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 26th February 2016, 08:04   #36501  |  Link
Uoppi
Registered User
 
Join Date: Oct 2015
Posts: 99
25 fps variable frame rate (according to Mediainfo) gets shown as 60 Hz, when using the "deint=Off" tag (smooth motion is off). Up until now, I don't think madVR has ever struggled with setting the correct refresh rate with variable frame rate.

On a more general level, a few times now I've seen the refresh rate being doubled, even though deinterlacing is off. I don't think that ever happened before, i.e. my TV's refresh rate previously remained at 24/25/30 Hz with 23.976 to 30 fps non-interlaced material. Have some changes been made recently to what values get shown in "Display" and "Composition rate"? At least previously they were showing the actual fps of the video file, not doubling it unless using deinterlacing or smooth motion.

Last edited by Uoppi; 26th February 2016 at 08:30.
Uoppi is offline   Reply With Quote
Old 26th February 2016, 12:22   #36502  |  Link
huhn
Registered User
 
Join Date: Oct 2012
Posts: 7,903
do you have a sample? and media info is not 100% correct.
huhn is offline   Reply With Quote
Old 26th February 2016, 13:57   #36503  |  Link
Stereodude
Registered User
 
Join Date: Dec 2002
Location: Region 0
Posts: 1,436
Quote:
Originally Posted by XMonarchY View Post
Off-topic, but wow.... ripping with HEVC produces BETTER image quality at HALF the size than ripping with medium-high quality h.264, but at the same time it seems kind of a tiny bit smoother-smeared looking, different from h.264, with less noise for sure.
That's not an entirely accurate summary. For UHD/4k content, yes, it's supposed to deliver equal results at half the bitrate. Does it? That's the design target of the codec. That doesn't mean it is delivering on the target today.

Quote:
Is HEVC expected to become the future for both 4K and 1080p content? That leaves everyone without GTX 950/960 at a pretty big disadvantage unless they plan on upgrading GPU when Pascal and AMD's new cards come out...
The standard for what? 1080p Blu-ray is obviously going to stay H.264. Will the "scene" pirate rippers move to HEVC? No idea, and I don't care. HEVC doesn't have the same sort of bitrate savings at 1080p against H.264 that it does for UHD/4k and the amount of devices, phones, tablets, etc. in people's hands that support H.264 and not HEVC means H.264 should be around for a long time.

And why are we having this discussion in the madVR thread?
Stereodude is offline   Reply With Quote
Old 26th February 2016, 16:15   #36504  |  Link
XTrojan
Registered User
 
Join Date: Oct 2015
Posts: 88
Does 3D subtitles work on all subtitles or does it need to contain 3D depth info (which only mkvmerge can get?)
XTrojan is offline   Reply With Quote
Old 26th February 2016, 19:36   #36505  |  Link
James Freeman
Registered User
 
Join Date: Sep 2013
Posts: 919
Fellows, AACS 2.0 is not going to be decrypted anytime soon so it is pointless to prepare to watch UHD discs on our HTPCs.
I bought Spider-Man 2 4k UHD to experiment with madVR and compare versions only to realize it will be dusting on the shelf for a very long time...
Bummer
__________________
System: i7 3770K, GTX660, Win7 64bit, Panasonic ST60, Dell U2410.

Last edited by James Freeman; 26th February 2016 at 19:39.
James Freeman is offline   Reply With Quote
Old 26th February 2016, 19:47   #36506  |  Link
flashmozzg
Registered User
 
Join Date: May 2013
Posts: 77
Quote:
Originally Posted by James Freeman View Post
Fellows, AACS 2.0 is not going to be decrypted anytime soon
Why you say that?
flashmozzg is offline   Reply With Quote
Old 26th February 2016, 19:47   #36507  |  Link
sneaker_ger
Registered User
 
