Hi everybody,

I've joined this forum recently to find out about possibilities for S3D on my Samsung 2233RZ, without using an Nvidia graphics card. I soon stumbled across this thread:link.

RAGEdemon had a great idea that I thought might also work for my setup with some modifications: namely using some sort of frame marking (f.e. blue line code) to drive a light sensor and generate a sync signal from that. He listed a few advantages of this approach. They include:

- Compatibility with any screen (except maybe slow LCD's), DLP projector (even non-3D certified), signal type, graphics card (as long as there is a possibility of showing marked frames - IZ3D can do this)
- Price (electronics altoghether are about $15, then you just need the IZ3D driver for $40 and some wired shutterglasses (there are lots of 4 wired shutterglasses for $50 total, and there are still lots of old ones available second hand - such as my Asus VR100 model). You could even fashion your own IR transmitter with a bit of effort. No need to switch GPU's or buy a different screen/projector.
- No eyeswitch - ever. Even in the case where the input lag of the projector varies all the time, the frames would still go to the correct eye because the frames are marked.

I am pleased to say that after some work, I've managed to get flawless 3D working on my Samsung 2233RZ with this method, using not the exact components that RAGEdemon suggested but with a phototransistor in the corner of the screen.

The circuitry for implementing this system on a DLP projector would be very simple, since there would probably be no need to decrease ghosting through electronics. I could make a schematic if anyone wants it, though I haven't tested it on my projector yet. The following should work though: use the Schmitt-triggered detector that RAGEdemon mentioned in his thread, use a hex inverter to buffer the signal and invert it for the other eye, amplify this to 12V amplitude and you've got your shutterglass signal. A simple switch can be used to switch the eyes around. Again, if anyone requests this I would be happy to make a schematic drawing.

Getting it to work on my LCD was a bit more complicated though. The detector used by RAGEdemon was not sensitive enough. I used a phototransistor and comparator to achieve the square wave I was looking for, but even then its duty cycle kept varying slowly. I eventually ended up hooking up an AVR microcontroller to the input signal, which triggered on one edge of the signal and then opened the shutterglasses at adjustable delays for each eye. I must say, the results are impressive! For anyone who has a 120Hz LCD and doesn't feel like buying 3D vision, the advantages are:

- Good chance this solution will remain compatible with whatever the future brings (only dependency is on the marked shutter output of IZ3D)
- Much cheaper than 3D Vision (think $70 for driver+glasses+electronics, then $15 for each additional set of glasses)
- More freedom to adjust the 3D effect (the microcontroller in my case gives the opportunity to adjust the exact timing, opening time of shutters etc, per eye individually, eye switch). I noticed that this way I can even sacrifice some brightness to enable games like Mirror's Edge to work without ghosting (3D Vision has a lot of trouble with that game from what I read)
- General coolness of using something you made yourself in a weekend's work!

I already posted some early results in the aforementioned thread, but the interest in this subject seems to have died down since RAGEdemon came up with it. That is why I started this thread. I will show my own progress, but if anyone wants to create this system for themselves I would be happy to go into detail, share source code, and even make a tutorial. It's just that I only want to do that if someone is actually going to use it. Also, it would require basic knowledge and experience with electronics.

One last thing: the marked shutter mode was removed from the IZ3D driver after version 1.10, so for now I am stuck with that version. However, I got a heads up on IZ3D forums that it is likely to return in version 1.14. This means that for now, there is no media player compatible with the marked shutter mode, and that the framerate in-game has to be high enough or the IZ3D driver will slow the shutters down to 60Hz. Both of these problems are solved in the latest driver though, so as soon as the marked shutter mode returns it'll be A-OK.

So, is there any interest?

Very interesting work. Nice job. Any chance you can take a picture of the device you created, or even a shot of a game through the shutter glasses?

Sure, I'll post some pics and a more detailed detailed description when I get home tonight. It's only been tested on my 2233RZ for now, but I have a DLP beamer waiting to be tried as well.

BTW, I still have some ghosting in the top of the screen. I didn't notice at first because I only tested the system on Flatout 2, where all the action is in the bottom and middle... I found out it is because the 2233RZ has a special display mode for 3D Vision that I cannot access (yet) from my AMD card - see this thread. Any info on how I can find the correct timings would be appreciated! Of course, this shouldn't be a problem for DLP projectors or CRTs...

-no theres no hidden mode, I don't know where this rumor originates from. In fact the 3D vision forum has 100 page long thread about top down ghosting. They have only 1.5ms of light every frame but there s still ghosting creeping in.

Basically that display has that ghosting because the pixels are updated line by line instead of some parallel scheme and it has some f-ugly persistence issues ( just like CRT's green phoshpor, thats rather benign) ,again, theres no hidden parallel mode.
(btw, can you imagine how good CRT would be with parallel electron guns?)
IDK which is worse example of rebadging, the 2233RZ kinda stuff , or LG 's stuff. Both so bad, they kinda cancel eachother out.

I like SLI with 3d vision, no plans for iz3d now, maybe with 28nanometer graphix.

