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 28th November 2016, 03:30   #40821  |  Link
leeperry
Kid for Today
 
Join Date: Aug 2004
Posts: 3,477
Thanks for the new version

Must admit I never understood the point to let us choose Jinc upscaling when doubling anyway, it's much clearer now.

But could you please make the NGU+SR nagscreen a one-timer and that's it? It currently shows up everytime I click on OK

Will run more tests tomorrow but yes so far it seems less confusing and no less efficient as far as I can tell so GG.

I still would like to be able to increase luma NGU and disable chroma NGU if need be, this currently seems impossible? I was doing it in the previous version and what's that new "downscale quality" setting exactly?

So doubling "downscale quality" is not related to what we picked in "image downscaling"? That's very confusing.

I would also like to have sxbr back please, sometimes its EE did the magic on low-res SD.

Quote:
Originally Posted by AngelGraves13 View Post
- Remove ringing artifacts is a must even with NGU. I was comparing a still of Futurama with it on and off and it removed the remaining ringing around the dark black line art.
That's one of the most dramatic filter in mVR to my eyes, turning it off just hurts my brain huh. I rest my case that my impression is that it repairs the ringing/fringing induced by camera optics and it looks so great with NGU

Last edited by leeperry; 28th November 2016 at 04:06.
leeperry is offline   Reply With Quote
Old 28th November 2016, 05:27   #40822  |  Link
Sunset1982
Registered User
 
Join Date: Sep 2014
Posts: 280
Quote:
Originally Posted by Sp00kyFox View Post
@madshi
but could you explain to us what the logic is behind the NGU options "chroma/downscale quality: automatic" and "activate only if it's useful"? this is some information the power user probably wanna know, me included.
I would like to know that too.

What settings are behind the chroma quality and downscale quality options?

How do these image upscaling options interact / depend on the normal chroma upscaling / image downscaling options?



If I select SuperXBR with SR in the chroma upscale tab, I can't choose the SR strength anymore. Bug or feature?!
__________________
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

Last edited by Sunset1982; 28th November 2016 at 05:30.
Sunset1982 is offline   Reply With Quote
Old 28th November 2016, 05:50   #40823  |  Link
JarrettH
Registered User
 
Join Date: Aug 2004
Location: Canada
Posts: 860
If you read the OSD and cycle the settings, you can see which scalers are being used.
JarrettH is offline   Reply With Quote
Old 28th November 2016, 06:04   #40824  |  Link
sauma144
Registered User
 
Join Date: Sep 2016
Posts: 89
@madshi
1) Why not creating presets based on quality or performance for newcomers?
2) Is something like "sync playback to display" from Kodi planned for madVR?

Last edited by sauma144; 28th November 2016 at 06:15.
sauma144 is offline   Reply With Quote
Old 28th November 2016, 07:26   #40825  |  Link
Asmodian
Registered User
 
Join Date: Feb 2002
Location: San Jose, California
Posts: 4,403
Quote:
Originally Posted by huhn View Post
it was
NGU very high below nnedi3 64
NGU high below nnedi3 32
NGU mid below nnedi 16

doesn't count for polaris GPUs.
Based on my experiance and other user reports both Maxwell and Pascal also seem to be slower than expected. NGU very-high on my pascal Titan X, with v0.91.2, still has a performance hit similar to NNEDI3 128 instead of 64.

It is hard to say where the quality is equivalent, they look pretty different.

Edit: I quite like the new settings page, there is a lot less to setup and it is great to have downscaling after doubling be a separate option. Bicubic 60 too.
Automatic chroma and downscale quality seems to pick exactly what I would set, at least when not in a placebo mind set.
__________________
madVR options explained

Last edited by Asmodian; 28th November 2016 at 07:41.
Asmodian is offline   Reply With Quote
Old 28th November 2016, 08:05   #40826  |  Link
Backflash
Registered User
 
Join Date: Jan 2016
Posts: 52
Quote:
Originally Posted by Schwartz View Post
Sorry but NGU doesn't hold a candle to XBR-75 for image doubling. At least with the material I'm upscaling, which is generally medium-high quality SD, NGU on various settings (tried low and high) gives me way more artifacts and obvious artificial sharpening. I'd like to suggest adding Super-XBR back, it's a great image doubler.
I've been of same opinion, what you need to do is remove sharpeners superres, some antiblaot settings, everything in image enchantment and image refinement except add grain and maybe thin edges. And it looks without artifact and sharp enough.
NGU conflicts with a lot of things like that creating artifacts. Play with settings more.