Join Date: Dec 2002
Posts: 5,565
See and discuss here:
http://forum.doom9.org/showthread.php?t=173234
sneaker_ger is offline   Reply With Quote
Old 26th February 2016, 20:38   #36508  |  Link
hu1kamania
Registered User
 
hu1kamania's Avatar
 
Join Date: Apr 2014
Posts: 46
Quote:
Originally Posted by ryrynz View Post
Actually if you could do that with your previous screenshot (Smaug on his treasure) that'd be great.
Same settings as earlier, here is a before and after doubling, with no refinements, aside from SR2:

http://screenshotcomparison.com/comparison/163469

A better example of SuperSampling without refinements
hu1kamania is offline   Reply With Quote
Old 26th February 2016, 20:43   #36509  |  Link
Nevilne
Registered User
 
Join Date: Aug 2010
Posts: 134
Likely just shows that you're doubling in gamma and downscaling in linear light.
In addition, if you make screenshots with sharpening filters, they will probably have different strengths on different resolutions.
Nevilne is offline   Reply With Quote
Old 27th February 2016, 00:35   #36510  |  Link
hu1kamania
Registered User
 
hu1kamania's Avatar
 
Join Date: Apr 2014
Posts: 46
Quote:
Originally Posted by Nevilne View Post
Likely just shows that you're doubling in gamma and downscaling in linear light.
In addition, if you make screenshots with sharpening filters, they will probably have different strengths on different resolutions.
This was purely to show that positive results can be had with super sampling. Doubling 1080p and bringing it back to 1080p.

This wasn't available in previous versions of MadVR.

The only refinement was SR2, no other sharpening filters.

Last edited by hu1kamania; 27th February 2016 at 00:37.
hu1kamania is offline   Reply With Quote
Old 27th February 2016, 01:59   #36511  |  Link
ryrynz
Registered User
 
ryrynz's Avatar
 
Join Date: Mar 2009
Posts: 3,646
Quote:
Originally Posted by Nevilne View Post
Likely just shows that you're doubling in gamma and downscaling in linear light.
Interesting point. If we're supersampling with NNEDI3 should we be downscaling in linear light?

Do you think there is there much point in allowing supersampling with the other scalers madshi? I suspect NEDI, NNEDI3 and super-xbr probably have us covered.
ryrynz is offline   Reply With Quote
Old 27th February 2016, 02:21   #36512  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
Quote:
Originally Posted by Werewolfy View Post
When I said your edge enhancement was the best I've seen I was only referring to actual edge enhancements like Adaptive Sharpen and the sharperning setting on my TV.

Reality Creation is not a real edge enhancement, it's a sort of intelligent high frequency booster. I wasn't a big fan of this filter on my Sony KDL-40W905A because it was associated with a denoiser filter (useful for DVD but I don't want that for Blu-Ray) and I've never found a setting that didn't boost also the flaws in the picture.

But Sony has made great progress and now the results are pretty impressive, I actually found a low setting that never harmed the picture (well, nothing I've seen so far). And like your idea, it works differently based on the resoltion of the picture (if you let the TV make the upscale of course). I don't know if it's thanks to the improve database but I like this filter now.
Too bad you can't do proper screenshots of what the Sony does. Would like to see it.

Quote:
Originally Posted by Stereodude View Post
I'm not sure about that one since the clips is 24/1.001Hz and so was my refresh rate. I tested again last night and confirmed several times that I can play the clip with 0 dropped frames if "don't rerender frames when fade in/out is detected" is checked. With it unchecked I always get dropped frames (in the same spots) and they appear to coincide with fades in the video. Are the internal rendering paths different when that option is checked?
It is quite possible that the feature still makes problems. But the log you created didn't seem to contain frame drops caused by the feature. My best guess right now is that you got frame drops for other reasons in that log and maybe stopped playback too early to get the frame drops caused by fade detection. So maybe you can try again creating another log, or two, and maybe let it run a bit longer and let a few more frame drops happen, for which you're sure they're caused by fade detection.

