Shutter Glasses - Active 3D Displays Like NVIDIA 3D Vision

Advanced display talk, display hackers, advanced game programmers, scientists, display researchers, display manufacturers, vision researchers. The masters on Blur Busters.
rustyk
Posts: 6
Joined: 09 Nov 2018, 14:22

Shutter Glasses - Active 3D Displays Like NVIDIA 3D Vision

Post by rustyk » 09 Nov 2018, 14:42

Hello to all the experts here! Please bear with me as this is related to high frequency displays.
Sorry for the rambling post but I'm hoping to get some useful feedback.

Over on the Nvidia 3d vision forum we're desperate to get good quality 3d and options are limited.
Nvidia's own solution these days basically comes down to 120HZ TN LCD screens with lightboost (3d vision native) or Nvidia 3d tvplay, which uses the normal HDMI framepacked 3d methods and requires a display which natively supports 3d.
So let's forget passive displays and 3dtv and just concentrate on high frequency displays which in theory could support active 3d.

The point I'm getting at, is that with a high enough refresh rate and a low pixel persistence, there are ways to get 3d vision to activate on any 120Hz display. The issue is that the results are all over the place.

I tried a LG C7 OLED and the results were appalling, presumably something to do with lack of backlight strobing? Even though the OLEDS have tiny pixel transition times, it just didn't work, there was terrible persistent crosstalk.
I understand that lightboost is equivalent to ULMB but also overdrives the panel voltage to regain lost brightness?

There are an increasing number of TVs these days which support 1920x1080x120Hz input natively, so I'm wondering, other than testing them with a 3d vision kit is there a systematic way of identifying suitable displays from specification alone?

In my head, in terms of pixel response times with current displays, it goes something in the order of DLP~OLED>TN>IPS>VA.

Most of the TVs I've seen are VA panels, but some of them support BFI @ 120Hz for a 120Hz input (according to RTINGS). I'm assuming that BFI will reduce the brightness, but will also mask the pixel transition times and crosstalk between the L-R lenses on the LCD shutter glasses.

Do you think there are any modern TVs that are suitable candidates for working with active 3d and why? Thanks for any feedback.

User avatar
Chief Blur Buster
Site Admin
Posts: 6509
Joined: 05 Dec 2013, 15:44

Re: 120hz+ Active 3d displays - Nvidia 3d vision

Post by Chief Blur Buster » 10 Nov 2018, 00:34

Great questions. I'm an expert in displays, and I totally understand why this happens.

However, the assumptions that many people make about pixel response is WRONG about glasses.
There are two assumptions that I need to educate, in order to explain better:

A) Pixel response time is not necessarily chief cause of crosstalk
A. Pixel response time (GtG) is not the same thing as pixel persistence (MPRT).
Useful reading:
1. See Why Does Some OLEDs have Motion Blur?
2. See Blur Busters Law And The Amazing Journey To Future 1000Hz Displays

B) Not all pixels are refreshed at the same time
C) NVIDIA 3D Vision uses a strobe backlight
Please see this high speed video. A screen (even OLED) is refreshed sequentially from top-to-bottom. Pixel response is only for one pixel, but that doesn't mean the whole screen's of pixels are refreshed simultaneously.
phpBB [video]

From High speed video of LightBoost
The LightBoost mode (normally for 3D Vision) actually helped Blur Busters become popular, because this mode eliminated display motion blur. Such a display mode, perfect for 3D Vision, was also perfect for eliminating LCD motion blur. Even LightBoost has much lower MPRT response than the world's fastest OLED.

Remember...
GtG response is not MPRT response
GtG response is not MPRT response
GtG response is not MPRT response

GtG is pixel transition time (pixel changes from one value to another in a finite amount of time)
MPRT is pixel visibility time (pixel stays a fixed visible color for a finite amount of time)
From: Blur Busters Law And The Amazing Journey To Future 1000Hz Monitors

For good 3D shutter glasses operation, it becomes more complicated than this, because of the screen scanout effect (which you see in the above video too).