Last edited by Backflash; 28th November 2016 at 08:21.
Backflash is offline   Reply With Quote
Old 28th November 2016, 08:11   #40827  |  Link
Q-the-STORM
Registered User
 
Join Date: Sep 2012
Posts: 174
Quote:
Originally Posted by sauma144 View Post
@madshi
1) Why not creating presets based on quality or performance for newcomers?
2) Is something like "sync playback to display" from Kodi planned for madVR?
1. madshi wanted to do that eventually, but to be honest, now that a bunch of algorithms have been dropped and performance of NGU is that good, there really is no need for presets... you only have a few options now and most people will simply be using NGU
2. that setting is also messing with audio, which madVR has nothing to do with... you can get the same result if you use 32bit madVR and reclock...
personally, I've always hoped that LAV would implement a slowdown/speedup feature, they would be able to slowdown/speedup audio and video and report the new framerate to the renderer... because that would mean every player that can use LAV would have a speedup/slowdown feature... would also mean I wouldn't have to keep around a 32bit player for PAL DVD content...

Last edited by Q-the-STORM; 28th November 2016 at 08:14.
Q-the-STORM is offline   Reply With Quote
Old 28th November 2016, 08:52   #40828  |  Link
Backflash
Registered User
 
Join Date: Jan 2016
Posts: 52
http://imgur.com/a/qyVl8

I lost ability to read OSD. high into med into "random" downscaler. Is it a bug? or am I confused about steps here?

"random downscaler issue": can we please leave it for downscaling tab domain?
or maybe add what settings mean inside brackets? high(ssim1d100 ar)
it has enough space I think, same with chroma
Only these two confused me, I guess they are great if you never saw previous UI.

In far future(because it's a pain to do) would be nice to have expert and novice mode.
Backflash is offline   Reply With Quote
Old 28th November 2016, 08:54   #40829  |  Link
toniash
Registered User
 
Join Date: Oct 2010
Posts: 131
Quote:
Originally Posted by huhn View Post
sorry i don't see that working for DVDs.

edit: and very common 4/3 AR 16/9 broadcast shouldn't work too.
For low-end cards it will be very useful a NGU 1,5X for 720p -> 1080p
toniash is offline   Reply With Quote
Old 28th November 2016, 09:03   #40830  |  Link
flossy_cake
Registered User
 
Join Date: Aug 2016
Posts: 601
Quote:
Originally Posted by madshi View Post
If you believe that using FreeSync/G-SYNC would be beneficial for video playback, you might be able to convince me to implement support for that by fulfilling the following 2 requirements:
1) Show me an API which allows me to define at which time in the future a specific video frame gets displayed, and for how long exactly.
Here's one example used by Frafs Test Pattern app (DX9). You can use the command line argument to select a fps limit to engage the hires timer and turn on the scrolling bar to check for dropped frames.

Source code of area where hires timer is implemented
https://sourceforge.net/p/frafstestp...ttern.cpp#l872
https://sourceforge.net/p/frafstestp...HiResTimer.cpp

Note: cpu usage is quite high with the hires timer (~15% CPU load ). I was able to eliminate it by adding a 1ms sleep on line 916 of FrafsTestPattern.cpp.

Also
https://msdn.microsoft.com/en-us/lib...(v=vs.85).aspx

Quote:
Originally Posted by madshi View Post
2) Donate a FreeSync/G-SYNC monitor to me, so that I can actually test a possible implementation. Developing blindly without test hardware doesn't make sense.
I hope someone sees that and does. Unfortunately I am broke after getting mine

Since it already works in MPC-HCx64 + MadVR + DX11 disabled + Full Screen Exclusive Mode + SVPTube , is it possible to just have a "windowed exclusive mode"? I have tried setting a hotkey to toggle exclusive on/off while in windowed mode but no effect.


