madVR - high quality video renderer (GPU assisted) [Archive] - Page 974

mclingo

26th January 2018, 00:28

I have created mk3D file using makemkv.
The file is created correctly because it exctract Mpeg4-3D track
My stand alone player play it correctly with 3D activated.
MadVr not, play it as a normal 2D film.
Is Madvr compatibe with mk3d SBS files?

Hi, lots of variables in playing stuff in 3D MVC frame packed mode, some have already touched on this.

What you've given us is like walking into the Doctors, stating you have a problem and then expecting an accurate diagnosis based on that alone.

Give us a full breakdown of what system you have, what settings you've enabled and what hardware / win software etc you have.

or try this, this how I always set mine up for testing as I know it works, other dont use filters externally but I prefer this method so you are always using the latest versions.

1. Install MPC-HC (easiest to configure for testing)
2. add the latest LAV filters as external filters - 1x audio 1x video 2x splitter - set to prefer
3. untick internal source and transform filters
4. set madvr back to defaults, (you can dial in your own preferences for scaling etc once its working again)
5. turn on all stereo 3d options in MADVR - (4 ticks)
6. Turn on FSE (full screen exclusive mode)
7. Set MPC to open in full screen but make sure autochange monitor mode is unticked.
9. set MADVR as your renderer in MPC

clsid

26th January 2018, 00:46

MPC-HC 1.7.14 (https://forum.doom9.org/showthread.php?t=175209) (includes latest LAV Filters)

jasonwc18

26th January 2018, 01:02

madshi,

Would you mind clarifying how to go about generating a 3D LUT for BT.2020 content mastered to the DCI-P3 colorspace (basically all UHD Blu-Rays thus far)? You mentioned before that DCI-P3 within a BT.2020 container is basically just DCI-P3. However, if I try to generate a 3DLUT for DCI-P3, wouldn't that result in wrong colors since my projector can only accept BT.2020? I've already generated a 3DLUT for D65/Gamma 2.4/Rec.709 with excellent results. I also did a quick manual calibration of SDR D65/Gamma 2.4/BT.2020 which gives pretty good gamma, greyscale and color tracking through 60-70% saturation, as expected, since the projector can't display higher. I plan to use this mode with madvr's HDR --> SDR tone mapping feature. However, I'm at a loss for how to generate a BT.2020 3D LUT. Both DisplayCal and Calman seem to target 100% saturation within BT.2020, which no current display technology can display. Is what I am trying to achieve currently possible? I've done some research online and all the guides for 3D LUT generation refer to Rec.709.

I'm going to answer my own question because I got some very helpful information from the developer of DisplayCal that answers all my questions. That discussion can be found here: DisplayCal Forum (https://hub.displaycal.net/forums/topic/question-about-bt-2020-3d-lut-generation/)

A TLDR version is this. If you want to generate a 3D LUT for displaying 4K Blu-Ray content tonemapped with madvr (pretty much all 4K BDs use 10 bit color, BT.2020 colorspace, and HDR10):

1) Create the 3DLUT for the BT.2020 colorspace and a gamma of your choice. Using DCI-P3 will give incorrect results.

2) It doesn't matter if the TV/projector is unable to display the entirety of BT.2020 (which is going to be true of basically all displays at this time). The profiler will figure out the limits of your display and generate a 3D LUT accordingly. Calman allows you to limit saturation sweeps to DCI-P3, but this is only useful for manual calibration and/or validation.

3) madVR always uses TV levels (16-235) for the 3D LUT encoding. Calman gives an option to set output to PC levels (0-255) but DisplayCal only shows TV levels as an option. The developer says that the 3D LUT should work fine with PC Levels so long as you set madvr and Windows to output PC levels.

Given what I was told, it should be possible to generate a working 3D LUT for BT.2020 with either DisplayCal/ArgylCMS or Calman. There are other options, but these are the two I investigated.