The LCD in a shutter glasses has a "response time"
(often slower than display!!)


Shutter glasses uses LCD -- the same technology in a common monitor.

There are only two pixels in shutter glasses: One massive lens-sized pixel for your left eye, and one massive lens-sized pixel for your right eye.

The big problem is that the shutter glasses have a finite response time itself (shutter fading between opaque-vs-transparent or transparent-vs-opaque in a finite time period), so the display actually has to behave unorthodoxly to make them work. LCD shutters are like one massive lens-sized LCD pixel.

But how the hell do I fix that?

Black frame insertion. Strobe backlight.

The main workaround for fixing shutter glasses crosstalk for the screen scanout problem is to use a form of black frame insertion to black out the screen BEFORE the glasses changes shutters (remember: LCD shutters "fade from opaque to transparent" over milliseconds.... so it's better for the screen to be blacked out during this moment!!!).

Black frame insertion can reduce persistence. Black frame insertion is used to black out the screen while the shutter glasses switch eyes. It can take a few milliseconds for a shutterglasses shutter to fade from "on-to-off" or "off-to-on". During this time period, The screen must be COMPLETELY BLACK in order to AVOID CROSSTALK. When a shutter glasses is switching eyes, one shutter is fading to black and the other shutter is fading to transparent (over milliseconds). That means both eyes can be semi-transparent. So even the world's fastest instant-pixel-response displays, WILL have very bad crosstalk.

Such screens that do this (aka NVIDIA LightBoost) naturally ends up having very low MPRT because a specific pixel is visible for a short time period. For years, Blur Busters had a LightBoost HOWTO and a LightBoost FAQ, how to commandeer and trick a 3D Shutter Glasses Mode into a Motion Blur Reduction Mode, as it was hugely much more profitable to sell this feature as a motion blur reduction feature. And now there are lots of LightBoost copycats but they're targetted for motion blur reduction, not for 3D shutter glasses.....

Shutter Glasses Fighting Against Display

Shutter glasses in one refresh cycle, this simultaneously occurs:
1. Left eye "LCD-shutter" gradually fades to black (or vice versa); and
2. Right eye "LCD-shutter" graduallt fades to clear (or vice versa)

Display
1. Top to bottom refresh over one refresh cycle
2. For each pixel, there's pixel response too (creates fade wipe effect in high speed video)

That's your attempt to hack most displays (LCD, OLED) display for 3D glasses

Shutter Glasses Working With Display

Shutter glasses in one refresh cycle:
1. Left eye "LCD-shutter" fades to black slowly (or vice versa)
2. Right eye "LCD-shutter" fades to clear slowly (or vice versa)

Display
1. Refreshes in total darkness (80-90%) of refresh cycle
2. Display brightly flashes fully refreshed frame at the perfect moment for the shutter glasses
(when closed shutter is at its most opaque, and when open shutter LCD is at its most clear.

That's what NVIDIA LightBoost is. But you can reverse-commandeer a blur-reduction backlight into a 3D glasses mode!

Reverse Hack? Yes, That's The Good News

Hacking a 3D glasses mode for use as 2D motion blur reduction
In the past, we commandeered a 3D Shutter Glasses Mode into a Blur Reducing Mode. That's what Blur Busters did. We're the website that popularized the "LightBoost Hack" for this use, and #1 when you google "LightBoost".

Reverse Hack is possible!!! Hacking a 2D motion blur reduction mode into becoming a 3D glasses mode
However, the vice-versa CAN be done with certain exceptions:
A. Look for a display that advertises motion blur reduction
B. Make sure 120 Hz is supported with that mode.

