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 17th July 2016, 03:15   #38761  |  Link
ryrynz
Registered User
 
ryrynz's Avatar
 
Join Date: Mar 2009
Posts: 3,646
Quote:
Originally Posted by QBhd View Post
p.s. Is there anyway to create a debug OSD that will show every setting being used? Even if it covers the entire screen... something we could use to narrow things down for you, and provide more detailed SS's. It could also be useful for hardcore users to tweak their own settings
There is not. The best way to narrow things down for him is to simply reset to defaults and test each option in your profile on both versions until you find what is causing it. You ideally want the coder to spend as little time hunting things down as possible so more time can be spent on adding new features or optimization.
ryrynz is offline   Reply With Quote
Old 17th July 2016, 07:33   #38762  |  Link
x7007
Registered User
 
Join Date: Apr 2013
Posts: 315
Quote:
Originally Posted by madshi View Post


The render queue is the first queue in the list which is nearly empty, so that's the bottleneck. It seems that your GPU can't keep up. Might have to do with the high refresh rate (100Hz). You may have to lower your scaling settings. E.g. use 16 neurons instead of 32 for chroma upscaling.

Hi,

I will test the 100Hz, but I tried everything on lowest, not NNEDI3, just everything was "worst quality" let's call it, and still it the rendered didn't go up at all. again, it worked well with FSE but not windowed mode.
x7007 is offline   Reply With Quote
Old 17th July 2016, 08:16   #38763  |  Link
nevcairiel
Registered Developer
 
Join Date: Mar 2010
Location: Hamburg/Germany
Posts: 10,344
Quote:
Originally Posted by leeperry View Post

BUT if you really wanna compare SR1 and SR2 AB75, only 4X zooms make sense to me as I run a 32" TV from a 1 meter distance: http://screenshotcomparison.com/comparison/178799
SR2 AB75 looks terrible, so many artifacts with higher contrast then the SR1 image.
__________________
LAV Filters - open source ffmpeg based media splitter and decoders
nevcairiel is offline   Reply With Quote
Old 17th July 2016, 08:19   #38764  |  Link
K2mil
Registered User
 
Join Date: Dec 2015
Location: Chicago
Posts: 2
full screen bug

Does anyone have that bug when double click into full screen mode is not instant there is some sort of video distortion / corruption for half a sec before full screen plays.

I'm on i5 2500K @ 4.5 and 1070 GTX with latest drivers Had it also on my old 680 GTX


Question two any recommended settings for 1070 GTX ?? Thanks
K2mil is offline   Reply With Quote
Old 17th July 2016, 09:36   #38765  |  Link
ryrynz
Registered User
 
ryrynz's Avatar
 
Join Date: Mar 2009
Posts: 3,646
Quote:
Originally Posted by K2mil View Post
Does anyone have that bug when double click into full screen mode is not instant there is some sort of video distortion / corruption for half a sec before full screen plays.

I'm on i5 2500K @ 4.5 and 1070 GTX with latest drivers Had it also on my old 680 GTX
Don't think anyone else does. Is it FSE mode? Are you using D3D11? Windowed overlay? Are you changing refresh rates? You could always just stay in windowed mode. Did you do a fresh install of your drivers of did you just install the 1070 straight in after taking out another Nvidia card?
ryrynz is offline   Reply With Quote
Old 17th July 2016, 10:04   #38766  |  Link
Iznogūd
Registered User
 
Iznogūd's Avatar
 
Join Date: Sep 2002
Location: Barcelona
Posts: 20
Quote:
Is it possible that with 0.90.20 the upscaling refinement was not applied? Please try unchecking those options and see if it makes a difference ot the render times.
Well, that seems to be it. For testing purposes, I've cranked the upscaling refinements all the way up.

With 0.90.20, the render times are virtually the same, comfortably sitting around 30ms, whether they are engaged or not. Also, there is no appreciable visual difference at all.

With 0.90.21, the performance drop is huge, and the visual difference is quite evident, so it would seem that doubling and refining simultaneously is something my GPU (GTX 750TI) can't cope with.

Thanks for looking into it.
Iznogūd is offline   Reply With Quote
Old 17th July 2016, 10:06   #38767  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
Quote:
Originally Posted by leeperry View Post
Well, screenshots are one thing, how AB fares in combination with FRC in 60Hz motion another, basically SR1 doesn't look sharp enough and SR2 too sharp.
Which would again only suggest that you use AB to fine tune SR strength, nothing else.

