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 12th May 2015, 23:44   #29801  |  Link
huhn
Registered User
 
Join Date: Oct 2012
Posts: 7,922
Quote:
Originally Posted by madshi View Post
Luma quadrupling yes. But chroma doubling/quadrupling? I'd like to see a comparison screenshot.

Does the test build fix the refresh rate problem for you?
i wasn't on the right PC haven't even installed that.

i go install 87.21 do some screens (found the video i did these chroma doubling/quadrupling test month ago) and than test your new 88 build.

nedi totally fails for pixel art by the way.
huhn is offline   Reply With Quote
Old 12th May 2015, 23:45   #29802  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
Quote:
Originally Posted by aufkrawall View Post
How'd I use chroma doubling with 0.87? Testing with 4:2:2 content? If yes, this is way too much effort.
The source format doesn't matter. Any source will do, 4:2:0, 4:2:2, 4:4:4 or even RGB. We're not talking about 4:2:0 -> 4:4:4 chroma upsampling here, but about "doubling" the chroma channels when the luma channel is doubled with NNEDI3 at the same time. With v0.87.21 you can double chroma in the "image doubling" settings page. Or you can switch "image upscaling" to Jinc3 AR, then that will be used for doubling chroma, as well, if NNEDI3 chroma doubling is disabled.

Quote:
Originally Posted by aufkrawall View Post
Yes, but afaik, C-R is hardly better in any situation compared to Jinc3.
Jinc3 AR does cost a lot more performance than Catmull-Rom AR. The difference should be noticeable even on a fast GPU. It might not be a night-and-day difference, but even saving 2-3ms can help if you want to use other algorithms, as well.
madshi is offline   Reply With Quote
Old 12th May 2015, 23:49   #29803  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
Quote:
Originally Posted by mark0077 View Post
My question is, although I know my gpu is literally at its absolute limits, how come some of the queues seem full, while others are nearly empty all the time, yet my gpu isn't under enough pressure to drop frames. Absolutely very low priority question, just thought it was worth mentioning as I'm not sure if the render and present queues not being full in this scenario may be something madVR is controlling and could potentially be improved to have them more full seeing as frames aren't being dropped.
I don't know. If your GPU is fast enough, the queues should be full. If it's not fast enough, you should be dropping frames. I don't know why your queues would be near empty, but playback is still smooth. Make sure in your NVidia control panel you have the max number of pre-rendered frames set to application controlled.

Quote:
Originally Posted by huhn View Post
i go install 87.21 do some screens (found the video i did these chroma doubling/quadrupling test month ago) and than test your new 88 build.

nedi totally fails for pixel art by the way.
Thanks.

NEDI works ok for pixel art, when combined with SuperRes. Without SuperRes it totally fails, that's true.
madshi is offline   Reply With Quote
Old 12th May 2015, 23:54   #29804  |  Link
mark0077
Registered User
 
Join Date: Apr 2008
Posts: 1,106
Quote:
Originally Posted by madshi View Post
I don't know. If your GPU is fast enough, the queues should be full. If it's not fast enough, you should be dropping frames. I don't know why your queues would be near empty, but playback is still smooth. Make sure in your NVidia control panel you have the max number of pre-rendered frames set to application controlled.
OK cool, when I go down to 16 neurons all queues fill up again, just thought it might be an algorithm in madVR perhaps to balance the queues based on some priority, but I guess not.
mark0077 is offline   Reply With Quote
Old 12th May 2015, 23:56   #29805  |  Link
XMonarchY
Guest
 
Posts: n/a
Quote:
1) With v0.87.21, what happens if you disable Chroma doubling and Chroma quadrupling (right side of the image doubling page)? Do you then get similar results to v0.88.5?
With build 87.21 (new 88.5 features not available in this build, Image Upscaling is set to Lanczos 3 tap, and FSE to disabled):
- Rendering times are about 20-22ms when I set Chroma Upscaling to NNEDI3 32 neurons, and Image (BOTH LUMA AND CHROMA) Doubling + Quadrupling to NNEDI3 32 neurons. No frames are dropped.
- Rendering times are about 18-20ms when I set Chroma Upscaling to NNEDI3 32 neurons, and Image (ONLY LUMA) Doubling + Quadrupling to NNEDI3 32 neurons. No frames are dropped.
- I do not get the same results with build 88.5. Please see below.