Overdrive artifacts causes a crosstalk problem. There's a way to massively reduce this. To reduce cross talk (caused by overdrive) further, make your colors slightly duller (about 10-20%) -- reduce dynamic range slightly (LightBoost-style poor colors) by raising digital black level and reducing digital white level (via NVIDIA Control Panel or AMD Catalyst Control Center) or configuring the computer to use only HDMI range (16 to 235). LCD panel overdrive is problematic at the fullblacks and fullwhites so artifacting occurs with those. http://www.testufo.com/crosstalk is a very good test pattern for 2D blur reduction but is also 100% applicable to quality you can achieve with a mode. This extra digital headroom (not backlight brightness) below black and above white, reduces the overdrive artifact problem that creates amplified crosstalk.

These modes work adequately with 3D glasses
A. NVIDIA ULMB (in monitors that don't have 3D Vision)
B. BENQ Blur Reduction on 1080p monitors with the Large Vertical Total trick
C. Samsung Blur Reduction
D. ASUS ELMB

Remember, however, to use the dynamic range reduction trick (make your blacks digitally brighter, and makes your whites digitally darker) to fix the overdrive ghosting that occurs with those end-range colors. Doing that reduces ~80% of the crosstalk visibility in the reverse hack (commandering a blur reduction mode into a 3D glasses mode). This is exactly a major reason why LightBoost had crappy color, it was pre-compressed dynamic range. But you can intentionally do it with most blur reducing modes to reduce crosstalk.

Remember, TN LCD screens are always best. Also, global-flash strobe backlights are better for 3D glasses than sequential scanning backlights (which attempt to compensate for the screen scanout effect by flashing fully refreshed rows away from the still-refreshing areas of the screen).

Pixel transition time is unimportant if the pixel transition time occurs in total darkness -- that's how LightBoost works and it's possible to have MPRT lower than GtG. (This is how Blur Busters got started... LOL) For 120Hz, a backlight is flashed for only 1ms to 2ms which means the screen is completely dark almost 90% of the time, with a photographic-flash-quickness of a screen flashing at the very moment that the shutter glasses is in its best situation (most opaqueness of the shuttered eye, and most clearness of the non-shuttered eye) because the shutter glasses eyes fade back and fourth in a sinewavey way (point a 1000fps camera through shutterglasses pointing at a bright lightbulb -- and you'll see what I mean), so making screen visibility brief only at the most optmium moment of the shutter glasses -- minimizes crosstalk. Thanks to this, this gives an LCD screen plenty of time to finish GtG pixel transition in total darkness -- completely unseen by eyes -- (1ms TN screens especially -- given LightBoost gives 6ms- 7ms of total darkness between screen flashes) -- that really lowers crosstalk.

For good "3D Shutter Glasses Hack Displays", make sure that the display supports 120Hz strobed (black frame insertion). Some displays that have a Motion Blur Reduction Backlight will make acceptably decent 3D shutter glasses displays.

You can shop for monitors with blur reduction. TN will generally have less crosstalk, IPS a bit more crosstalk, and VA even more crosstalk. Study the Large Vertical Total tricks, those crosstalk-reduction tricks also helps 3D glasses too. And obviously, the little-known dynamic-range-reduction trick (yes, colors will be 10% - 20% duller, but 80% of the crosstalkv can disappear).

What Screen Tech Works Best?

Currently, it is in this following order:
1. Strobed TN LCD
2. Strobed IPS LCD
3. Strobed VA LCD
(NOTE: Strobed VA LCD is ghosty/crosstalky; only Samsung seems to produce OK quality -- make sure it's warmed up hot (warm VA has less crosstlak) and make sure to really raise the digital black levels quite a bit to help-out the overdrive overshoot-compensation range below your raised digital black)

Strobed OLED will be even better but they don't exist in simultaneous "120Hz-and-BFI" yet. You need a true 120Hz mode that also simultaneously has black frame insertion.

Also, if your/their 3D glasses site wants me to write a guest article -- I'd be happy to write one. Please contact me at mark[at]blurbusters.com ... I do work with monitor manufacturers from time to time, and have an scientific understanding of the limitations of shutter glasses too, and how to coax displays to work better with them.