One more tip: If you have an existing BT.709 3D LUT but no 3D LUT in the BT.2020 or DCI-P3 slots, and you want to watch BT.2020 content, you'll need to set madvr to "This Display is Already Calibrated" for BT.2020 and the Gamma you are using. Otherwise, madvr will try to use the BT.709 3D LUT resulting in oversaturated and incorrect colors (at least it does for me as I require a different mode with a color filter for BT.2020 on my JVC RS600). If you have a 3DLUT in either the DCI-P3 or BT.2020 slot, this will be used instead.

huhn

26th January 2018, 01:04

Waiting for the display to match the refresh rate of the title takes all of about 1 second at the beginning of playback and once again at the end of playback. It doesn't bother me but I understand it does others. Once I tried forcing 60Hz for a 23.976 title and using 'smooth motion' algo to see what the improvement was (before understanding those with 60Hz don't really have any other choice). Blur and judder was terrible. I know you say there should be no difference and respect your opinion and how you arrive at your deductions but this is simply my experience. Matching refresh rate has always been very pleasant and smooth as butter for me. My present 4k display is true native 120Hz as was my previous 1080 (both Samsungs). Judder during panning scenes is a pet peeve of mine over any other irregularities. I simply will never own a display under 120Hz and fwiw, both displays advertise as 240Hz using interpolation which I find even more attractive since imo, (through interpolation) adds to the sharpness, clarity, and realism many deem fake and full of miscalculated frames although I've have NEVER experienced those claims.

i hope you didn't forget to disable frame interpolation with SM because SM and frame interpolation will nearly for sure turn into a mess. which alone is enough of a reason for some user to never ever use it.

and if you want to see frame interpolation artifacts just watch a corner of your TV in an action scene it often has to interpolate images without an reference. even through i have to say how can you not notice all these calculations error 3 jointed arms and legs....

a sceen that can't do 23p properly will usually be smooth when frame interpoaltion is engaged and there are screen that can do 23p properly with a 60 hz sony has no problem with that.

and i just repeate my self here again Samsung TV usually can't do 4:4:4 at any other refreshrate as 30 or 60 hz.

Asmodian

26th January 2018, 01:35

@Asmodian. Sorry. Since the entire discussion was about 8bit vs 10bit or higher depending on resolution Hz, thought it went without saying, so rephrasing for better understanding:
HDMI 2.0 doesn't support RGB Full 444 above 30Hz with 10bit or higher color depth but it does support 10bit or higher under 30Hz.

I was trying to point out the pointless fixating on getting 10bit output when it really is the least important aspect. To the point of using subsampled chroma, pure madness.

I simply will never own a display under 120Hz and fwiw, both displays advertise as 240Hz using interpolation which I find even more attractive since imo, (through interpolation) adds to the sharpness, clarity, and realism many deem fake and full of miscalculated frames although I've have NEVER experienced those claims.

I even said matching refresh rates is better. I am not trying to claim smooth motion is just as good, only that it is quite close. Smooth motion breaks any TV's motion enhancements so maybe the native 24Hz studder (especially obvious during pans) is bothering you, not true judder (because smooth motion takes care of that very well). Have you compared 24Hz to 60 Hz with smooth motion with all your TV's processing disabled for both tests? Maybe smooth motion confused your TV's processing so it made the judder even worse (I have seen it happen). Not that you should use it, just you might see why it doesn't bother me too much.

I don't think you have a true 120Hz TV. Samsung's TVs accept up to 60Hz and then in internal processing decide which frames to multiply or interpolate extra frames from. If it isn't madVR presenting the frames and the TV displaying them 1:1 then it is not the true refresh rate. Samsung processing is known to have pretty bad handling of some content. I suppose some true 120Hz 1080p TVs may exist now, e.g. my OLED can take a true 120 Hz input but only at 1080p, but when 120 Hz 4K displays are actually available I will be very happy.