Quote:
Originally Posted by XTrojan View Post
What Freesync/G-Sync will do is just multiply framerate to minimize judder, there will still most likely be judder.
Absolutely not! And why would you think there would be judder? For each 1fps increase or decrease the monitor is matching a perfect 1:1 cadence for that fps. The zig zag line does not mean "this zone is judder", its just showing which multiple of refresh rate is used. As long as it's an exact multiple the cadence will remain 1:1 and no judder or jitter is seen. In practical testing I can assure you that it is seamless even at low frame rates
http://testufo.com/#test=stutter&dem...max=12&pps=720


Quote:
Originally Posted by e-t172 View Post
If you don't match them exactly (such as in your example), you get presentation jitter, and that's bad.
I think you are making an assumption about frame time deltas of 33.4ms (1 dropped frame at 60hz) and trying to apply that to a delta of +/- 5-10ms, which proves you don't own one of these variable monitors because if you did then you would already know that in D3D applications while the fps/hz meter is fluctuating up and down all over the place the motion is still seamless because the inter-frame delta is still much much lower than 1 dropped frame at 60hz. You are still thinking about everything snapping to the nearest 16.7ms which is not the case here. A rapid rise from say 24-90 in gsync will still have an average inter-frame delta that is very low and nowhere near the 33.4ms caused by 1 dropped frame at 60hz.

I would urge you to actually see or use one of these monitors at rapidly varying refresh rates before making judgement based on theory.


Quote:
Originally Posted by e-t172 View Post
Windows is not a real-time OS, so there are no guarantees that madVR will be able to make that call at the appropriate time.
Of course there are no guarantees. At the end of the day the CPU is still running the application regardless of the scenario and relies on not being interrupted by other processes. If the vsync implementation is done by the CPU (such as waiting for a vsync signal) then it will be vulnerable to the same problem if some other process interrupts it.


Quote:
Originally Posted by e-t172 View Post
I think you misunderstood my point. My point was, most people who are in this case (i.e. take the time to learn how to use madVR, worry about cadence…) do not typically use gaming monitors to play their videos on. Instead they use large HDTVs or projection systems, which don't support G-Sync. It's a bit weird to spend all this time getting to the perfect playback system only to use it on a 28" gamer-optimized panel just because it has G-Sync. I'm sure there are people in this case, but it's a very small niche, even among madVR users.
I use it mostly for watching youtube which is primarily a PC activity. The picture quality is just so much better with the edge enhancement, dithering and scaling algorithms etc. plus the colour space is wrong in Chrome (oranges are more red/crimson).

Having to change monitor modes for each video file or youtube video is an annoyance and many monitors don't support the necessary modes (24hz and 50hz which I use quite a lot , most European motorsports broadcasts) which forces me to use the artificial smoothing algorithm at 144hz which is kind of a patch solution, but works quite well actually and full credit to Madshi there .

But Alt+Enter and I get 50hz gsync, perfectly smooth motion without any tricks thus proving gsync is doable and has no significant technical limitations in this application therefore everything you said about it being impossible to do smoothly is untrue because I've seen it with my own eyes (and in games with external FPS limiter RivaTuner )

Last edited by flossy_cake; 28th November 2016 at 09:25.
flossy_cake is offline   Reply With Quote
Old 28th November 2016, 09:24   #40831  |  Link
ryrynz
Registered User
 
ryrynz's Avatar
 
Join Date: Mar 2009
Posts: 3,645
0.91.2 is making some of my profiles disappear and saying there's an error in the rules because it can't find the profile.
Reverting back to 0.91.1 resolves it. Will post more info later if required, no time ATM.
ryrynz is offline   Reply With Quote
Old 28th November 2016, 09:29   #40832  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
Quote:
Originally Posted by Skwelcha View Post
As soon as it tries to upscale it just hangs with the following message [...]
Thanks, will be fixed in the next build.