...And please support Blur Busters by shopping monitors in the monitor lists below.
Thank you!
Head of Blur Busters - BlurBusters.com | TestUFO.com | Follow @BlurBusters on Twitter

       To support Blur Busters:
       • Official List of Best Gaming Monitors
       • List of G-SYNC Monitors
       • List of FreeSync Monitors
       • List of Ultrawide Monitors

rustyk
Posts: 6
Joined: 09 Nov 2018, 14:22

Re: Shutter Glasses - Active 3D Displays Like NVIDIA 3D Visi

Post by rustyk » 10 Nov 2018, 05:48

Hi Mark and thanks the fantastically detailed reply.

I've been a lurker here for some time and was aware of some of what you're written, but it's brilliant to have it all in one place :-)

One thing I didn't know about was:

"To reduce cross talk further, make your colors slightly duller (about 10-20%) -- reduce dynamic range slightly (LightBoost-style poor colors) by raising digital black level and reducing digital white level (via NVIDIA Control Panel or AMD Catalyst Control Center) or configuring the computer to use only HDMI range (16 to 235). LCD panel overdrive is problematic at the fullblacks and fullwhites so artifacting occurs with those. http://www.testufo.com/crosstalk is a very good test pattern for 2D blur reduction but is also 100% applicable to quality you can achieve with a mode."

This is a great tip and I'll experiment with this on my monitors. I have 3x BenQ XL2420T monitors, which although old do work really well and have very limited crosstalk. I was aware that tweaking brightness and contrast could help, but never thought along your lines before.

I'd totally forgotten all about the fact that the shutter glasses have transition times as well, that completely explains why my results were so poor on the OLED that I tried.. The lack of backlight (i.e. constantly lit pixels) meant that during shutter transitions I was seeing both frames. Would that be correct?
I did take some high speed videos of this using a P20 Pro, I'm happy to share them with you? When viewing them I couldn't really understand what was going on though and I wasn't sure if the phone was applying any digital processing.

I do still have a couple of questions if you'd be so kind.

1. How come DLP projectors work so well with 3d vision when they don't have black frame insertion/ULMB? Compared to any other display (active or passive), they have virtually 0 crosstalk. *edit* actually just thinking about this, do the DLP micromirrors inadvertantly introduce BFI as they tilt on/off all the time because of the colour wheel?

2. Given that nearly all new TVs are VA panels, do you think it's likely that I could get good results using active 3d? This was the main reason for posting. For example, looking at the details reviews on RTINGS which analyse the strobe effect in detail, is there anything that I can pick out as a good candidate or do I just need to try some displays and see how I get on?
I think I just need to find a display where the BFI coincides best with the shutter timings.

For example, the Samsung QE55Q6FN seems like a reasonable candidate:
https://i.rtings.com/images/reviews/tv/ ... -large.jpg

It's quite easy to 'fool' 3d vision to work with any non 3d 120hz input screen now. It will just pick it up as a 'Generic CRT'.

Thanks and regards,

Russell.

User avatar
Chief Blur Buster
Site Admin
Posts: 6509
Joined: 05 Dec 2013, 15:44

Re: Shutter Glasses - Active 3D Displays Like NVIDIA 3D Visi

Post by Chief Blur Buster » 11 Nov 2018, 00:44

rustyk wrote:One thing I didn't know about was:

"To reduce cross talk further, make your colors slightly duller (about 10-20%) -- reduce dynamic range slightly (LightBoost-style poor colors) by raising digital black level and reducing digital white level (via NVIDIA Control Panel or AMD Catalyst Control Center) or configuring the computer to use only HDMI range (16 to 235). LCD panel overdrive is problematic at the fullblacks and fullwhites so artifacting occurs with those. http://www.testufo.com/crosstalk is a very good test pattern for 2D blur reduction but is also 100% applicable to quality you can achieve with a mode."