It is funny how some issues bother some people a lot and others not at all. Frame interpolation is not my friend, interpolation artifacts bother me way too much, but 24Hz studder is only a minor annoyance.

I want to see each frame of the source as presented and I always disable as much of the TV's internal processing as possible. All motion clarity enhancements are always disabled. LG doesn't have great processing anyway. ;)

brazen1

26th January 2018, 03:57

Ok. I turned off motion in display settings so its processing doesn't interfere with madVR's. Checked GPU is at 60Hz. Turned off switch to matching display mode in madVR. Enabled smooth motion frame rate conversion for output above 24fps to always. At this point hoping upgrades of GPU drivers, madVR, W10, MPC-BE and LAV will have a different outcome than the last time I did this (not that I need it). I play Cowboys and Aliens opening long pan scene. It's a one to one iso rip. I check display notification no longer shows 3840x2160p at 24Hz and now shows 60Hz. No difference. The blur and judder are unwatchable to me because I'm spoiled, never experiencing it using my regular setup. Put everything back and it's all smooth as butter. No idea why your experiences are better but for sure, mine are night and day differences. I wouldn't go as far as saying Samsung processing is superior but these are my results. Perhaps madVR smooth motion works best with 60Hz displays and with 120Hz doesn't? I don't know? Madshi would. I don't think it should matter.

Yes, my display is 120Hz and not 60Hz. You keep telling me using motion interpolation aka phoney 240Hz, I'm seeing 2 headed people with 7 arms, 3 legs, 11 fingers, and 3 eyes. Oh and to confirm by looking in the corners. Again, I respect you and your opinions and all but that's simply not the case. Samsung is known for very good motion. I know you disagree. Maybe you got a peeve with Samsung or something but I'm honestly telling you the truth. I have nothing to gain. I wish you could sit in front of this set for 20 minutes and have your mind blown to be honest.

Quick google search not including anything from Samsung unless these reviews are liars too:

https://www.rtings.com/tv/reviews/samsung/js7000/settings
Do you think the JS7000 is better at 60Hz native, full array LED, or the JS8500 with 120Hz native and Edge-lit LED?
Between the two TVs, the JS8500 have a better picture quality. 120Hz will allow judder-free 24p

https://www.rtings.com/tv/reviews/samsung/js8500
this is the best TV we've seen for watching sports, partly because you get very little blur, which is ideal for the quick pace of sporting events.

The response time is very fast. We measured it at 11.9 ms on our Samsung UN55JS8500. Motion blur is not a problem on this TV.

There is no judder in movies when using a 24p source (like a Blu-ray player). Sometimes it can do the reverse 3:2 pulldown, but not always, so you might notice a little bit of judder in movies when watching over a 60p or 60i source (like on cable). You can get rid of it when using 'Auto Motion Plus'

This TV is very good for motion blur, so unless you're very sensitive and your last TV was extremely good for motion, you shouldn't have problems. The motion doesn't look exactly like a plasma TV though, because they flicker more than LED TVs.

You won't find a 70" TV with the same picture quality of the JS8500.

Thanks for the great review. Is the extra $500 between the JS7000 and JS8500 worth it? It seems like only difference is 3D, which I don't need, and the JS7000 shows 120 motion rate vs 240 for the JS8500, but I've heard those ratings can be misleading anyway.
You're right that those motion rates are misleading. For Samsung's 4k TVs, the motion rate is double the refresh rate, so the JS7000 has a 60 hz refresh rate, and the JS8500 has a 120 hz refresh rate.
The JS7000 has judder when playing 24 fps movies, and its screen is PLS, so it has a wide viewing angle, but weak contrast. For those reasons, the JS8500 is worth getting instead.

https://www.cnet.com/news/ultra-hd-4k-tv-refresh-rates/
Even high-end 4K TVs, like this Samsung JS8500, have a true refresh rate of at most 120Hz.
Sarah Tew/CNET