Quote:
Originally Posted by Sp00kyFox View Post
could you explain to us what the logic is behind the NGU options "chroma/downscale quality: automatic" and "activate only if it's useful"?
"chroma quality: automatic" chooses Bicubic60 AR for doubling chroma, except when using NNEDI3-128/256 or NGU-VeryHigh. In that case NNEDI3-16 respectively NGU-Med is used instead.

"downscale quality: automatic" chooses Bicubic150 AR for downscaling, except when using NNEDI3-128/256 or NGU-VeryHigh. In that case SSIM1D100 is used instead.

"activate only if it's useful" currently activates doubling with a scaling factor of 1.2x or higher, and quadrupling with a scaling factor of 3.0x or higher.

Quote:
Originally Posted by JarrettH View Post
Also noticed that chroma quality is unselectable at low, medium settings - probably because that shouldn't be a selection then.
Correct. At low/medium settings, speed is king, and using any sort of NGU for chroma doubling would slow things down too much. If you have enough GPU power to do chroma doubling, you really should be using NGU-High for luma doubling instead.

Quote:
Originally Posted by JarrettH View Post
What is bicubic60ar for? As a doubler?
It serves as a quick chroma doubler for best "bang of the buck".

Quote:
Originally Posted by Schwartz View Post
Sorry but NGU doesn't hold a candle to XBR-75 for image doubling. At least with the material I'm upscaling, which is generally medium-high quality SD, NGU on various settings (tried low and high) gives me way more artifacts and obvious artificial sharpening. I'd like to suggest adding Super-XBR back, it's a great image doubler.
Can't you use NNEDI3-16 instead? That's the one I left in for low quality SD material.

When I released v0.90.0/1, I asked for feedback which doubling algos could be deleted, and everybody said super-xbr could go as soon as NGU has the same speed. I don't remember anyone saying they would want to keep super-xbr. There were votes for keeping NNEDI3, though.

I'm not saying I won't bring super-xbr back, but I need to be convinced that it's really useful/necessary.

Quote:
Originally Posted by JarrettH View Post
I don't know what you've done with the speed-ups, madshi, but this is a whole lot faster than 0.91.1.
Not sure why it's a lot faster for you, it was only supposed to be slightly faster. But I won't complain.

Quote:
Originally Posted by aufkrawall View Post
Why no more old high + high quadrupling?
Current max quality doesn't look nice to me for quadrupling cartoons. I'd really like to be in charge how luma is scaled, any automatism is just a painful restriction...
The plan is for NGU to soon quadruple directly without needing 2 steps. Which is why I designed the new settings in such a way that you can't choose separate doubling/quad quality levels for NGU. Of course if you pick "High" I could use "High" for both doubling and quadrupling, but there will be users which have a GPU which is too slow for that, while NGU-High + NGU-Med might work for them.

I was thinking that NGU-Med would not show a noticeable difference to NGU-High when used for quadrupling. Maybe I was wrong about that? Do you have a nice screenshot which shows the difference?

Quote:
Originally Posted by MariaX9 View Post
What is the difference between "image downscaling" and the "downscale quality" setting in image upscaling?
Quote:
Originally Posted by Q-the-STORM View Post
image downscaling is for actually downscaling, like 2160p video on a 1080p display, or 1080p video in a 720p window.
downscale quality is for what downscaling algorithm is used after doubling, so for e.g. 720p on a 1080p display, it doubles to 1440p and then downscales to 1080p using what is set in downscale quality...
^

Quote:
Originally Posted by MariaX9 View Post
Also i cant select the very high setting in chroma upscaling, it always switches back to high.
Will be fixed in the next build.

Quote:
Originally Posted by x7007 View Post
Should I use

Chroma Upscaling - NGU Very High + SuperRes 3 ( Do I need SuperRes 3 when using NGU in Chroma Upscaling ) ?
Image upscaling NGU - Luma Very High , Chorma Quality - Auto , Downscale Quality - Auto , Activate only if it's useful
Upscaling Refi - All Disabled
For chroma upscaling NGU Very High might be overkill. But if your GPU can do it, it shouldn't harm.