This is a great tip and I'll experiment with this on my monitors. I have 3x BenQ XL2420T monitors, which although old do work really well and have very limited crosstalk. I was aware that tweaking brightness and contrast could help, but never thought along your lines before.
This is because overdrive require overshoot headroom above white, and undershoot headroom below black, in order to become more perfect (ghost-free / corona-free). Sometimes only one or the other headroom works (display dependent) or both is needed. The ghosts become much dimmer at http://www.testufo.com/crosstalk as you add those headroom,

rustyk wrote:I'd totally forgotten all about the fact that the shutter glasses have transition times as well, that completely explains why my results were so poor on the OLED that I tried.. The lack of backlight (i.e. constantly lit pixels) meant that during shutter transitions I was seeing both frames. Would that be correct?
I did take some high speed videos of this using a P20 Pro, I'm happy to share them with you? When viewing them I couldn't really understand what was going on though and I wasn't sure if the phone was applying any digital processing.
Again, look at both ends - shutter glasses transition is not the only problem.

Even if you make shutters instant, you will still also get some crosstalk on OLED due to the rolling scan and very short blanking intervals on current panels (0.5ms). You'd lose over 90 percent of light if you opened a shutter for only 0.5ms (the length of the blanking intercal during 120Hz VESA signal timing). Outside the VBI, a pixel somewhere on the screen is undergoing transition.

CRTs avoid most crosstalk because it's not sample-and-hold, and back then, 60Hz CRTs (30Hz 3D) had much longer blanking intervals, and crosstalk was only more severe at the top/bottom edges (much like the crosstalk gradient of blur reduction modes).

Sample and hold displays means a new refresh cycle "overwrites" the old refresh cycle. Even if it is CRT-style scanning refresh (top to bottom), even 0ms instant pixel response. So even though VBI size the same, and scan velocity the same, the difference between CRT (bright scanlines with lots of black above/below, see high speed videos of CRT) and sample-and-hold displays( (new refresh cycle is overwriting on top of continuously displayed previous refresh in a vertical wipe effect).... you get strange tearing-crosstalk artifacts (top half or bottom half is switched eyes) even with 0ms instant shutterglasses combined with 0ms instant GtG response screens, unless your shutter is open only during an ultrashort VBI.

Remember: During the refresh scan of a sample-hold display, the top edge is the new refresh surface area, bottom edge is the old refresh surface area, it'll manifest itself as a torn/tearing 3D effect for an instant-pixel response that eliminates the GtG fuzzyzone into a sharp tearline instead, if such a display existed.

To compensate for the scan effect of an OLED display, and get perfect 3D with a 120Hz display, you can use a mechanical spinning wheel with a slit. Thats how 3D movies were tested nearly 100 years ago.

Image

It would be easy to setup today with a custom Arduino project connected to a synchronous motor to spin a opaque disc (with a slit) in front of your eyes, in sync with a VSYNC tick-tock signal, with an analog knob for phase offset (to calibrate via watching a crosstalk pattern).

Theoretically, instead of LCD shutter glasses, one can use two spinning mechanical wheels with a slit or pie curour, designed to scan a slit downwards (like a camera rolling shutter) on alternate eyes, motors synchronized (via an Arduino).

This will eliminate the need for black frame insertion by forcing the eyes to see only the fully refreshed part of an OLED, while the rest of screen surface is refreshing. The rolling scan mechanical shutter glasses in sync with the rolling scan of display. You'd have to size the slit in a way that not the whole display can be seen at once ever (when wheel stationary) but scans the whole display. This allows hiding of the still-refreshing part of the display, as well as hiding the wrong-refresh-cycle portion of a sample-and-hold display. Viola. Perfect OLED without BFI, zero crosstalk!

But obviously would be massively unwieldly, noisy mechanical, though I am happy to assist a hobbyist with creating such a fun experimental project.
rustyk wrote:I do still have a couple of questions if you'd be so kind.