Quote:
2) What are your rendering times with: a) v0.87.21 with your original settings. b) v0.87.21 with disabled Chroma doubling/quadrupling. c) v0.88.5?
A) With build 87.21 (new 88.5 features not available in this build, Image Upscaling is set to Lanczos 3 tap, and FSE to disabled):
Original Settings
- Rendering times are about 20-22ms when I set Chroma Upscaling to NNEDI3 32 neurons, and Image (BOTH LUMA AND CHROMA) Doubling + Quadrupling to NNEDI3 32 neurons. No frames are dropped.

B) With build 87.21 (new 88.5 features not available in this build, Image Upscaling is set to Lanczos 3 tap, and FSE to disabled):
Chroma Doubling + Quadrupling disabled
- Rendering times are about 18-20ms when I set Chroma Upscaling to NNEDI3 32 neurons, and Image (ONLY LUMA) Doubling + Quadrupling to NNEDI3 32 neurons. No frames are dropped.

C) With build 88.5 (ALL new features turned off, Image Upscaling is set to Lanczos 3 tap, and FSE to disabled):
- Rendering times are about 18-20ms when I set Chroma Upscaling to NNEDI3 32 neurons, and Image Doubling + Quadrupling to NNEDI3 32 neurons. No frame drop this time!

I even went back and set Image Upscaling to Jinc 3 tapm, enabled the Shiandow's Deband Algorithm, and enabled SuperRes filter for Chroma Upscaling and it produced no frame drops! Quadrupling issue is SOLVED with 88.5! . Using SuperRes fot Image Refinment in addition to settings above resulted in frame drops, but as you said - those settings are only for testing.

Last edited by XMonarchY; 13th May 2015 at 00:35.
  Reply With Quote
Old 12th May 2015, 23:59   #29806  |  Link
AngelGraves13
Registered User
 
Join Date: Dec 2010
Posts: 255
Will you be adding 60 frames to render ahead for Windowed and Fullscreen now that D3D11 can sync to the VSYNC rate? Or am I misunderstanding the whole D3D11 mode that was recently added?

I've found setting Windowed and Fullscreen to 16 and GPU que to 16 with CPU que at 20 provides very low ms (10-20 ms) playback.
AngelGraves13 is offline   Reply With Quote
Old 13th May 2015, 00:00   #29807  |  Link
detmek
Registered User
 
Join Date: Aug 2009
Posts: 463
Quote:
Originally Posted by madshi View Post
Ouch. I'm sorry that you spent so much time with different OSs just to find out it didn't help at all. Unfortunately I don't know what to do. I can't reproduce the problem you have on my PC. And it seems nobody else has the same problem which you have. I've no clue why.

If it would be very easy for you, maybe you could create a short digicam video, showing the problem you're seeing? But don't bother if it would be hard for you to do.
No problem. Like I said, I wanted to try Windows 10. I guess I should tried fresh install as Win 8.1 > Win 10 didn't work for me.

Anyway, I created 2 clips, one with Bandicam and one with my compact digicam.

P.S. That half a second black screen is because I restarted playback with Ctrl + E.


I also noticed that Render Queue is not full in D3D11 mode although rendering times are low but I can't remember if I saw that under Windows 8.1.