I wouldn't use SuperRes with NGU. Other than that the settings look good to me.

Quote:
Originally Posted by AngelGraves13 View Post
- Image Downscaling settings aren't used when using Image Upscaling with NGU?
No, that's what the "downscaling quality" option in the "image upscaling" section is for. I intentionally separated "image downscaling" from "downscaling after doubling", because there might be different needs for these 2 different situations, and because I didn't want to use settings from the "image downscaling" page when we're actually *up*scaling the video. For downscaling after doubling in most cases Bicubic150 AR should be good enough, while for true "image downscaling" using SSIM1D might be beneficial.

Quote:
Originally Posted by AngelGraves13 View Post
- NGU Upscaling - Would like to be able to use "User" setting for downscale quality, which would use my User Specified settings in Image Downscaling.
Why?

Quote:
Originally Posted by AngelGraves13 View Post
Also, please add Very High for downscale quality, which would probably use SSIM 2D100%
No, sorry, but that's total overkill, IMHO. I want to protect users from wasting GPU resources on algos which don't really bring any noticeable quality benefit. Using SSIM 2D100 after doubling is such a case, IMHO.

Quote:
Originally Posted by leeperry View Post
But could you please make the NGU+SR nagscreen a one-timer and that's it? It currently shows up everytime I click on OK
Oh well. I suppose that should be possible.

Quote:
Originally Posted by leeperry View Post
I still would like to be able to increase luma NGU and disable chroma NGU if need be, this currently seems impossible?
You can disable chroma NGU by setting "chroma quality" to "normal". You can select luma NGU directly, not sure what you mean there.

Quote:
Originally Posted by leeperry View Post
So doubling "downscale quality" is not related to what we picked in "image downscaling"? That's very confusing.
Is it? Why would the "image downscaling" settings page play any role when we actually upscale the video? I think you're stuck in the old way the settings worked. Once you understood it, it made sense, but it was kinda backwards. The new settings are very strict: The "image downscaling" options are now only used when we actually downscale the video. Every option that plays a role when we upscale the image is now actually in the "image upscaling" settings page. That is not confusing but more logical, IMHO, if you try to forget what you were used to from older madVR builds and look at this with a fresh set of eyes.

Quote:
Originally Posted by leeperry View Post
I would also like to have sxbr back please, sometimes its EE did the magic on low-res SD.
JFWIW, did you ask for super-xbr to stay when I asked for feedback when releasing v0.90.0/1?

Quote:
Originally Posted by Sunset1982 View Post
If I select SuperXBR with SR in the chroma upscale tab, I can't choose the SR strength anymore. Bug or feature?!
Bug. Will be fixed in the next build.

Quote:
Originally Posted by sauma144 View Post
2) Is something like "sync playback to display" from Kodi planned for madVR?
I don't even know what that is/does?

Quote:
Originally Posted by Asmodian View Post
Automatic chroma and downscale quality seems to pick exactly what I would set, at least when not in a placebo mind set.


Quote:
Originally Posted by Q-the-STORM View Post
2. In image upscaling, it would be great if "chroma quality" and "downscale quality" would actually state what alogrithms are being used... sure you can enable it and check OSD, but it shouldn't be too much of an issue to have it say "chroma quality: high (NGU low + bicubic60 AR)"
I'm not sure I want to do that. It might result in new users thinking that Bicubic60 AR can't be good enough and that they should activate some sort of NGU for chroma doubling.

Quote:
Originally Posted by Q-the-STORM View Post
4. When luma high is selected, chorma veryhigh automatically reverts to high. Instead of automatically reverting, maybe display a message that it would be better to select higher luma quality than chroma. The way it is now feels like a bug.
The old "image doubling" logic behaved in a very similar way and nobody ever complained about it.