Quote:
Originally Posted by baii View Post
Interesting addition, wondering which algorithm combination will work for 4k YouTube.
Or do it even give benefit vs just doing image enhancements. YouTube really need some sharpening~.
Sharper downscaling should already help a lot. If you want cheap performance with sharp downscaling, try Bicubic150.

Quote:
Originally Posted by evil-doer View Post
.11 Not only had black OSD for me, it also played the video with an all black screen, until i resized or full screened the video.

.12 didnt fix the black screen issue

edit: It also depends on the video.. Some open to a black screen while others do not.

edit2: One difference I see with the problem files is that they are 1280x718
Quote:
Originally Posted by Scyna View Post
I have the same problem on an amd 290. I find that changing the image doubling from NNEDI3 to anything else fixes it.
Can you guys upload a small sample, together with your "settings.bin" file and a screenshot of the black screen with the Ctrl+J OSD turned on? Maybe also a debug log would help. Thanks!

Quote:
Originally Posted by clsid View Post
Using the rate of frames that madVR gets as input would be more consistent with the values reported by other renderers, and imo be more useful information wise. For example if I play at 2x speed, the reported value doubles with the other renderers.
Other renderers also seem to use less values for calculating the average, so they converge faster to the current rate. Vanilla EVR even resets the calculation when changing the playback speed.
Ok, but do you actually make use of this information for anything important? Or is it just a cosmetical issue? I've so many things on my plate right now that I'd like to pass on cosmetical stuff atm.

Quote:
Originally Posted by Georgel View Post
I have created a little introductory page on my project's main page, on how to properly install and set madVR for best experience, especially a few tips for when using laptops.

I found myself in a pinch because of Nvidia Optimus, but now I seem to have figured out that there were a few simple tricks to be done.
Your images don't work for some reason. Can you do a very short summary of which tricks you had to use exactly to make Optimus work better?

Quote:
Originally Posted by Manni View Post
By the way, while looking at this pattern (once I had solved the scaling issue), some of the new chroma upscaling algos fail this test. For example, reconstruction soft fails and Super-XBR above 75 significantly raises the brightness in a way which can't be good or accurate, even if it does look sharper.
Which test pattern is that? It's always "dangerous" to base test results on just one test pattern. I've seen different chroma upscaling algorithms do very bad on one test pattern and then suddenly very good on another. So judging final chroma upscaling quality is pretty hard to do.

Quote:
Originally Posted by Manni View Post
Yes, I know about these settings in the player, as I say they have NOTHING to do with MadVR's decision to rescale a frame if it has black borders when the auto A/R is enabled. My player is set to normal size, no zoom, original A/R etc, and if I switch to EVR-CP (or if I switch MadVR's auto A/R off), as mentioned earlier, there is no scaling/zooming problem (but then of course, there is no auto A/R from MadVR).
Quote:
Originally Posted by Manni View Post
Sure, here are two screenshots and the auto AR options:

Auto AR off: https://www.dropbox.com/s/hzek8rabi5...f.png?dl=0Auto
AR on: https://www.dropbox.com/s/aiwl6wnldj...%20On.png?dl=0
A/R options: https://www.dropbox.com/s/jom6ms9kr7...tions.png?dl=0

My player is always set to 100% / no zoom. The issue goes away if I deselect the black bar detection in MadVR or if I switch to EVR-CP in MPC-BE.
Quote:
Originally Posted by Manni View Post
Thanks, I fixed the link.

Video frame is set to touch window from the inside and keep aspect ratio. That's the way it should be.
Your posts are somewhat contradicting. Sometimes you say you have set the player to "100% / no zoom", and sometimes to "touch window from inside", which are 2 very different settings (only if the cropped video resolution doesn't match the screen resolution, though).

Generally, madVR tries to understand what the media player wants, which should be what *you* want, because the media player is supposed to have control over zooming behaviour. Things are rather complicated, though, because the whole "zoom control" settings somewhat conflict with that approach.