No big deal. I just wanted to try D3D11 for better performance. I don't need 10-bit output or so. Madshi
__________________
Intel Pentium G3220, Asus H81M-K, 4GB DDR3@1600MHz (1333MHz working), LG W1934S, Windows 8.1 Pro x64
detmek is offline   Reply With Quote
Old 13th May 2015, 00:13   #29808  |  Link
agustin9
Registered User
 
Join Date: Aug 2008
Posts: 86
Quote:
Originally Posted by madshi View Post
I've done some digging and hacking and raping Direct3D, I think I have the refresh rate problems sorted out. Give this build a try:

http://madshi.net/madVR885refreshRateFix.zip

Should work for D3D9 and D3D11 now, and also for 64bit. Once again, I've spent a lot of time to fix Microsoft's stupidity. They actually *intentionally* use 59.940Hz (23.976Hz) even if you ask for 60.000Hz (24.000Hz). There's no way to convince Direct3D to use 60.000 (24.000), except by going behind their back and twisting and binding their fingers, like I've done. I wonder if there's any other software besides madVR which can switch FSE mode to 24.000/60.000? I kinda doubt it...
With this build it switches to 59.93, and leave the screen in that refresh rate even after I go back to windowed. Previously when going back to windowed it switched to 59.99. Windows 8.1 here
agustin9 is offline   Reply With Quote
Old 13th May 2015, 00:17   #29809  |  Link
huhn
Registered User
 
Join Date: Oct 2012
Posts: 7,922
Quote:
Originally Posted by madshi View Post
NEDI works ok for pixel art, when combined with SuperRes. Without SuperRes it totally fails, that's true.
ok quadrupling is not that huge sorry but placebo? not sure.chroma doubling is quiet something!

chroma upscale
jinc3AR: http://abload.de/img/nnedi3jinc3arx2syl.png
nnedix2: http://abload.de/img/nnedi3chromax2posri.png
nnedix4: http://abload.de/img/nnedi3chromax424swl.png
editv2: catmull-rom AR http://abload.de/img/nnedi3catv2misgl.png

and with some more jinc3ar scaling on top:
jinc3AR: http://abload.de/img/nnedi3jinc3arg2sj3.png
nnedix2: http://abload.de/img/nnedi3chromax2rtsql.png
nnedix4: http://abload.de/img/nnedi3chromax4z7ssj.png

i wouldn't be shock if scaling to UHD would show some more potential for quadrupling on this source. but this source is just ...

edit2:added the right frame

Last edited by huhn; 13th May 2015 at 14:01.
huhn is offline   Reply With Quote
Old 13th May 2015, 00:28   #29810  |  Link
mark0077
Registered User
 
Join Date: Apr 2008
Posts: 1,106
Quote:
Originally Posted by agustin9 View Post
With this build it switches to 59.93, and leave the screen in that refresh rate even after I go back to windowed. Previously when going back to windowed it switched to 59.99. Windows 8.1 here
Maybe you guys already know about this utility, but its what I have used in windows 7 / 8 to stop my PC setting my TV to 59.9X mode.

http://www.monitortests.com/forum/Th...on-Utility-CRU

It lets you set what resolutions your display is allowed to use. Once setup correctly, you can make your display only be allowed use 60hz for example by removing the other ones you don't want to allow. This worked for me with a bit of fiddling.
mark0077 is offline   Reply With Quote
Old 13th May 2015, 00:35   #29811  |  Link
pirlouy
_
 
Join Date: May 2008
Location: France
Posts: 692
Quote:
Originally Posted by madshi View Post
Did it ever work for you, in D3D9, I mean?
No. As i said earlier, I already had this problem with 0.87.21 (no D3D11 if I understood correctly).
I noticed the problem with x64 build since the 64 build switches to 59Hz.
I never noticed the problem with x86 builds before because in fact the x86 build auto switches to 60Hz and I was already in 60Hz.

I don't know if I'm clear but:

With x86 build ; desktop to 59Hz; No D3D11; launching file; goes into FullScreen Exclusive --> Switches to 60Hz

With x86 build ; desktop to 60Hz; No D3D11; launching file; goes into FullScreen Exclusive --> No Switches

With x64 build ; desktop to 60Hz; No D3D11; launching file; goes into FullScreen Exclusive --> Switches to 59Hz

With x64 build ; desktop to 59Hz; No D3D11; launching file; goes into FullScreen Exclusive --> No Switches.


But during my extensive tests today, I've most of all noticed Fullscreen Exclusive causes dropped frames and other nasty things. For now, the windows renderer is way better for me. But I don't think it is linked with 0.88 since it is the same if I don't use D3D11.