Quote:
Originally Posted by Q-the-STORM View Post
- with 720p -> 2160p on chroma quality normal OSD shows bicubic60 AR twice "chroma > bicubic60 AR > bicubic60 AR" shouldn't bicubic60 AR do this in one step? maybe just an OSD bug
this only happens with full 2160p, when I have a window resized to slightly smaller than 2160p, it only shows it once..
It's what's actually being done. Don't worry about it. It might seem wrong/wasteful, but it's done intentionally for best speed with NGU, due to how NGU works internally.

Quote:
Originally Posted by Q-the-STORM View Post
- not really a bug but maybe display the same message about superres and NGU when using NGU and superres in chroma upscaling
I thought about it, maybe I should add it.

Quote:
Originally Posted by Backflash View Post
I lost ability to read OSD. high into med into "random" downscaler. Is it a bug? or am I confused about steps here?

"random downscaler issue": can we please leave it for downscaling tab domain?
or maybe add what settings mean inside brackets? high(ssim1d100 ar)
it has enough space I think, same with chroma
Only these two confused me, I guess they are great if you never saw previous UI.
Please try to forget the previous UI logic, that will help understanding the new one better.

Downscaling quality settings are really simple: Low = Bilinear; Normal = Bicubic150 AR; High = SSIM1D100. I don't want users worry about which is which, but just trust in that my preselection makes sense. I think it's much better this way for new/normal users. Maybe power users don't like it, but seriously, would you want to use a downscaling algo other than Bicubic150 or SSIM1D100 after doubling? If so, please explain why.

Quote:
Originally Posted by toniash View Post
For low-end cards it will be very useful a NGU 1,5X for 720p -> 1080p
Yes, but it's not technically possible. NGU will always be a pure integer (2x, 3x, 4x etc) upscaler.

Quote:
Originally Posted by flossy_cake View Post
Here's one example used by Frafs Test Pattern app (DX9). You can use the command line argument to select a fps limit to engage the hires timer and turn on the scrolling bar to check for dropped frames.
It's real time rendered, similar to games, not a movie with pre-recorded frames.

Quote:
Originally Posted by flossy_cake View Post
Source code of area where hires timer is implemented
A hires timer doesn't help. Have you read my FAQ? I asked for an API which defines at which exact point in time a frame will be shown in the future and for how long the frame will stay on screen. I'm not interested in trying to do this myself with a hires timer. The API needs to provide this functionality (and with a "hardware interrupt" supported backend, not with hires timers working in the background), otherwise it's useless for video rendering. Or do you want to have stuttering motion in situations where the CPU is busy for a few milliseconds (which is pretty normal for a Windows PC)? Relying on a hires timer to do the frame syncing would be a big step *back*, compared to the reliable motion smoothness madVR achieves today with a conventional VSync display, and I'm not interested in spending time and money on a solution which is actually a step back in motion smoothness reliability.
madshi is offline   Reply With Quote
Old 28th November 2016, 09:31   #40833  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
madVR v0.91.3 released

http://madshi.net/madVR.zip

Code:
* fixed: NNEDI3 doubling could crash in certain situations
* fixed: NGU chroma upscaling didn't allow "veryHigh"
* fixed: SuperRes strength controls were disabled for chroma upscaling
madshi is offline   Reply With Quote
Old 28th November 2016, 09:33   #40834  |  Link
HillieSan
Registered User
 
Join Date: Sep 2016
Posts: 176
Comparing 0.91.1 and 0.91.2 using RX 480

Code:
0.91.1 settings
------------------
Chroma upscaling: Catmull-Rom
Image upscaling: Catmull-Rom
Luma doubling: disabled
SuperRes: off

0.91.1 with SE and AG both off
-----------------------------------
low    7.5 ms
med   14.3 ms
high  36.5 ms

0.91.1 with SE and AG both 1
-----------------------------------
low    8.5 ms
med   15.0 ms
high  36.5 ms

0.91.2 Settings
------------------
Chroma upscaling: Cubic (=BiCubic50/Catmull-Rom)
SuperRes: off
Quality settings are left on Automatic.