Anyway, if the media player is set to "100% / no zoom", and if madVR understands that, then madVR will not zoom, even if you activate any of the madVR zoom control settings. So if you really do not want madVR to do any zooming, then please set your media player to "100% / no zoom". However, and here comes a small catch: Only MPC-HC so far really tells madVR which zoom setting is active in the media player. All other players (including MPC-BE) do not do that, although I asked the devs to, so for all other players madVR has to guess the media player zoom mode. And that doesn't always work reliably. E.g. if you play a 1080p movie on a 1080p display, "100%" and "touch window from inside" have the same effect. In this situation madVR isn't able to guess properly which zoom mode the media player is set to. In this situation MPC-HC produces the best results because it reliably tells madVR which zoom mode the user has selected.

You can work around the issue by activating "crop black bars" *and* "notify media player [...]". This combination of settings results in the media player being forced to send a proper zoom rectangle to madVR which allows it to do a better guess which zoom settings your media player is set to. So if you're using MPC-BE, set it to 100% and activate those 2 madVR settings and you should be set.

Of course setting your media player to 100% means SD stuff is not upscaled, anymore...

Quote:
Originally Posted by thewebchat View Post
madVR since 0.90.11 has crashed w/ Skylake iGPU (HD 520) on videos with dimension 702x480. Default settings, 64-bit.
Crash report?

Quote:
Originally Posted by chros View Post
Madshi, can make it multithreaded, since especially 4k can profit of the cropping feature?
It's on my to do list.

Quote:
Originally Posted by Uoppi View Post
Could someone clarify what "crop black bars" is supposed to do? Should it zoom/scale the bars away or keep them visible (providing player is not set to zoom and madVR is set to lose no image content)?
"crop black bars" physically removes the black bars, so madVR has less pixels to work with. Should improve performance a bit, and of course logos (and hard coded subtitles!) are cut off, too.

If you disable "crop black bars", the image should still be shown with the same (or nearly the same) zoom factor and position, but the black bars are not physically cut off, so if you manually zoom out, logos and subtitles in the black bars are visible again.

Quote:
Originally Posted by huhn View Post
depends.

but in most cases you have to scale the image more which could be slower.
I think there should always be a (small) performance benefit. Scaling factor should not be affected. The media player might change the window size, though (if "notify media player" is activated), which *may* affect scaling factor, once again. <sigh> Complicated...

Quote:
Originally Posted by David View Post
What do i need to watch hdr with madVR if my TV is 10 bit and WCG?

- Last LAV nightly build
- Setting madVR ex. fullscreen+d3d11+10 bits and (?) DCI P3 in set is already calibrated
- Maximizin luminance and contrast TV setting(?)
- Setting TV WCG
- Ajusting madVR nits

With this settings i think madVR will send 10 bits DCI P3 (REC 2020) with dinamic range compressed to luminance settings. Am i correct?

What about PQ EOTF (ST 2084)? What does madVR with it?
Sounds about right. 10bit output is not really necessary, thanks to high quality dithering 8bit should suffice, but of course 10bit doesn't harm, either. Currently madVR always converts the PQ EOTF to a gamma curve. Whether your screen really supports full 10bit is something you might want to test. Simply use some test pattern to check whether you get less banding letting madVR output 10bit or 8bit.

Quote:
Originally Posted by Uoppi View Post
I'm seeing an insignificant performance boost so I've decided to just disable "crop black bars".

Half of the time I don't even understand why madVR is scaling, so I guess I'm better off not using that feature, LOL. For example, because of the scaling, some videos even end up playing under a different/unexpected madVR profile (i.e. not the profile that would be used if "crop black bars" was disabled).
There are variations of the profile vars available if you want profiles to be selected according to the original resolution (before cropping). E.g. you can use "uncroppedSrcWidth" instead of "srcWidth".

Quote:
Originally Posted by Uoppi View Post
25 fps variable frame rate (according to Mediainfo) gets shown as 60 Hz, when using the "deint=Off" tag (smooth motion is off). Up until now, I don't think madVR has ever struggled with setting the correct refresh rate with variable frame rate.
"shown as" (meaning listed in the OSD) or "shown at" (meaning actual GPU output is) 60 Hz? Maybe a screenshot of the OSD could help clarify that, together with a list of your madVR display mode switching settings.