Anyway, very happy with my current settings and will just continue to use and enjoy them. Thanks for a great discussion. You guys are good teachers. Perhaps we've all learned a little something from each other and thanks for your input.

huhn

26th January 2018, 08:37

you should compare SM and native 23p playback without any post processing.
if you don't like motion comparable to cinema. fine move on.
Yes, my display is 120Hz and not 60Hz. You keep telling me using motion interpolation aka phoney 240Hz, I'm seeing 2 headed people with 7 arms, 3 legs, 11 fingers, and 3 eyes. Oh and to confirm by looking in the corners. Again, I respect you and your opinions and all but that's simply not the case. Samsung is known for very good motion. I know you disagree. Maybe you got a peeve with Samsung or something but I'm honestly telling you the truth. I have nothing to gain. I wish you could sit in front of this set for 20 minutes and have your mind blown to be honest.

it's not like i can't go into a shop and see the hole samsung TV line up.
if you don't see it fine but there is no flawless interpolation algorithm and there is no way it will ever be flawless. if it doesn't have a reference for interpolation it can't work at all.

sorry but you are not going to blow me away with an old VA 3.5k CR VA panel.

and yet again you don't need a 120 hz panel to display 23p properly it just looks like Samsung need it...

mytbyte

26th January 2018, 08:45

@brazen1: I think you should really set your TV to game mode to make sure all processing is off and then evaluate for 24p stutter or motion blur. (there is nothing about LCD natively that can produce good motion, there is always some underlying processing going on, be it something as simple as PWM blinking...) ..it's not the question whether the panel can do 120 Hz, it's if it can accept 120Hz signal and not process it...

mytbyte

26th January 2018, 08:50

and yet again you don't need a 120 hz panel to display 23p properly it just looks like Samsung need it...

My Panasonic which is 120Hz sure as hell needs some processing, when set to "4K direct" it just reverts to pulldown judder even when fed 24 Hz from GPU...it's odd becuse Panasonic has always had proper 24Hz support (old PV8 plasma deals with it perfectly with no processing)..I was always under impression that LCD panels are multisync and that they can run 24 Hz input as 48Hz (not that it matters for flicker since they're basically sample-and-hold)

huhn

26th January 2018, 09:13

my panasonic has this bug too.
they even send me a techguy to replace my mainboard and i was like "yeah sure doesn't make any sense but good luck".

i'm done with that company.

it would be really nice if someone in Japan could bring this up to them.

ryrynz

26th January 2018, 11:13

it would be really nice if someone in Japan could bring this up to them.

Shouldn't be hard to find someone to talk to about it and get it sorted. Panasonic support on Facebook? or tweet em? Usually easy enough to find someone on a forum who knows someone who works there.
I like the brand, I'd like to see it improve.. If need be I'll contact the support team here if you provide me with the right info, but chances are you could contact higher up just as easily.

huhn

26th January 2018, 11:53

there support is not trained for this. which is not unusual.
the hotline gave me a email and that email said to me.

a PC can't send UHD so you can't use the pure direct mode 4K so there is no issue. please by an panasonic UHD BD player.

so they clearly not trained for this too. and this is as high as i could get. and trust me i tried.

but if you want some info here you go:

model: TX-55CXW704

serial number: ND 5640071

issue:
3:2 judder with pure direct mode.
the TV doesn't add judder if it is not in pure direct mode.

pure direct mode is Panasonic name for "PC mode"

mclingo

26th January 2018, 12:00

Just some info for 60hz vs 23hz – MADVR / OLED users -

I use 23 hz to ensure i'm getting matched rates and zero stutter or drops, AMD cards do this really well with zero fuss as we all know.