Quote:
Originally Posted by leeperry View Post
AB gives SR an analog look and does exactly what you initially said: "The new anti-bloating tries to concentrate sharpening on higher frequencies and to remove lower frequencies. I like the look that produces, but your mileage may vary".
I can't see much of this in the screenshots. The madVR SR implementation by design doesn't bloat that much, if at all.

Quote:
Originally Posted by leeperry View Post
BUT if you really wanna compare SR1 and SR2 AB75, only 4X zooms make sense to me as I run a 32" TV from a 1 meter distance: http://screenshotcomparison.com/comparison/178799

Apart from the top left corner I far prefer the SR2 AB75, it even makes the shadow in the bottom left corner lighter & closer to the ground truth, very obviously less bloated™
Don't you see all those extra artifacts introduced by SR2 AB75? I consider them far more serious than any ever so slight improvement in the "look" of the image.

I'm sorry to say, but based on this screenshot comparison I'd say SR1 is clearly better than SR2AB75. Maybe you have two or three other images where SR2AB75 looks better than SR1?

Quote:
Originally Posted by leeperry View Post
Oh so now a ground truth has become mandatory? So we went from "I like the look that produces, but your mileage may vary" to "got ground truth?"

Maybe the OP would be able to provide it, but it seems rather hard to deny that mVR's SR looks way oversharpened in his example IMHO. I'm certainly not interested in such a sharp image myself, and neither would be others as it would appear.
Having an objective ground truth is very helpful, but not always a requirement. Have you actually carefully compared the guitar image in madVR yourself, toggling between different SR settings? Or have you just quickly looked at two screenshots, and noticed that one is way softer than the other (which should be no surprise if one is SR3 with no AB and the other one AB100, which removes almost all of what SR did)?

FWIW, I had downloaded the original guitar image and had done several tests myself before I wrote my previous comment. I had compared the guitar image in madVR with different settings, and had also compared it to Shiandow's new algorithm, before I came to any conclusions. And the conclusion I ended up with was that with this specific image I still prefer my SuperRes over Shiandow's, and I prefer sharper over softer, but it's hard to do a definite judgement without having a ground truth.

Quote:
Originally Posted by leeperry View Post
Apparently the only way to try it would be that MPDN thingie but I've never used and I have no interest in learning as it seems tied to MPC.
So you've spent 5 seconds looking at 2 images, which don't even have the same scaling factor, without having a ground truth to compare to, and you're not willing to do some tests yourself to at least get comparison images with the same scaling factor? But you still think you are in a position to judge which algo looks better, and to request that I spend several hours of my time to implement it?

Quote:
Originally Posted by ShiftyFella View Post
Ok, here it is:

SuperRes turned on:
http://i.imgur.com/3khJ82K.png
http://i.imgur.com/p6JGGyZ.png
SuperRes turned off:
http://i.imgur.com/57jtmWP.png
http://i.imgur.com/IwiMxG6.png

heck, even did a gif switching sr back and forth to show changes:
https://gfycat.com/FloweryIdolizedCaterpillar

In HDR processing everything turned on, preserve hue is on high and peak luminance is at 400 nits. Having different chroma upscaling, image downscaling, doubling, image upscaling algos has no effect on image color shift including turning on or off superres in upscaling refinement, only turning off superres for chroma gets rid of color shit.
Ah, thanks, I've found out that the problem only occurs if the luminance measurement feature is activated at the same time as SuperRes for chroma upscaling!

Quote:
Originally Posted by har3inger View Post
No screenshots for the OSD for now, but literally everything in it is the same, except render time (35ish -> 45ish) , and the render queue emptying slowly over time.

I have no upscale refinement algo's checked except SR (which is definitely on in both cases), so it shouldn't be that, unless one became forced on regardless of the UI setting.
Quote:
Originally Posted by QBhd View Post
p.s. Is there anyway to create a debug OSD that will show every setting being used?
Not exactly, but something that approaches that: Create an empty file named "ShowRenderSteps" in the madVR folder. Then make a screenshot of the OSD with the 2 different madVR versions. That should allow us to see if there's any difference in the configuration of the render steps (shader passes).

Quote:
Originally Posted by x7007 View Post
I will test the 100Hz, but I tried everything on lowest, not NNEDI3, just everything was "worst quality" let's call it, and still it the rendered didn't go up at all. again, it worked well with FSE but not windowed mode.
Can you please make a screenshot of the Ctrl+J OSD with these "worst quality" settings where the render queue doesn't fill in windowed mode?