Quote:
Originally Posted by Uoppi View Post
On a more general level, a few times now I've seen the refresh rate being doubled, even though deinterlacing is off. I don't think that ever happened before, i.e. my TV's refresh rate previously remained at 24/25/30 Hz with 23.976 to 30 fps non-interlaced material. Have some changes been made recently to what values get shown in "Display" and "Composition rate"? At least previously they were showing the actual fps of the video file, not doubling it unless using deinterlacing or smooth motion.
Again I'm not sure what you mean. E.g. are you just talking about some numbers in the OSD, or are you talking about the actual refresh rate the GPU is outputting?

Quote:
Originally Posted by XTrojan View Post
Does 3D subtitles work on all subtitles or does it need to contain 3D depth info (which only mkvmerge can get?)
Should work on all subtitles, even external SRT files, as long as you use the latest nightly LAV build. However, without the proper Blu-Ray playlist madVR receives a bunch of depth information without knowing which of that is meant for which subtitle track or which is meant for OSD etc. In that situation madVR will "guess" which depth (out of all the depths it receives) is likely to be a reasonable choice. Only when the playlist is available, LAV/madVR know exactly which depth information was meant for exactly the one subtitle track you've selected.

Quote:
Originally Posted by ryrynz View Post
Do you think there is there much point in allowing supersampling with the other scalers madshi?
I considered it, but found no good way to offer it in the settings dialog. Maybe some time in the future.

Last edited by madshi; 27th February 2016 at 02:24.
madshi is offline   Reply With Quote
Old 27th February 2016, 02:32   #36513  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
When releasing v0.90.11 about a week ago I asked you guys for feedback about the following things:

1) Agressive downscaling anti-ringing filter: Is it working well for you, or do you see any problems?
2) New sharpening anti-ringing filter: Is it working well for you, or do you see any problems?
3) crispen edges: Do you prefer the linear light option to be enabled or disabled?

So far from what I remember I've received zero feedback about any of that. Of course you don't *have* to provide feedback, but if you don't, then I get the impression that I develop too fast for you guys to keep up. Which means I'd have to slow down development accordingly to allow you to catch up. Your choice...
madshi is offline   Reply With Quote
Old 27th February 2016, 02:52   #36514  |  Link
clsid
*****
 
Join Date: Feb 2005
Posts: 5,643
Quote:
Originally Posted by madshi View Post
Ok, but do you actually make use of this information for anything important? Or is it just a cosmetical issue? I've so many things on my plate right now that I'd like to pass on cosmetical stuff atm.
It's not important. Put it somewhere low on the todo list.
__________________
MPC-HC 2.1.7.2
clsid is offline   Reply With Quote
Old 27th February 2016, 03:17   #36515  |  Link
hu1kamania
Registered User
 
hu1kamania's Avatar
 
Join Date: Apr 2014
Posts: 46
Quote:
Originally Posted by madshi View Post
When releasing v0.90.11 about a week ago I asked you guys for feedback about the following things:

1) Agressive downscaling anti-ringing filter: Is it working well for you, or do you see any problems?
2) New sharpening anti-ringing filter: Is it working well for you, or do you see any problems?
3) crispen edges: Do you prefer the linear light option to be enabled or disabled?

So far from what I remember I've received zero feedback about any of that. Of course you don't *have* to provide feedback, but if you don't, then I get the impression that I develop too fast for you guys to keep up. Which means I'd have to slow down development accordingly to allow you to catch up. Your choice...
The most testing I did was with SuperSampling, and these were my preferred settings:

Image Downscaling: SSIM 1D 100% AR LL
Image Upscaling: Jinc AR
Image Doubling: NNEDI3, 32
Upscaling Refinement: Crispen edges: 1.0 w/ Linear Light and AR
Upscaling Refinement: Enhance detail: 1.0
Upscaling Refinement: SuperRes: 2 w/ linear light
Chroma Upscaling: NNEDI3 32

The short answer would be that I prefer the new settings.