However with OLED, TVs panning shots can sometimes look pretty bad compared to LCD as LCD blurs the frames slightly on pans and OLED does not. With OLED you can sometimes see jagged edges as bright elements are panned across the screen, some people don’t noticed this, some do, this is a technology issue though, the screen response times on OLEDs are much quicker than LCDs, this is a known quantity.
This initially didn’t bother me but over time I kept noticing it so I decided to have a play with a few tools. I tried MADVR and SVP, both worked well to deal with these issues but I couldn’t get rid of the odd frame stutter now and again, granted it was only maybe 2-3 per movie but this is still unacceptable to me.

It was then I thought I’d try the TV’s own TRUMOTION, I’d been advised by many people not to go near this but I found it worked perfectly as long as I kept it on a low setting. I now have a very clean image with zero stutters and drops. I use position 3 out of 10

A good scene to test this on is Elysium at the beginning. This is a super long panning shot and at one point a very bright earth pans to one about 1:17. You can see the jagged edges on this and the the space station as that pans too, this looks pretty awful with no motion interpolation. With TRUMOTION its 95% free of issues.

This is also why I dont use PC mode on this TV, I use bluray mode as PC mode disables trumotion.

huhn

26th January 2018, 12:31

ha the good old rubbish of OLED and perfect motion...
while in theory they can have the best they don't have it in practise the closest is the sony A1E with 60 HZ BFI for really good results you need 120 HZ BFI and the "strobing" is only 50 % which is not amazing at all.

so that leaves the good old eye tracking motion resokution at 300-600p LCD can easily do 2160p.

OLED got only one part correct and that's response time. which maeks them generally better than VA panels without BFI but that's it.

and the screen with the best motion handle is an LCD
http://i.rtings.com/images/reviews/monitor/asus/vg248qe/vg248qe-bfi-large.jpg
6 bit TN panel AKA the godfather of motion (CRT are better but what ever...)
even the sony OLEd is clearly not up to the task but LG OLEd can't even reach this: http://i.rtings.com/images/reviews/tv/sony/a1e/a1e-bfi-large.jpg

ryrynz

26th January 2018, 12:31

Yeah same situation for me, I use IFC set on low, it's slightly better than without it and I never see any artifacts, medium is too much.. Wish I had more control over it but oh well.

j82k

26th January 2018, 12:48

Opinions on motion interpolation seem to differ greatly.

When I got my LG Oled I played around with Truemotion alot, tried pretty much all settings but came to the conclusion that it just causes way to many errors for a enjoyable viewing experience, even at a low setting.

It does a good job with simple panning shots, but whenever the scene is more complex and there is a lot of movement of smaller parts going on it often messes up really badly.

Also LG's truemotion can absolutely not deal with flickering lightning, like if a scene is lit by fire or a torch it causes heavy artifacts over the whole picture. Game of Thrones was unwatchable with it enabled.
Or just 2 people going past each other from different directions pretty much always causes visible artifacts.
As much as I hate the slideshow that 24 fps is, motion interpolation with it's errors is even worse for me, I also tried SVP which isn't much better.

Whats really sad is that the technology for smooth 60 fps content is there, as it's within UHD Bluray specs but nobody is making use of it. The only exception being billy lynn.

mclingo

26th January 2018, 13:17

lets not get into an OLED VS LCD flame war please, its subjective, pointless and has been done to death, lets just agree that Micro LED is the way forward , we should all be able to agree on this.

I dont see any artifacts at all and its not just a bit better it loads better on panning shots, I do get artefacts above 3 though so maybe knock it down a notch until they disappear - not sure about later OLED though, mine is first gen and I genuinely have a really crisp clear picture.

NOTE - there have been some firmware related issues with motion on LG OLEDs, mine did have problems at one point but current FIRMWARE is spot on.

mytbyte

26th January 2018, 15:33

@mclingo: MicroLED is not the way, it's signage tech brought down into rich man's batcave to replace a projector...they will never be able to bring it down to the PPI required for normal screen sizes...when you disabled all processing, did you do that with Clear LED option (i.e BFI) as well? When you bring the brightness all the way up, is motion any worse?