1. How come DLP projectors work so well with 3d vision when they don't have black frame insertion/ULMB? Compared to any other display (active or passive), they have virtually 0 crosstalk. *edit* actually just thinking about this, do the DLP micromirrors inadvertantly introduce BFI as they tilt on/off all the time because of the colour wheel?
But they ARE black frame insertition, not because of the mirror switching (they're already ~1440Hz 1-bit displays essentially; Hz varies but are in the high triple digit or low quadruple digit. The pixels flip many times per refresh cycle to generate colors temporally, turning 1bit into 8bit...). Generating colors temporally using ultrafast near-global-refresh. Unlike for OLED/LCD, all DLP/plasma needs internal scan conversion to the unique refreshing techniques of such temporal-per-pixel-color-generating displays)

That is because "3D Mode" settings on DLP projectors always activate intentional BFI (it has to be intentiona, cuz it is much longer duration than a mirror swtitch speed).

They just dont call it BFI but it is. But all mirrors are turned off for a few milliseconds. In fact, they've also been commandeered as blur reducing modes.
rustyk wrote:2. Given that nearly all new TVs are VA panels, do you think it's likely that I could get good results using active 3d? This was the main reason for posting. For example, looking at the details reviews on RTINGS which analyse the strobe effect in detail, is there anything that I can pick out as a good candidate or do I just need to try some displays and see how I get on?
Look for 120 Hertz BFI'd pursuit captures that are super razor sharp with zero or near zero crosstalk. That's your huge clue of good shutterglasses candidate.

Remember, it should be BFI-based blur reduction, not interpolation-based blur reduction
rustyk wrote: I think I just need to find a display where the BFI coincides best with the shutter timings.
You can also use the strobe crosstalk pursuit photography as the WYSIWYG reference point. Timing problems can theoretically be adjusted via a phase-offset adjustment if the glasses setup or emitter or driver-tweaking has one. Then I can trust I can calibrate the phase/delay offset to roughly match the amount of crosstalk I see in the RTINGS 120Hz BFI pursuit camera capture. (120Hz single strobe capable displays).
1. 120Hz
2. BFI
3. Single strobe, interpolation free
4. Look at the pursuit capture
5. That's the approx crosstalk you can get (at least for thise colors of cyan, red, and white)
Head of Blur Busters - BlurBusters.com | TestUFO.com | Follow @BlurBusters on Twitter

       To support Blur Busters:
       • Official List of Best Gaming Monitors
       • List of G-SYNC Monitors
       • List of FreeSync Monitors
       • List of Ultrawide Monitors

rustyk
Posts: 6
Joined: 09 Nov 2018, 14:22

Re: Shutter Glasses - Active 3D Displays Like NVIDIA 3D Visi

Post by rustyk » 11 Nov 2018, 06:33

That's fantastic and I've learnt a lot, thanks so much :-) Your level of knowledge is vast!

I'm spending some time looking at the pursuit test results for a few displays. It's interesting when you compare the pixel response times and the static screen shots of the test. For one, you can see that for pixel response times only, TN>IPS>VA, but you can also see that at 80% the VA based Samsung Q7FN pixel response time is equal to a IPS gaming monitor, the ASUS PG279Q. That to me illustrates why reducing the brightness can help with crosstalk, as it will speed up the pixel transition times significantly.
If you look at the TN based Dell S2716DG by comparison, you can see that the response times are at least twice as good.

However, as you say BFI is also key and the Samsung does show good results in the pursuit test, so it does seem like a fairly good candidate with a few tweaks to brightness etc.

I'll refer people back to this site to read this and if I ever find a good display I'll let you know.

User avatar
Chief Blur Buster
Site Admin
Posts: 6509
Joined: 05 Dec 2013, 15:44

Re: Shutter Glasses - Active 3D Displays Like NVIDIA 3D Visi

Post by Chief Blur Buster » 16 Nov 2018, 21:48

You are very welcome!

