Stroboscopic Glasses

[daniloberserk]

So, I was thinking. Wouldn't be possible to just create "stroboscopic glasses" that flickers at an set frequency that hides the vision for both eyes be an "better" solution for sample and hold motion blur on displays instead of backlight strobing?

I mean, I've never used those Nvidia 3D glasses back then, but I'm pretty sure they work like that (but instead of closing the lid for both eyes, it just flickers between then).

It may sound weird, but wouldn't this work for basically anything? Even VRR wouldn't be a problem, although I'm not sure how our brains would "interpret" the scene.

I did a quick search and it actually exists some stroboscopic glasses for sports training, although I have no idea of how "versatile" it might be when trying to match configurations like, 360 Hz flickering frequency with 1ms of exposition or something like that.

It might be an stupid idea, but I would actually like to try something like that.

[Chief Blur Buster]

So, I was thinking. Wouldn't be possible to just create "stroboscopic glasses" that flickers at an set frequency that hides the vision for both eyes be an "better" solution for sample and hold motion blur on displays instead of backlight strobing?

It definitely a doable concept and there are already multiple threads about this.

viewtopic.php?t=5837
viewtopic.php?t=8509

The second link is the most useful.

The problem is strobe crosstalk is worse than if the LCD and backlight co-operates more perfectly, including the Quick Frame Transport tricks documented at www.blurbusters.com/xg2431

However, it definitely reduces motion blur of displays that does not support motion blur reduction at all.

Also it was discovered that mechanical strobing was superior (less crosstalk) to electronic strobing (shutter), because a rolling shutter slit can slide downwards in sync with the LCD scanout since not all LCD pixels refresh at the same time (high speed videos www.blurbusters.com/scanout ...)

A spinning cardboard disc (connected to Arduino) mounted in front of the monitor, and staring at the monitor through it, works too. Or putting the mechanical shutter in front of an LCD/LCoS projector -- which works great too (better than DLP strobing, in fact!) once the cardboard is cut and slitted properly and the phase is adjusted correctly (need to use a VSYNC monitor, but there are many ways to do that). To avoid spinning the disc fast (60 spins per second) multiple slits in the mechanical rolling shutter is recommended.

If you build a mechanical strobing system, please post here!

[daniloberserk]

Oh! Just missed those topics! Thanks! I'm glad to know that such thing might be possible/feasible. Maybe as technology evolves it might have some uses.

Although I'll pretty much prefer the 1000Hz journey with FRAT =)

[Chief Blur Buster]

Oh! Just missed those topics! Thanks! I'm glad to know that such thing might be possible/feasible. Maybe as technology evolves it might have some uses.

It's already feasible.

You just have to tune it carefully, and mechanical shutters sometimes produce vastly superior results to the backlight strobing. Especially in monitors with fully adjustable overdrive-gain (100-levels) and you have a phase adjustment (VSYNC time offset) in an Arduino-based synchronous spinning motor, perhaps with ad8e's open source VSYNC-listening algorithm.

But it is a lot of simultaneous skills if you want to build a mechanical strobing rig:

- Understanding rasters (Tearline Jedi)
- Understanding LCD scanout (High Speed Videos of LCD Refresh)
- Understanding Electronic Circuits (wiring a synchronous motor to an Arduino that's listening to VSYNC/VBI heartbeats over USB to PC, with USB--jitter-filtering)
- Understanding Arduino programming (Arduino program that spins the wheel precisely in sync with a dejittered/filtered VSYNC/VBI heartbeat, with a phase-offset adjustment, to allow you to carefully strobe out-of-phase with LCD GtG)
- Understanding Optical Physics (correct slit size, larger disc with more slits to allow slower spin speeds, correct distance of spinning cardboard disc away from screen, correct head position if using a screen instead of a projector)

But if you have all these skills concurrently, please... build it and post your results!

The most successful mechanical strobing is an Arduino-driven spinning rolling shutter in front of a LCD/LCoS projector lens. Then head position does not matter. But you will want to use very thin slit sizes & a large wheel at least several inches away from the projector lens, to create the sharpest shadow edges in the rolling shutter, and the rolling shutter should scan downwards in sync with the LCD/LCoS projector scanout behavior.

A high speed camera (At least 8x your strobe Hz) such as the 960fps feature of a Samsung Galaxy, will make development easier of a mechanical strobing rig.

[elexor]

phpBB [video]

took apart an old set of 3d glasses and connected directly to a 5v arduino direct driving the lcd glass. qft60hz on my lcd works better then I expected. I might end up using this with the dell ultrawide qdoled when I get one. since the monitor has zero blur reduction features global shutter shouldn't be as much of an issue with qft modes + oled g2g.