@huhn: I may be wrong but I don't see how response time that is less than a frame's duration can attribute to motion blur...it's this misconception that led people to conclude that OLEDs should have great motion because they have minimal response time while in reality they still exhibit motion blur because of sample-and-hold technique....that TN panel you linked iz 144 Hz, so 144 fps motion should be pretty good,but how's with 24 or 60 fps? The photo is taken with BFi on, so motion is sharp because of BFI, not because it's repsonse time is 1 ms

mclingo

26th January 2018, 16:11

@mclingo: MicroLED is not the way, it's signage tech brought down into rich man's batcave to replace a projector...they will never be able to bring it down to the PPI required for normal screen sizes...when you disabled all processing, did you do that with Clear LED option (i.e BFI) as well? When you bring the brightness all the way up, is motion any worse?
s

lol, I know someone would disagree, Samsung have claimed they will have a home sized / priced microled TV sometime in the next 1-9 years, yes this maybe vaporware, we'll just have to see, i'm personally confident they'll manage it.

I always have TRUMOTION on manual, i've not experimented with brightness, however I have the same settings on my 3D profile which has brightness set to 100, no motion issues at all, its just as perfect.

However, if you move to "4" with 3D you get 3D artifacts around moving figures in the scene, these completely disappear for me on 3 and below.

j82k

26th January 2018, 16:45

mclingo

26th January 2018, 16:48

el Filou

26th January 2018, 16:55

@huhn: [...] that TN panel you linked iz 144 Hz, so 144 fps motion should be pretty good,but how's with 24 or 60 fps? The photo is taken with BFi on, so motion is sharp because of BFI, not because it's repsonse time is 1 msWell the photo of the OLED that huhn linked to is also taken with BFI. Nothing would prevent OLED panel manufacturers to use BFI at 144 Hz or even 240, except brightness issues.

this is for deinterlancing and i have no clue what suppose to happen with it because toggling has no effect last time i check and i clearly know what telecine.AFAIK, DXVA's ivtc only works in DXVA Native, and maybe also only with EVR.

j82k

26th January 2018, 17:07

is this the same one?

https://www.dropbox.com/s/ghuz4qpt8vmgzrj/24%20fps%20motion%20test.mp4?dl=0

Yes and in real content I can see the same kind of artifacting which that video produces too.

mclingo

26th January 2018, 17:13

Yes and in real content I can see the same kind of artifacting which that video produces too.

I've used this to check for judder but I wasnt looking at it very closely for artifacts, i'll have a look when I get home later.

el Filou

26th January 2018, 17:16

@mclingo:
Download this youtube video [...] and turn truemotion on.
My C6 can't even interpolate the moving square on the left side without artifacts, It basically breaks apart when it hits the edge, even with dejudder set to 1.That's a nice one! I tested it just out of curiosity on my 2013 Panasonic plasma:
- IFC (renamed '24p Smooth Film') at Min: the squares are pretty much OK, but distracting artefacts on the bigger circle, smaller circle fine
- at Medium: the artefacts on the bigger circle strangely become less obvious, but now the squares especially the left one start showing distracting artefacts
- at High, which switches the panel to 60 Hz instead of 96 Hz: the squares and the bigger circle now are pretty much free of distracting artefacts which is impressive, however the smaller circle now produces massive ghosting artefacts! Also this level gives an awful soap opera effect to everything so it's useless for me anyway.
Back to IFC off. :D

feelingblue

26th January 2018, 17:44

You need the Intel Media SDK plugin (http://files2.codecguide.com/intel_libmfxsw.exe) for decoding 3D H.264 MVC video. This plugin will be used by LAV Video decoder.

MadVR handles SBS video as 2D. Your TV should have an option to turn it into 3D.

