
Schwartz is offline   Reply With Quote
Old 10th July 2015, 14:16   #31630  |  Link
Hyllian
Registered User
 
Hyllian's Avatar
 
Join Date: Jun 2014
Posts: 42
Quote:
Originally Posted by Schwartz View Post
Ideally, a scaler shouldn't introduce ringing in the first place.
Some ringing is benefic to the output. If you don't introduce it, the output will be blurred some way. The AR function is to get rid of bad ringing.
Hyllian is offline   Reply With Quote
Old 10th July 2015, 14:17   #31631  |  Link
aufkrawall
Registered User
 
Join Date: Dec 2011
Posts: 1,812
The ringing is very obvious with lanczos/spline and afaik, the ringing filter isn't known to really do any harm.
The comparisons with super-xbr and NNEDI3 for chroma led me to the conclusion that Jinc3 AR is doing an extraordinary job.
The only disadvantage is that it can look too soft with extreme synthetic sample upscaling, but this isn't really a realistic case.
aufkrawall is offline   Reply With Quote
Old 10th July 2015, 14:22   #31632  |  Link
leeperry
Kid for Today
 
Join Date: Aug 2004
Posts: 3,477
Quote:
Originally Posted by aufkrawall View Post
The comparisons with super-xbr and NNEDI3 for chroma led me to the conclusion that Jinc3 AR is doing an extraordinary job.
The only disadvantage is that it can look too soft with extreme synthetic sample upscaling, but this isn't really a realistic case.
Agreed, J3AR hits the spot for chroma IME. But I don't like a very hard-edged chroma as it then tends to interfere with luma too much and look artificial, I kinda suspect that some encoders process it somehow in order to counterbalance the drastic 4:2:0 downsampling so there's really nothing to dig for IMHO(on live-action movies at least, not anime or test patterns eventually).
leeperry is offline   Reply With Quote
Old 10th July 2015, 16:07   #31633  |  Link
Braum
Registered User
 
Join Date: Jan 2015
Posts: 48
Madshi do you plan on introducing a frame interpolation function in Madvr ?
Braum is offline   Reply With Quote
Old 10th July 2015, 16:34   #31634  |  Link
surgical
Dopax
 
Join Date: Dec 2009
Posts: 17
Quote:
Originally Posted by Asmodian View Post
That is correct if your TV does not support full RGB.

If you want the desktop correct as well you would need to switch Nvidia to limited RGB and madVR to full. This does reduce the quality of madVR because the GPU does a full to limited conversion which damages the video compared to madVR outputting limited range and the GPU leaving it untouched.
Many thanks Asmodian
surgical is offline   Reply With Quote
Old 10th July 2015, 17:00   #31635  |  Link
aufkrawall
Registered User
 
Join Date: Dec 2011
Posts: 1,812
Suggestion: Could we have a checkbox for image enhancement to just enable it if no luma upscaling is applied?
This would be useful to prevent sharpening of sharpening artifacts when also using it as an upscaling refinement.
I think it would be neat to help NNEDI3 with AdaptiveSharpen, but I don't want it combined AdaptiveSharpen of image enhancements, which on the other hand is neat to fight chroma blur a bit.
aufkrawall is offline   Reply With Quote
Old 10th July 2015, 17:58   #31636  |  Link
James Freeman
Registered User
 
Join Date: Sep 2013
Posts: 919
Quote:
Originally Posted by Braum View Post
Madshi do you plan on introducing a frame interpolation function in Madvr ?
Did you try SVP?
I would love to have a quality interpolation algo in madvR in the future myself.
__________________
System: i7 3770K, GTX660, Win7 64bit, Panasonic ST60, Dell U2410.
James Freeman is offline   Reply With Quote
Old 10th July 2015, 18:43   #31637  |  Link
aufkrawall
Registered User
 
Join Date: Dec 2011
Posts: 1,812
Some more NNEDI3 vs super-xbr doubling test. Sorry about the image size, I don't know how to crop always exactly the same pixels.
However, I could do this in future if someone explained this to me.
720p -> WQHD, Jinc3AR chroma

super-xbr 100 is sharper than NNEDI3 64, but ringing gets more obvious and lines aren't as clean.
NNEDI3 64:


super-xbr 100:


With AdaptiveSharpen of 0.2 as an upscaling refinement for NNEDI3, the image still looks natural and clean, but is sharper than super-xbr 100 without sharpen:


Edit: as a reference, the original frame:
http://abload.de/img/frame6ro6g.png

I also notice the worse line reconstruction of super-xbr e.g. with clothes of news speakers.
Even though it's "just" doubling, the advantage of NNEDI3 is already noticeable.

Last edited by aufkrawall; 10th July 2015 at 18:50.
aufkrawall is offline   Reply With Quote
Old 10th July 2015, 18:57   #31638  |  Link
leeperry
Kid for Today
 
Join Date: Aug 2004
Posts: 3,477
Oh you nail it, indeed sxbr rings a lot....I was hoping that lowering its strength and increasing the SR sharpness could be helpful but maybe not, I still wish I could try sxbr37(25 is too soft and 50 rings a lot) on untouched DVD material though as temperature is seriously too hot around here these days to run NNEDI3 anyway. For now I slightly increase the SR softness in .15 in order to somewhat tame down the sxbr50 ringing.

Last edited by leeperry; 10th July 2015 at 19:28.
leeperry is offline   Reply With Quote
Old 10th July 2015, 19:24   #31639  |  Link
aufkrawall
Registered User
 
Join Date: Dec 2011
Posts: 1,812
Too hot for NNEDI3?
I remember my 780 Ti getting loud with NNEDI3, the MSI Gaming design was too weak for the power consumption to remain silent.
With a MSI Gaming 970, you can set a manual fan curve so that it remains close to totally silent, if you don't mind high VRM temperatures and some coil whine.
aufkrawall is offline   Reply With Quote
Old 10th July 2015, 19:31   #31640  |  Link
leeperry
Kid for Today
 
Join Date: Aug 2004
Posts: 3,477
...and HDD temperature going up as well, I can keep my o/c HD7850 fairly silent as long as NNEDI3 isn't on.
leeperry 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 12:38.


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