I want to try mechanical strobing too but this is much easier to setup and more practical to use.

[Chief Blur Buster]

took apart an old set of 3d glasses and connected directly to a 5v arduino direct driving the lcd glass. qft60hz on my lcd works better then I expected. I might end up using this with the dell ultrawide qdoled when I get one. since the monitor has zero blur reduction features global shutter shouldn't be as much of an issue with qft modes + oled g2g.

I want to try mechanical strobing too but this is much easier to setup and more practical to use.

Excellent to see!

As long as the LCD GtG of the shutter is fast enough, and the LCD GtG of the panel is fast enough, your blur-reducing shutter glasses can do some good motion blur reduction duty on almost any non-strobed LCD.

If you decide to try a mechanical spinning wheel, I recommend the table-edge-mounted version, with perhaps 10, 12 or 15 slits, so you only need to spin the wheel 6, 5, or 4 times a second for 60 Hz. The mechanical spinning wheel can also make QFT less necessary by "sweeping" a tiny slit out of phase of the LCD GtG fade zone. it could be very handy for LCD or LCoS projectors as well.

The bonus is a mechanical spinning wheel works wonderfully with LCD projectors (Even cheap $100 chinese LCD projectors). Just make sure the cardboard wheel is designed big enough and calibrate lens-to-shutterwheel distance, if you're going to do mechanical "rolling scan" of the clearly refreshed pixels between the LCD GtG fadezones (not all pixels refresh at same time -- LCD projectors do the same as www.blurbusters.com/scanout ...).

The lack of overdrive in many chinese LCD projectors does add a smidge of crosstalk. On the other hand, the high heat of chinese LCD projectors does accidentally speed up the GtG (run it for a while to make it hot and toasty for minimum crosstalk). But adding software-based overdrive actually creates major reductions to LCD crosstalk.

Incidentally, this is one of the reasons I want to see the existence of an open-source virtual windows indirect display driver with software-based overdrive (superior clone of ATI Radeon overdrive). If you know windows indirect display driver programming, send a PM to say hi!

If you have a ~1000fps high speed camera mode (Samsung Galaxy), this is very handy in debugging a homebuilt external strobing accessory (e.g. whether shutter wheel or modified shuttter glasses)

[elexor]

Love your ideas really want to see more people experimenting with this stuff! especially the virtual display driver stuff, software bfi sorely needs this. Speaking of cheap chinese projectors do you know of any that can display 60hz without much processing delay?

[kokoko3k]

Why both eyes at the same time?
Why not just alternating?
Blur would be gone because each eye would be unable to track the moving object, but the perceived brightness woukd stay the same (in the common R/L eyefield) because that's what happens if you close one eye.

Something I'm overlooking maybe? It seems so simple to me.

[Chief Blur Buster]

Why both eyes at the same time?
Why not just alternating?
Blur would be gone because each eye would be unable to track the moving object, but the perceived brightness woukd stay the same (in the common R/L eyefield) because that's what happens if you close one eye.

Something I'm overlooking maybe? It seems so simple to me.

1. It doubles frame rate requirement. If you don't double frame rate, you get a double image effect.
3. It makes stutters worse.

You must be gametime:photontime correct the moment the eye sees the frame, regardless of whether 1 or 2 eyes see a game's temporal event (a frame). So if you alternate, you need different unique time-correct frames, and you now have 2x render rate needed.

By alternating eyes (for no reason, because you're not doing 3D), you've doubled the number of frame-flash events and required frame-render events. That's more inefficient on the game engine side per motion blur reduction benefit. It's more efficient to just flash the same frame simultaneously to both eyes at a lower frequency, and still get the SAME motion clarity benefit.

It's much like how 120fps 120Hz with 1ms PWM strobe has less motion blur than 240fps 240Hz with 2ms PWM strobe.

Also, another factor is if you're strobing 120fps 240Hz, you get double images like CRT 30fps at 60Hz, which I can reproduce via software BFI at this animation just like this TestUFO BFI Double Image Effect Caused by Half Frame Rate Effects.

And if stutters occurs in between the eye swap, you have some nasty stutter artifacts worse than a plain stutter. This is because one eye's stutter is out of sync with the other eye's stutter. It's like the real world stuttered only to one eye, then temporally later, a very different-looking stutter happens to other eye, out-of-sync. This can amplify motionsickness if even a single framedrop stutter look much more unnatural because the stutter did not look the same to both eyes simultaneously.

[kokoko3k]

Thanks for answering, it is all theory, so I still haven't fully understood.
But just to be clear, you say that if the left eye sees even frames and right eye odd ones, you'd have the impression of double frames?