Tks, problem solved.
Intel Media SDK is necessary to correctly decode 3D video.
Madvr was not the problem, it works perfectly.

e-t172

26th January 2018, 18:09

I'm going to answer my own question because I got some very helpful information from the developer of DisplayCal that answers all my questions. That discussion can be found here: DisplayCal Forum (https://hub.displaycal.net/forums/topic/question-about-bt-2020-3d-lut-generation/)

A TLDR version is this. If you want to generate a 3D LUT for displaying 4K Blu-Ray content tonemapped with madvr (pretty much all 4K BDs use 10 bit color, BT.2020 colorspace, and HDR10):

1) Create the 3DLUT for the BT.2020 colorspace and a gamma of your choice. Using DCI-P3 will give incorrect results.

2) It doesn't matter if the TV/projector is unable to display the entirety of BT.2020 (which is going to be true of basically all displays at this time). The profiler will figure out the limits of your display and generate a 3D LUT accordingly. Calman allows you to limit saturation sweeps to DCI-P3, but this is only useful for manual calibration and/or validation.

Careful, this really depends on the rendering intent. If you use some kind of perceptual rendering intent for example, you will generate a 3DLUT that tries to preserve all gradients from the source gamut (here BT.2020) when converting to the destination gamut (which is smaller). In other words it will compress, not clip. In the case of BT.2020, which virtually no content make full use of, you really don't want that because it will waste huge swaths of your destination colorspace leaving room for compressed extreme tones from the source colorspace that will virtually never occur in practice. You need to make sure that you're using absolute or relative intent, which will clip instead of compress the out of gamut colors. (Basically it's the difference between "make sure gradients are preserved, even for out of gamut colors" and "make sure the in-gamut colors are reproduced accurately, even if that means losing out of gamut gradients".)

Fortunately, AFAIK relative colorimetric is the default in most tools (and is the suggested choice in the ArgyllCMS video docs (https://www.argyllcms.com/doc/Scenarios.html#TV1) for example), so I believe most people are getting this right.

jasonwc18

26th January 2018, 18:34

You need to make sure that you're using absolute or relative intent, which will clip instead of compress the out of gamut colors. (Basically it's the difference between "make sure gradients are preserved, even for out of gamut colors" and "make sure the in-gamut colors are reproduced accurately, even if that means losing out of gamut gradients".)

Fortunately, AFAIK relative colorimetric is the default in most tools (and is the suggested choice in the ArgyllCMS video docs (https://www.argyllcms.com/doc/Scenarios.html#TV1) for example), so I believe most people are getting this right.

Any idea what Calman uses by default for 3D LUT generation, and if it's wrong, whether it's possible to change this setting?

mytbyte

26th January 2018, 18:44

Well the photo of the OLED that huhn linked to is also taken with BFI. Nothing would prevent OLED panel manufacturers to use BFI at 144 Hz or even 240, except brightness issues.

Yes, I somehow missed the link to sony OLED...the difference must be in the implementation of BFI, not response time (OLED is better than TN)...also I am a bit sceptic of the precision of how Rtings take those photos, I've seen them match the camera move with the onscreen moving logo by hand :eek:

el Filou

26th January 2018, 20:05

also I am a bit sceptic of the precision of how Rtings take those photos, I've seen them match the camera move with the onscreen moving logo by hand :eek:Wait, seriously, like motorsport photographers do? :eek: They should use a fixed high-speed camera and some burst mode.
If that's really the case then that test is of no value whatsoever.

mytbyte

26th January 2018, 20:34

Wait, seriously, like motorsport photographers do? :eek: They should use a fixed high-speed camera and some burst mode.
If that's really the case then that test is of no value whatsoever.

See for yourself: https://youtu.be/vNb3X1AM6uI?t=96

:D

Note: there is a way, explained in the video, to calibrate the camera move, probably try several times before they get it right..the camera must move with the moving object because motion blur occurs when eyes track/focus the motion object, if you look at the motion object with your peripheral vision you'll see there is no smeared motion blur, but stepped trails or no blur...

d3rd3vil

26th January 2018, 20:52

Generally what settings are the best for optimizing 4k HDR material regarding quality? Not that it is needed in general ofc :)
What to select for the sharpest best most brutal result? Super-XBR and SSIM or NGU and Jinc or SSIM or what?