Remember to do the dynamic-range-reduction trick digitally (raising black levels and decreasing white levels) -- not via backlight. You want actual LCD pixel black to be above the black level used for the blackest black. And the actual LCD pixel white to be below the white level used for the whitest black. This is to avoid most of the overdrive-range-related crosstalk artifacts.
Head of Blur Busters - BlurBusters.com | TestUFO.com | Follow @BlurBusters on Twitter

       To support Blur Busters:
       • Official List of Best Gaming Monitors
       • List of G-SYNC Monitors
       • List of FreeSync Monitors
       • List of Ultrawide Monitors

rustyk
Posts: 6
Joined: 09 Nov 2018, 14:22

Re: Shutter Glasses - Active 3D Displays Like NVIDIA 3D Visi

Post by rustyk » 19 Jan 2019, 06:46

Hi there,

Just before Christmas I had to scratch the itch and purchased a BenQ TK800 4k projector. It supports 120hz input at 1920x1080 and I've verified this using your UFO test and other tools such as https://www.displayhz.com/.

I've been able to activate 3d vision mode but I'm having a desynchronisation issue between the shutter glasses and the display.
When you activate 3d mode, you'll see the 3d image displayed almost perfectly but there will be a 'ghosting' effect in one the colours, usually red or blue. If you wait and watch, you can see the colours red, blue, yellow, green etc. come and go as a double image, before you finally see double everything and the image inverts before popping back to a perfect 3d image and the cycle then repeats.

It appears to me that the 3d emitter and the display are out of sync by a small margin, so that error is magnified over and over until they naturally get back in sync. I've tried using different display modes in the 3d driver and I've spent literally hours playing with ToastyX's CRU tool to try different refresh rates but I just can't seem to get them in sync.

Please do you have any ideas or insight? Interestingly, other users on the 3d vision forms have tried other DLP models such as the Optoma UHD51 and they are having exactly the same issue.
https://forums.geforce.com/default/topi ... 4/#5962264

The only projector proven working so far is the Optoma UHD40/50, which natively supports 3d frame sequential input at 1920x1080x120Hz, so it syncs using a DLP flash and is not reliant on the IR signal from the Nvidia emitter.

If I could solve this problem it would open up a huge number of options to us 3d vision users.

User avatar
Chief Blur Buster
Site Admin
Posts: 6509
Joined: 05 Dec 2013, 15:44

Re: Shutter Glasses - Active 3D Displays Like NVIDIA 3D Visi

Post by Chief Blur Buster » 20 Jan 2019, 00:46

Obligatory TestUFO mention: www.testufo.com/refreshrate
Head of Blur Busters - BlurBusters.com | TestUFO.com | Follow @BlurBusters on Twitter

       To support Blur Busters:
       • Official List of Best Gaming Monitors
       • List of G-SYNC Monitors
       • List of FreeSync Monitors
       • List of Ultrawide Monitors

rustyk
Posts: 6
Joined: 09 Nov 2018, 14:22

Re: Shutter Glasses - Active 3D Displays Like NVIDIA 3D Visi

Post by rustyk » 20 Jan 2019, 07:12

I believe displayhz gave exactly the same frequency as your UFO refresh rate test, although obviously your UFO tests are vastly superior :-)

I'm not quite clear on what you're proposing though? Do I need to create a custom resolution in CRU and make sure it matches exactly what the UFO test says?

The BenQ manual is quite specific in terms of its timing config for this mode:
Resolution Mode= 1920x1080@120Hz : Refresh rate (Hz)= 120.000 : H-frequency (kHz)= 135.000 : Clock (MHz)=297.

I tried this and I'm sure I already tried checking the exact refresh rate and setting it to that, i.e. 119.984 for example.

User avatar
Chief Blur Buster
Site Admin
Posts: 6509
Joined: 05 Dec 2013, 15:44

Re: Shutter Glasses - Active 3D Displays Like NVIDIA 3D Visi

Post by Chief Blur Buster » 20 Jan 2019, 14:40

rustyk wrote:I believe displayhz gave exactly the same frequency as your UFO refresh rate test, although obviously your UFO tests are vastly superior :-)
Yes, displayhz and TestUFO uses the same algorithm.