0.91.2 with SE and AG both off
-----------------------------------
low        5.3 ms
med        7.2 ms
high      14.0 ms
very high 42.5 ms

0.91.2 with SE and AG both 1
-----------------------------------
low        6.4 ms
med        8.0 ms
high      15.0 ms
very high 43.8 ms
I am sensitive to judder and GNU med with SR=1 and AG=1 is ok.
GNU high with SR=off and AG=off is a bit too slow.
I hope that the performance can improve a bit more.

EDIT: Did the performance change in 0.91.3?

Last edited by HillieSan; 28th November 2016 at 10:09.
HillieSan is offline   Reply With Quote
Old 28th November 2016, 09:39   #40835  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
Quote:
Originally Posted by ryrynz View Post
0.91.2 is making some of my profiles disappear and saying there's an error in the rules because it can't find the profile.
Reverting back to 0.91.1 resolves it. Will post more info later if required, no time ATM.
Ouch. It's probably due to the removed image doubling settings page. I'm not sure if there's an easy fix...

Would it be a lot of work to redo the profiles in 0.91.1 without image doubling as part of the profiles?

Quote:
Originally Posted by HillieSan View Post
Comparing 0.91.1 and 0.91.2 using RX 480

Code:
0.91.1 with SE and AG both off
-----------------------------------
low    7.5 ms
med   14.3 ms
high  36.5 ms

0.91.2 with SE and AG both off
-----------------------------------
low        5.3 ms
med        7.2 ms
high      14.0 ms
very high 42.5 ms
Not sure why very high measures slower, that shouldn't be the case, unless it's a different downscaling algorithm after doubling?

Generally, I hope that RX480 users will get a significant speed improvement from future drivers.

Quote:
Originally Posted by HillieSan View Post
EDIT: Did the performance change in 0.91.3?
No.
madshi is offline   Reply With Quote
Old 28th November 2016, 09:44   #40836  |  Link
HillieSan
Registered User
 
Join Date: Sep 2016
Posts: 176
Quote:
Originally Posted by madshi View Post
Not sure why very high measures slower, that shouldn't be the case, unless it's a different downscaling algorithm after doubling?
I left downscaling on automatic. I use 1080 -> 2160. There should be no downscaling enabled, right?
HillieSan is offline   Reply With Quote
Old 28th November 2016, 09:47   #40837  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
Right. Not sure why it's slower then.
madshi is offline   Reply With Quote
Old 28th November 2016, 09:52   #40838  |  Link
burfadel
Registered User
 
Join Date: Aug 2006
Posts: 2,229
In 0.91.3, if you enable NGU SuperRes shouldn't be used, and the warning message shows.

HOWEVER, what happens if you select one of the conditions for using NGU, such as 'Activate only if it's useful', or 2.0x scaling? if those conditions aren't met, it would use your other settings (for example, Jinc). In that situation, you would want SuperRes .
burfadel is offline   Reply With Quote
Old 28th November 2016, 09:52   #40839  |  Link
HillieSan
Registered User
 
Join Date: Sep 2016
Posts: 176
New test

Code:
0.91.2 with SE and AG both off
-----------------------------------
low        5.3 ms
med        7.2 ms
high      13.3 ms with CQ: normal, DQ:low
very high 35.0 ms with CQ: normal, DQ:low
CQ = Chroma Quality, DQ = Downscale Quality.

mmm, Chroma Quality set to Automatic is not clear to me. What is the criteria?

Last edited by HillieSan; 28th November 2016 at 10:06.
HillieSan is offline   Reply With Quote
Old 28th November 2016, 09:53   #40840  |  Link
ryrynz
Registered User
 
ryrynz's Avatar
 
Join Date: Mar 2009
Posts: 3,645
Quote:
Originally Posted by madshi View Post
Ouch. It's probably due to the removed image doubling settings page. I'm not sure if there's an easy fix...
Yeah that's what I thought.

Quote:
Originally Posted by madshi View Post
Would it be a lot of work to redo the profiles in 0.91.1 without image doubling as part of the profiles?
Nah, not really.. but if you can't improve things then might want to release an official warning of potential breakage.
ryrynz 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 11:35.


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