Megalith

26th January 2018, 21:09

Is DXVA2 image downscaling supposed to be bad? I am using a 1080p display, and jaggies are especially noticeable when 4K content is downscaled.

mclingo

26th January 2018, 23:19

@mclingo:
Download this youtube video so you can play it through madvr or whatever and turn truemotion on.
My C6 can't even interpolate the moving square on the left side without artifacts, It basically breaks apart when it hits the edge, even with dejudder set to 1. Same result when playing it directly to the TV through usb.
https://www.youtube.com/watch?v=TVtoxUohG5E

No artifacts?

something odd going on here with my setup, I cant play this smoothly at all never mind the artefacts, something in my chain isnt playing nice with 24hz.

First I played with with just 2160p23 in display modes, then I added 2160p24, no diference, then I tried added 24hz in CRU, no difference, then I removed them and ran it in 60hz, the all look the same, its stuttery as hell. If I play it using MPC HC using standard EVR renderer though it plays smoothly with no artefacts.

Any idea why my MADVR isnt playing 24hz properly?

EDIT - found the problem, its my crappy RX550 it doesnt have enough power to upscale a 720 video to 2160p using NGU sharp medium, its smooth again in 1080p or or 2160p using lanczos instead, I need to upgrade this, very disapointing.

Have to say though, zero artifacts, mine is set on dejudder 3 and deblur 0

mclingo

27th January 2018, 00:00

any progress on the NVIDIA 3D frame drop / repeat bug anyone?

SamuriHL

27th January 2018, 01:14

I'm still on vacation so have not been home to mess with 3d. We'll see if I have any time Sunday or not.

Sent from my Pixel XL using Tapatalk

ashlar42

27th January 2018, 01:20

so that leaves the good old eye tracking motion resokution at 300-600p LCD can easily do 2160p.Easily how? Because "easily" using 6 bit panel is frankly uninteresting from an home theatre point of view.
Also, would you be so kind as to point out the LCD TV offering 2160 lines of motion resolution?

huhn

27th January 2018, 01:57

sony BFI LCDs.
the sony OLED is most likely in the 1080p region for the line testing if not 2160p.

BFI has it's own problem but cinemas use it too.

nsnhd

27th January 2018, 02:03

Is DXVA2 image downscaling supposed to be bad? I am using a 1080p display, and jaggies are especially noticeable when 4K content is downscaled.
Can you link an image of that, what GPU card you're using? I'm using Intel iGPU Hd4600 to downscale 4k on 1080p display with DXVA and haven't noticed any jaggy.

el Filou

27th January 2018, 02:26

DXVA scaling can be very different depending on the GPU.
On my 1050 Ti, upscaling SD to 1080 with DXVA is much sharper than bilinear, but with much more artefacts than madVR's Lanczos or Jinc (I checked the image 'enhancements' are disabled in NVIDIA control panel to be sure).

mytbyte

27th January 2018, 09:38

sony BFI LCDs.
the sony OLED is most likely in the 1080p region for the line testing if not 2160p.

BFI has it's own problem but cinemas use it too.

AFAIK, the motion resolution test that is commonly used is from FPD Bluray and can only measure up to 1080 lines. Those 1080 lines have nothing to do with FHD being 1080 lines vertically, it's just a coincidence because the pattern measures horizontal motion blur and the lines to observe are drawn vertically. This test is, like the whole motion blur thing, largely dependant on perception, different people can state different motion resolution from the same TV

Yes, classic cinema projectors had multiple motion trail and wide area flicker due to the shutter (3x), but most digital projectors don't use shutters except for 3D AFAIK...