Quote:
Originally Posted by K2mil View Post
Does anyone have that bug when double click into full screen mode is not instant there is some sort of video distortion / corruption for half a sec before full screen plays.
Switching to fullscreen exclusive mode does take some time, there's not much I can do about that. You can try if disabling fullscreen exclusive mode helps, but it's the most reliable playback mode, so I'm not sure I would recommend sticking with that configuration, even if it helps get rid of the distortion.
madshi is offline   Reply With Quote
Old 17th July 2016, 10:09   #38768  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
Quote:
Originally Posted by Iznogūd View Post
Well, that seems to be it. For testing purposes, I've cranked the upscaling refinements all the way up.

With 0.90.20, the render times are virtually the same, comfortably sitting around 30ms, whether they are engaged or not. Also, there is no appreciable visual difference at all.

With 0.90.21, the performance drop is huge, and the visual difference is quite evident, so it would seem that doubling and refining simultaneously is something my GPU (GTX 750TI) can't cope with.
Good to hear we've solved the mystery, in your case at least.

Quote:
Originally Posted by ryrynz View Post
Don't think anyone else does.
I think he just means the short moment when the video can appear slightly stretched on screen when going into FSE mode.
madshi is offline   Reply With Quote
Old 17th July 2016, 10:20   #38769  |  Link
jerryleungwh
Registered User
 
Join Date: Jun 2016
Posts: 39
Now I am having another problem that when I am doing audio bitstreaming with the latest lav audio filter(nightly), and enters D3D11 FSE mode, there is no audio and the video freezes, even though the seek bar is still moving

Last edited by jerryleungwh; 17th July 2016 at 13:30.
jerryleungwh is offline   Reply With Quote
Old 17th July 2016, 10:51   #38770  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
Quote:
Originally Posted by jerryleungwh View Post
Now I am having another problem that when I am doing audio bitstreaming with the latest lav audio filter(nightly), and enters D3D11 FSE mode, there is no audio and the video freezes, even though the seek bar is still moving
This problem only occurs with audio bitstreaming but not with PCM? And it only occurs in FSE mode? Sounds strange...

-------

Got a reply from Graeme when asking about whether I should reduce more luminance or more saturation:

Quote:
it's very dependent on the situation. If it's out of
gamut due to there being an "exposure adjustment" problem,
then you would reduce luminance more than saturation.

If its luminance is about right, but it is simply more
vivid than the destination gamut can accommodate, then
reducing both luminance and saturation equal perceptual
amounts might be a good first approximation. The exactly
ratio depends on the role the color plays in the overall image.
So the next build is going to have a new option where you can choose how much luminance vs saturation reduction you want to have. Not sure if the option will stay long term, but it should allow us to do some experiments to maybe find a good setting for all content.
madshi is offline   Reply With Quote
Old 17th July 2016, 11:10   #38771  |  Link
jerryleungwh
Registered User
 
Join Date: Jun 2016
Posts: 39
Quote:
Originally Posted by madshi View Post
This problem only occurs with audio bitstreaming but not with PCM? And it only occurs in FSE mode? Sounds strange...

-------

Got a reply from Graeme when asking about whether I should reduce more luminance or more saturation:


So the next build is going to have a new option where you can choose how much luminance vs saturation reduction you want to have. Not sure if the option will stay long term, but it should allow us to do some experiments to maybe find a good setting for all content.
I'll need to run more tests to say for sure as I'm switching between 2 TVs and I'm not sure if that matters. When I uncheck the audio formats on LAV audio filter for bitstreaming and use the TV headphone socket as output, it works fine in D3D11 FSE mode, but when I switch to another setup which I can use an AV receiver for HD audio bitstream(computer->AV receiver->TV), I check the audio formats for bitstreaming in LAV audio filter and theres is no audio as soon as I switch to FSE mode. The audio doesn't return even after I switch back to windowed mode without quitting the player.
jerryleungwh is offline   Reply With Quote
Old 17th July 2016, 11:43   #38772  |  Link
ShiftyFella
Registered User
 
Join Date: Apr 2016
Posts: 17
Quote:
Originally Posted by madshi View Post
Ah, thanks, I've found out that the problem only occurs if the luminance measurement feature is activated at the same time as SuperRes for chroma upscaling!
I see, that's the only thing I didn't try turning off in hdr. I tried turning off compress highlights but that made no difference to the image and because gui greys out other options, I just assumed they were turned off as well, so never gave second thought to play with them them.