However, as inventor of TestUFO, I obviously will advertise TestUFO from time to time :D
rustyk wrote:I'm not quite clear on what you're proposing though? Do I need to create a custom resolution in CRU and make sure it matches exactly what the UFO test says?
No.

Don't worry about fractional refresh rates. It's normal because it's a clock accuracy difference between CPU clock and GPU clock.

Changing the refresh rate will change TestUFO result. There'll always be a fractional of a percent (e.g. 0.1%) difference between what CRU says and what TestUFO says, but they will move lockstep in each other. So if you change CRU to 115Hz, TestUFO might say 114.985Hz or 115.003Hz or 115.123Hz etc.

Also, most BenQ monitors are multisync in refresh rate, supporting approximately 48Hz through 144Hz. Even with the spec, there is often an error tolerance of a few percent.

(History Lesson precedent: That's also incidentally why they standardized on NTSC 59.94Hz to add color signal to the original NTSC 60.00 monochrome signal in the 1950's back in our grandparent's era -- they had to change refresh rate 1000/1001ths to avoid a stationary beaded chroma artifact -- and they knew the monochrome TVs were forgiving enough to accept 59.94Hz even though the TVs were specified at 60.00Hz).

3D shutter glasses are designed to adapt to minor differences, that's what the emitter is designed for -- synchronization of refresh cycles. However, since the LCD shutter glasses have an LCD GtG, it's best to put the monitor in strobed mode (LightBoost mode, ULMB mode, or other strobe backlight mode) so the monitor is OFF while the shutters are switching eyes.

If you're trying to hack 3D shutter glasses on a 120Hz LCD display with no strobe backlight, you will have much more ghosting -- it is just not possible to fix crosstalk on those nearly as well as with a strobed display. I've heard of people hacking 3D glasses onto a strobe-backlight display even if there's no 3D support (e.g. no 3D Vision) through third party glasses, third party emitter, and custom tweaks.

Your problem is NEVER the fraction digits. Your problem is a sync problem, not a fractions problem.
rustyk wrote:If you wait and watch, you can see the colours red, blue, yellow, green etc. come and go as a double image, before you finally see double everything and the image inverts before popping back to a perfect 3d image and the cycle then repeats.

It appears to me that the 3d emitter and the display are out of sync by a small margin, so that error is magnified over and over until they naturally get back in sync. I've tried using different display modes in the 3d driver and I've spent literally hours playing with ToastyX's CRU tool to try different refresh rates but I just can't seem to get them in sync.
Looks like an emitter tracking error. (Emitter phase/tracking adjustments needed).

That's a problem at the emitter side, something is not causing the emitter to communicate the correct blanking intervals between the computer and the emitter.

If you are 100% sure that the emitter is in sync with the blanking intervals, then there is something wrong with that chain.

Also, be careful about using screen mirroring for emitters, since the mirrored Hz may not be in perfect sync with the original Hz. Be careful about emitter connected to home theater receivers, because the VBI may slew out of sync if the receiver is framebuffering the refresh cycles (and accidentally doing a very tiny scan-conversion -- e.g. different Hz by 0.001Hz). This does not always happen, but this will wreck 3D sync.

A longshot is the DLP Projector is not synchronizing its colorwheel to the refresh rate, but I know DLP projectors typically does synchronize the wheel.

Do you have access to a 960fps smartphone high speed camera such as a Samsung Galaxy S9 or a Sony Xperia XZ/XZ2 Premium? We've begun using these for fantastic display analysis
Head of Blur Busters - BlurBusters.com | TestUFO.com | Follow @BlurBusters on Twitter

       To support Blur Busters:
       • Official List of Best Gaming Monitors
       • List of G-SYNC Monitors
       • List of FreeSync Monitors
       • List of Ultrawide Monitors

Post Reply