I've been using them when applicable, I don't always refine with crispen edges. Hopefully the experts weigh in with a more detailed analysis!
hu1kamania is offline   Reply With Quote
Old 27th February 2016, 04:22   #36516  |  Link
Sparktank
47.952fps@71.928Hz
 
Sparktank's Avatar
 
Join Date: Mar 2011
Posts: 940
I use:
chroma > ReconSoft (SR3)
luma < SSim 1D200 AR LL
Upscaling Refinement: AR ~2x
Crispen edges: 1.0 LL
SR3 LL
Image Enhancements: AR
CE LL

and it hasn't blown up the computer (yet). On bluray remuxes.
That's including using SVP to interpolate to double framerates. (at ~48)
__________________
Win10 (x64) build 19041
NVIDIA GeForce GTX 1060 3GB (GP106) 3071MB/GDDR5 | (r435_95-4)
NTSC | DVD: R1 | BD: A
AMD Ryzen 5 2600 @3.4GHz (6c/12th, I'm on AVX2 now!)
Sparktank is offline   Reply With Quote
Old 27th February 2016, 05:57   #36517  |  Link
neno
Registered User
 
Join Date: Feb 2014
Posts: 10
Hi, madshi:
There is a problem when I use madvr.(I tried the lastest version madvr and several players like MPC-BE and potplayer. Nothing changes. This problem disappears when I use other renderers, or change my videocard from R260X to GTX950.)
The problem is: the topmost line and the leftmost line of the video picture seems strange. For example, when the whole picture changes from white to black, the topmost line and the leftmost line keep white a few seconds and then gradually change to black with flash, just like they can not keep pace with the whole picture.
I tried to change some settings like scaling in madvr, and still can not solve it.
Can you please give me some suggestions? Thanks!
neno is offline   Reply With Quote
Old 27th February 2016, 07:07   #36518  |  Link
6233638
Registered User
 
Join Date: Apr 2009
Posts: 1,019
Quote:
Originally Posted by madshi View Post
1) Agressive downscaling anti-ringing filter: Is it working well for you, or do you see any problems?
I've just spent a good few hours doing testing with downsampling - mainly looking at Blu-rays instead of test patterns - and what I've seen so far is that the anti-ringing filter is massively improved.
Something may show up in artificial tests, but nearly every filter is now a viable option with linear light downscaling with the new anti-ringing filter, and I haven't noticed any drawbacks.

And speaking of downscaling, SSIM seems like a pretty good option.
Personally I think it's a bit sharp - especially on fine details - as aliasing can appear, but at the same time that kind-of seems to be the point of it. Get the image as sharp as can be, without adding any ringing.
I don't know if something like a 10% or 15% option might be possible, but if I'm not looking at it with a critical eye, 1D 25% looks pretty good most of the time, with the exception of very detailed/busy shots, or bad sources which already contain some degree of ringing.
6233638 is offline   Reply With Quote
Old 27th February 2016, 10:05   #36519  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
Quote:
Originally Posted by clsid View Post
It's not important. Put it somewhere low on the todo list.
Ok, will do.

Quote:
Originally Posted by hu1kamania View Post
The most testing I did was with SuperSampling, and these were my preferred settings:

Image Downscaling: SSIM 1D 100% AR LL
Image Upscaling: Jinc AR
Image Doubling: NNEDI3, 32
Upscaling Refinement: Crispen edges: 1.0 w/ Linear Light and AR
Upscaling Refinement: Enhance detail: 1.0
Upscaling Refinement: SuperRes: 2 w/ linear light
Chroma Upscaling: NNEDI3 32

The short answer would be that I prefer the new settings.