Last edited by pirlouy; 13th May 2015 at 00:39.
pirlouy is offline   Reply With Quote
Old 13th May 2015, 00:43   #29812  |  Link
huhn
Registered User
 
Join Date: Oct 2012
Posts: 7,922
refreshratefix doesn't fix the issue and yes the issue is there with normal FSE mode too.

only 32 bit FSE works fine interesting.

Last edited by huhn; 13th May 2015 at 00:46.
huhn is offline   Reply With Quote
Old 13th May 2015, 01:44   #29813  |  Link
Dogway
Registered User
 
Join Date: Nov 2009
Posts: 2,361
Quote:
Originally Posted by madshi View Post
Profiles are not supported yet for displays. I'm not totally sure I understand what you need/want. madVR should only switch to 1080p24 for 24fps sources, anyway. So what else do you want?
If I only have 1080p24 (or 1080p23 for that matter) in the setting box, every video that goes to fullscreen will turn the TV to 1080p24 mode regardless of source fps. You say this shouldn't happen right? I use madVR v0.87.21, and a bit older MPC-HC (for a reason) v1.6.9.7418 Lite, not sure this is the culprit.
Dogway is offline   Reply With Quote
Old 13th May 2015, 02:17   #29814  |  Link
agustin9
Registered User
 
Join Date: Aug 2008
Posts: 86
Quote:
Originally Posted by mark0077 View Post
Maybe you guys already know about this utility, but its what I have used in windows 7 / 8 to stop my PC setting my TV to 59.9X mode.

http://www.monitortests.com/forum/Th...on-Utility-CRU

It lets you set what resolutions your display is allowed to use. Once setup correctly, you can make your display only be allowed use 60hz for example by removing the other ones you don't want to allow. This worked for me with a bit of fiddling.
It's a good workaround, thanks!
agustin9 is offline   Reply With Quote
Old 13th May 2015, 02:35   #29815  |  Link
pirlouy
_
 
Join Date: May 2008
Location: France
Posts: 692
After some tests, here is what I've noticed with 2-3 minutes from the same movie.
From best config to worst

madVR x86, No Fullscreen exclusive (windowed), No D3D11 (D3D9): 0 frames dropped/glitches
madVR x64, No Fullscreen exclusive (windowed), No D3D11 (D3D9): 2 frames dropped/glitches
madVR x86, Fullscreen exclusive, No D3D11 (D3D9): 5 frames dropped/glitches

After, all configs have problems, did not really count, but D3D11 adds dropped frames/glitches, not as much as FullScreen Exclusive+64 bits though.
And so the worst config is:
madVR 64 bits, Fullscreen exclusive, D3D11: 50 frames dropped/glitches

Some comparison with madVR 0.87.21:
madVR x86, No Fullscreen exclusive (windowed): 0 frames dropped/glitches
madVR x64, No Fullscreen exclusive (windowed): 0 frames dropped/glitches
madVR x86, Fullscreen exclusive: 1 frame dropped/glitche
madVR x64, Fullscreen exclusive: 10 frames dropped/glitches

Sorry I read you wanted feedback on debanding for now, but I had some time to do tests, and I'm not interested at all by upscaling/debanding/dithering, I'm not strong enough to notice image quality.

Last edited by pirlouy; 13th May 2015 at 02:50.
pirlouy is offline   Reply With Quote
Old 13th May 2015, 02:39   #29816  |  Link
Asmodian
Registered User
 
Join Date: Feb 2002
Location: San Jose, California
Posts: 4,407
Quote:
Originally Posted by Dogway View Post
If I only have 1080p24 (or 1080p23 for that matter) in the setting box, every video that goes to fullscreen will turn the TV to 1080p24 mode regardless of source fps. You say this shouldn't happen right? I use madVR v0.87.21, and a bit older MPC-HC (for a reason) v1.6.9.7418 Lite, not sure this is the culprit.
If you only have one setting madVR will switch to that setting.