Quote:
Originally Posted by pose View Post
Any RX 480 users here? Id like to compare my rendering times on 7870 to the new card! Mine could do 1280x720 24FPS video to 1080p with luma doubling 64 NNEDI3, smooth motion, error diffusion dithering and low/high debanding with around 35ms rendering time. Trade quality for performance - all unchecked. Win 7 Catalyst 13.12 driver.
According to benches, 480 should be roughly equal to my 290x, so this might be good reference point for you, as I can do your settings at around 16ms but you can crank it up to Chroma ReconSoft@SR1, Doubling NNEDI3 128 luma and 32 for chroma, CE 0.6 AR in upscaling refinement and end up at around 32ms with smooth motion, error diffusion 2 http://i.imgur.com/SJO2IF8.png
If I go for super-xbr for doubling instead of nnedi3 render time drops to 18ms with almost non existent load on the gpu, which is pretty nice if you ask me
ShiftyFella is offline   Reply With Quote
Old 17th July 2016, 12:21   #38773  |  Link
pose
Registered User
 
Join Date: Jul 2014
Posts: 69
ShiftyFella, performance of your 290x is amazing! I really hope that RX 480 will perform as well but i doubt it after seeing Sunset1982 test. He compared RX 480 to GTX 970 and they are pretty close in terms of performance. If i remember correctly, back in the days NNEDI3 worked faster on AMD cards and my 7870 were faster than 970. Maybe things changed since then but this is really concerns me.
Quote:
Excel File:
MadVRComparison.xlsx
pose is offline   Reply With Quote
Old 17th July 2016, 12:26   #38774  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
Quote:
Originally Posted by jerryleungwh View Post
I'll need to run more tests to say for sure as I'm switching between 2 TVs and I'm not sure if that matters. When I uncheck the audio formats on LAV audio filter for bitstreaming and use the TV headphone socket as output, it works fine in D3D11 FSE mode, but when I switch to another setup which I can use an AV receiver for HD audio bitstream(computer->AV receiver->TV), I check the audio formats for bitstreaming in LAV audio filter and theres is no audio as soon as I switch to FSE mode. The audio doesn't return even after I switch back to windowed mode without quitting the player.
Don't know, might be that in your Optimus system one GPU doesn't pass the bitstream properly to the other GPU or something like that. I don't think there's anything I can do about it.
madshi is offline   Reply With Quote
Old 17th July 2016, 12:27   #38775  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
madVR v0.90.23 released

http://madshi.net/madVR.zip

Code:
* updated AdaptiveSharpen to version 2016-07-10
* custom output levels can now be adjusted in 0.25 steps
* HDR: added option to define how to handle too bright & saturated pixels
* fixed: HDR luminance measurements + chroma SuperRes produced green screen
* fixed: #420: chroma was doubled despite being below scaling factor threshold
madshi is offline   Reply With Quote
Old 17th July 2016, 13:03   #38776  |  Link
leeperry
Kid for Today
 
Join Date: Aug 2004
Posts: 3,477
Quote:
Originally Posted by madshi View Post
I'm sorry to say, but based on this screenshot comparison I'd say SR1 is clearly better than SR2AB75. Maybe you have two or three other images where SR2AB75 looks better than SR1?
I'm afraid that whatever screenshots I would provide they would never be to your taste as you just said yourself that you "prefer sharper over softer" when I'm looking for the best sharp/natural compromise and again I use SSIM 2D 100% downscaling which is very sharp. SR's AB looks stunning to me in 60Hz FRC on low-res low-bitrate footage.

Quote:
Originally Posted by madshi View Post
Have you actually carefully compared the guitar image in madVR yourself, toggling between different SR settings?
I just like the look of Shiandow's, it looks very natural as in neither too sharp or too soft. My goal is to have the feeling of looking through a window and his screenshot seems more prone to fool my brain.

Quote:
Originally Posted by madshi View Post
I had downloaded the original guitar image and had done several tests myself before I wrote my previous comment. I had compared the guitar image in madVR with different settings, and had also compared it to Shiandow's new algorithm, before I came to any conclusions. And the conclusion I ended up with was that with this specific image I still prefer my SuperRes over Shiandow's, and I prefer sharper over softer, but it's hard to do a definite judgement without having a ground truth.
Right, so even though several people in this thread made clear that they thoroughly enjoyed SR's AB you're still not keen on bringing it back and again you just said that you "prefer sharper over softer", I see little point in me continuing to waste your time about it. I rest my case that I especially like SR's AB in combination with 60Hz FRC and 2X SR refinement. I'll just stick to 0.90.21 for a (long) while then because as others in this thread we like the look that it produces, but our mileage may very well vary indeed

Quote:
Originally Posted by madshi View Post
So you've spent 5 seconds looking at 2 images, which don't even have the same scaling factor, without having a ground truth to compare to, and you're not willing to do some tests yourself to at least get comparison images with the same scaling factor? But you still think you are in a position to judge which algo looks better, and to request that I spend several hours of my time to implement it?
I figured that at this point it would only be a matter of rolling a script in mVR as the implementation would appear to already be there but tbh I must admit that I'm kinda getting tired of constantly whining in this thread for my favorite settings to stay and you must feel even more irritated reading it. Please allow me to apologize for this situation as I sincerely do not mean to abuse your patience and I certainly do not take anything for granted, we're already extremely lucky to have you sharing your most tremendous work with us so let's call it a day then

Too bad stupid MPC can't do seamless playback and provide a D3D GUI, the W95 WMP GUI really nails it huh....it would feel like a Formula 1 engine in a Trabant or something, oh well c'est la vie otherwise I would happily give MPDN a shot as scripts would appear to be user-customizable for a change.