I've been using them when applicable, I don't always refine with crispen edges. Hopefully the experts weigh in with a more detailed analysis!
Quote:
Originally Posted by Sparktank View Post
I use:
chroma > ReconSoft (SR3)
luma < SSim 1D200 AR LL
Upscaling Refinement: AR ~2x
Crispen edges: 1.0 LL
SR3 LL
Image Enhancements: AR
CE LL

and it hasn't blown up the computer (yet). On bluray remuxes.
That's including using SVP to interpolate to double framerates. (at ~48)
Thanks guys. Did you have a chance to compare Crispen edges with Linear Light and/or AR turned on vs off? Did you like LL on better? Or was turning on LL more of a "random" choice?

Quote:
Originally Posted by neno View Post
There is a problem when I use madvr.(I tried the lastest version madvr and several players like MPC-BE and potplayer. Nothing changes. This problem disappears when I use other renderers, or change my videocard from R260X to GTX950.)
The problem is: the topmost line and the leftmost line of the video picture seems strange. For example, when the whole picture changes from white to black, the topmost line and the leftmost line keep white a few seconds and then gradually change to black with flash, just like they can not keep pace with the whole picture.
I tried to change some settings like scaling in madvr, and still can not solve it.
Can you please give me some suggestions? Thanks!
That sounds quite weird, haven't ever heard of something like this before. Are you using any sort of hardware decoding? E.g. CUVID, native DXVA or copyback DXVA? If so, try software decoding to make sure it's not caused by hardware decoding.

Quote:
Originally Posted by 6233638 View Post
I've just spent a good few hours doing testing with downsampling - mainly looking at Blu-rays instead of test patterns - and what I've seen so far is that the anti-ringing filter is massively improved.
Something may show up in artificial tests, but nearly every filter is now a viable option with linear light downscaling with the new anti-ringing filter, and I haven't noticed any drawbacks.
Wonderful, very helpful feedback! And of course I very much like your findings! So you did not find any aliasing caused by the new AR? What is your opinion about LL scaling now? I'm considering moving both LL and AR options for downscaling into the "trade quality for performance" section. Which would mean I'd consider highest quality to be achieved with both LL and AR turned on. Would you agree with that assessment?

Do you like sharpening algos? If not, I don't want to bother you with tests. But if you do, I'd like your feedback about Crispen edges linear light on vs off, and whether the sharpening AR filter works well for you or not.

Quote:
Originally Posted by 6233638 View Post
And speaking of downscaling, SSIM seems like a pretty good option.
Personally I think it's a bit sharp - especially on fine details - as aliasing can appear, but at the same time that kind-of seems to be the point of it. Get the image as sharp as can be, without adding any ringing.
I don't know if something like a 10% or 15% option might be possible, but if I'm not looking at it with a critical eye, 1D 25% looks pretty good most of the time, with the exception of very detailed/busy shots, or bad sources which already contain some degree of ringing.
Ouch, I find 1D 25% already so low that it barely makes a difference for me. Do you really want/need less than that? It would not be a big problem to add, but I somewhat question if you'll see much of a detail improvement at all with such low values...
madshi is offline   Reply With Quote
Old 27th February 2016, 10:34   #36520  |  Link
Sunset1982
Registered User
 
Join Date: Sep 2014
Posts: 280
Quote:
"crop black bars" physically removes the black bars, so madVR has less pixels to work with. Should improve performance a bit
Ok, this has nothing to do with feedback to you actual features, but I find this one interesting so I'll post it.

Just did some testing with the crop black bars feature. Tested a 720p File which is upscaled to 2160p by my GTX960. Activating the crop black bars option brings me a better performance by 12ms rendering time.

man, thats a HUGE improvement!!!
__________________
Intel i5 6600, 16 GB DDR4, AMD Vega RX56 8 GB, Windows 10 x64, Kodi DS Player 17.6, MadVR (x64), LAV Filters (x64), XySubfilter .746 (x64)
LG 4K OLED (65C8D), Denon X-4200 AVR, Dali Zensor 5.1 Set
Sunset1982 is offline   Reply With Quote
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 23:33.


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