Quote:
Originally Posted by madshi View Post
But I'm willing to bow to the majority. What does everybody else say? Should I "allow" Jinc3 to be used for chroma doubling? Instead of blocking it and using Catmull-Rom instead?
Personally I am happy to have Catmull-Rom AR used for doubling the chroma. I can see some improvements in huhn's example, but that is Jinc3 v.s. NNEDI3 not Catmull-Rom v.s Jinc3. For this use case maybe a trade quality for performance option that used chroma scaling, not image scaling, so one could still get NNEDI3 chroma doubling without any extra settings? Is Jinc3 enough of an improvement over Catmull-Rom AR with that source to want the trade quality for performance option?

I take this phrasing to mean that chroma is doubled to match luma, the video is converted back to RGB, and then downscaled with image downscaling (assuming a <2 scaling factor)? Sorry if I am reading too much into your choice of words.
__________________
madVR options explained
Asmodian is offline   Reply With Quote
Old 13th May 2015, 04:14   #29817  |  Link
huhn
Registered User
 
Join Date: Oct 2012
Posts: 7,922
don't forget my source was lossless RGB a normal 4:2:0 like 99 % of all videos needs to use nnedi3 on chroma first and that means the chroma doubling will lose even more of it's power.
huhn is offline   Reply With Quote
Old 13th May 2015, 04:18   #29818  |  Link
Warner306
Registered User
 
Join Date: Dec 2014
Posts: 1,127
Quote:
Originally Posted by Asmodian View Post
If you only have one setting madVR will switch to that setting.



Personally I am happy to have Catmull-Rom AR used for doubling the chroma. I can see some improvements in huhn's example, but that is Jinc3 v.s. NNEDI3 not Catmull-Rom v.s Jinc3. For this use case maybe a trade quality for performance option that used chroma scaling, not image scaling, so one could still get NNEDI3 chroma doubling without any extra settings? Is Jinc3 enough of an improvement over Catmull-Rom AR with that source to want the trade quality for performance option?

I take this phrasing to mean that chroma is doubled to match luma, the video is converted back to RGB, and then downscaled with image downscaling (assuming a <2 scaling factor)? Sorry if I am reading too much into your choice of words.
The improvement in that image comparison is very minimal.

A trade quality for performance checkbox that forces Catmull-Rom chroma scaling would make sense. NNEDI3 is meant to be the best of the best scaling algorithm. If the opposite logic was used, place a checkbox next to Image Doubling that allows for Jinc3 chroma scaling. Either way, it is only one checkbox.

Last edited by Warner306; 13th May 2015 at 04:25.
Warner306 is offline   Reply With Quote
Old 13th May 2015, 04:25   #29819  |  Link
Warner306
Registered User
 
Join Date: Dec 2014
Posts: 1,127
I don't find image doubling too complex or confusing in its current iteration. It is Image Enhancements and Upscaling Refinement that require simplification.
Warner306 is offline   Reply With Quote
Old 13th May 2015, 04:25   #29820  |  Link
Anime Viewer
Troubleshooter
 
Anime Viewer's Avatar
 
Join Date: Feb 2014
Posts: 339
Quote:
Originally Posted by madshi View Post

You won't get highest quality by using the most expensive algorithm for every part of the processing pipeline. You will get highest quality by spending your GPU for algorithms which make a difference. Instead of worrying about chroma doubling, have a look at FineSharp and SuperRes.

But I'm willing to bow to the majority. What does everybody else say? Should I "allow" Jinc3 to be used for chroma doubling? Instead of blocking it and using Catmull-Rom instead?
One other option you could consider regarding Chroma is possibly instituting a SuperChromaRes option (similar to what MPDN has done) like you have done with SuperRes. It might be a less resource draining option than chroma doubling.
__________________
System specs: Sager NP9150 SE with i7-3630QM 2.40GHz, 16 GB RAM, 64-bit Windows 10 Pro, NVidia GTX 680M/Intel 4000 HD optimus dual GPU system. Video viewed on LG notebook screen and LG 3D passive TV.
Anime Viewer 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 03:46.


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