Last edited by leeperry; 17th July 2016 at 13:19.
leeperry is offline   Reply With Quote
Old 17th July 2016, 13:26   #38777  |  Link
James Freeman
Registered User
 
Join Date: Sep 2013
Posts: 919
Quote:
Originally Posted by madshi View Post
madVR v0.90.23 released

Code:
* HDR: added option to define how to handle too bright & saturated pixels
Looks beautiful, Thank You!
Samsung and Panasonic eating their hats with jealousy.

EDIT:
Bug: Luminance/Saturation % takes effect even if "preserve hue in" is unchecked.
__________________
System: i7 3770K, GTX660, Win7 64bit, Panasonic ST60, Dell U2410.

Last edited by James Freeman; 17th July 2016 at 13:34.
James Freeman is offline   Reply With Quote
Old 17th July 2016, 13:26   #38778  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
Quote:
Originally Posted by leeperry View Post
I'm afraid that whatever screenshots I would provide they would never be to your taste
Taste got nothing to do with it. I've proven in the past that if users provide objective screenshot proof, I'm very willing to change my mind. Unfortunately your screenshot comparison didn't convince me. Which again has nothing to do with taste, but everything to do with that your preferred settings produced very visible artifacts.

Quote:
Originally Posted by leeperry View Post
I just like the look of Shiandow's, it looks very natural as in neither too sharp or too soft. My goal is to have the feeling of looking through a window and his screenshot seems more prone to fool my brain.
Shiandow's image might have a soft vs sharp balance that you like. But it is clearly more "bloated" looking than anything madVR's SuperRes algo produces. Which you would have seen if you'd have done a proper comparison.

Quote:
Originally Posted by leeperry View Post
Right, so even though several people in this thread made clear that they thoroughly enjoyed SR's AB
Huh? Please give me a list of those several people saying they "thoroughly enjoyed SR's AB". And no, ryrynz writing a short comment that he prefers AB in one specific image doesn't satisfy your claim.

Quote:
Originally Posted by leeperry View Post
Too bad stupid MPC can't do seamless playback and provide a D3D GUI, the W95 WMP GUI really nails it huh....it would feel like a Formula 1 engine in a Trabant or something, oh well c'est la vie otherwise I would happily give MPDN a shot as scripts would appear to be user-customizable for a change.
MPDN got nothing to do with MPC.
madshi is offline   Reply With Quote
Old 17th July 2016, 13:28   #38779  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
Quote:
Originally Posted by James Freeman View Post
Looks beautiful, Thank You!
Samsung and Panasonic eating their hats with jealousy.
Hehe - thanks! So which luminance vs saturation setting do you prefer (see new option in the HDR settings page)? Would be great if you could check with multiple different demos, though, because what looks good in one scene/video might not look so well in another scene/video.
madshi is offline   Reply With Quote
Old 17th July 2016, 13:45   #38780  |  Link
Uoppi
Registered User
 
Join Date: Oct 2015
Posts: 99
Quote:
Originally Posted by madshi View Post


Thanks. That's looks like quite a strong difference. Is it possible that with 0.90.20 the upscaling refinement was not applied? Please try unchecking those options and see if it makes a difference ot the render times.

There was a user who complained about upscaling refinement settings not becoming active when upscaling by a relatively small scaling factor. I've changed that, which will of course slow things down if upscaling refinement suddenly becomes active when it didn't before.
EDIT: Oops, I missed the post above where this was just explained. So SR was not activated previously on v.20, most likely.

(Hmm, me personally, I use SR1 or SR2 for every scaling profile and no other refinements. So SR might not have kicked in previously?

All I know is that I've needed to revert from SSIM 2D to SSIM1D in most of my profiles after v.20, so yeah, the difference must be several ms at least. Not a huge deal to me but was just intrigued if anyone else was having the issue. So hence no real testing or screenshots, sorry for that.)

Last edited by Uoppi; 17th July 2016 at 13:51.
Uoppi 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 09:39.


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