I read this thread as well, and I'm not saying I can make this monitor completely ghost-free. However, that doesn't mean that there isn't any special display mode for 3D. Several things point to the conclusion that there probably is:

- User manual of the monitor lists a max horizontal refresh of 190kHz, which is not used in any standard mode AFAIK.
- When in 3D mode, any 3D Vision owner can confirm that the panel reports a 185kHz horizontal refresh, while in normal 120Hz mode it only reports 129kHz.
- In some thread about getting 3D Vision to work on Linux, someone was able to let Powerstrip analyze the output during 3D mode - Powerstrip also reported 185kHz refresh (although some of the other results were kind of strange).

It is at the very least obvious that something changes when 3D Vision is active. The higher horizontal refresh suggests that there is more "stable time" for the shutter glasses to sync with the screen.

Also, in that same thread I saw many people who resolved their ghosting issues by doing something about their USB latency to the NVidia emitter.

I am pretty hopeful that accessing this mode would at least improve ghosting.

As promised, here are some pics, a video of the action and a more detailed description of the device!


By narcoticv at 2011-07-06: This picture shows what the device looks like (beautiful isn't it? ). On the left the cable leading to the sensor is plugged in, on top the shutterglasses (Asus VR100 wired shutterglasses, but any 3.5mm wired ones should work). From the right a 12V power supply is connected.


By narcoticv at 2011-07-06: This image shows the box up close. Pardon the crappy styling ... the left side has a 3.5mm jack for sensor input, right side a DC chassis for 12V input, at the top (though not visible here) there are 4 connectors for shutterglasses (all recieving the same signal). At the bottom of the image, you see a power switch, an eye switch, a slider switch with 4 positions that selects the parameter to be adjusted (left eye delay, right eye delay, shutter opening time and something for debugging that is not worth mentioning here). The two pushbuttons can then be used to increase/decrease that particular parameter. So what is the job of this box? It recieves the signal from the phototransistor, and syncs only on the leading edge of this signal (meaning it syncs only once in two frames - so it syncs everytime both eyes have received a frame). At first it synced on both edges, but since the duty cycle of the phototransistor signal varied all the time (see my posts in this thread), it was better to let it sync on one frame only and then have a fixed delay for the other eye. Relative to this sync, the delays for each eye and the shutter opening time can be adjusted. Pressing both buttons at the same time saves the current settings to EEPROM memory, so it remembers upon next startup what your last settings were.


By narcoticv at 2011-07-06: This shows how the sensor is mounted on the monitor. Initially, it was just a phototransistor with a cable attached to it, leaving all electronics in the small black box. However, it seemed the signal was to weak to go over the cable without interference. Although covered in tape, you can see it became a bit bulkier: it is now a phototransistor that sends current through a resistor (generating voltage signals whose shape can be seen in my posts in this thread), which is then lead directly into an operational amplifier (LM741) and compared to a reference voltage which is configurable by a potentiometer. Basically, the voltage signal is compared to this reference voltage and thereby turned into either High or Low, meaning the output becomes a square wave which goes to the little box. The duty cycle of this square wave is varying all the time in the case of my 2233RZ (again, see my post in the aforementioned thread for an explanation why), but sync with the monitor is stable.



By narcoticv at 2011-07-06: These images show the inside of the box. If anyone iis interested I can draw up full-blown schematics, but for now I will just give an overview of the circuitry. The long black chip in the lower left corner is an operational amplifier. Basically it does the same thing that the circuitry embedded in the sensor does: compare the input to a fixed reference determined by the potentiometer next to it, and generate a square wave as a result. It's just a precaution: if the signal from the sensor wasn't a square wave yet, it is now.
To the left there is a bunch of wires going to all the connectors, buttons and switches in the top half of the box. The "three-legged" chip in the middle is a 7805 voltage regulator that converts the 12V into 5V supply for the logic elements in the circuit. It has some required capacitors around it.
The longest chip is an AtTiny2313 microcontroller. This has my software on it, and controls every aspect of the shutter timing based on the signal it receives from the operational amplifier. It outputs the signals for each eye of the shutterglasses. Next to it is a connector for uploading code into the device.
Finally, there is a pair of transistors (BC550) and a pair of resistors near the wire connector on the left. These two transistors amplify the output signal of the microcontroller, which is still 5V range, to 12V. They also have the power required to drive the shutter glasses (though I have not tested how much power is actually needed for this, they seem to do well).

...But of course, all you really want to see is the result! Since taking crappy pictures with my phone only resulted in photos that were smeared all over the place, I chose making a small video instead:


In the video you can see that while there is no noticable ghosting in the middle of the screen, the bottom half of the car seems a tad ghosted. I found that no matter how short I make the shutter opening, or how I adjust the timings, there is never a combination without any ghosting anywhere. The best I can do is move the ghosted part around the screen (top or bottom). I am still hoping that activating the "hidden display mode" of the monitor (which, to be fair, I am not certain exists) will solve this issue. Anyway, I can be sure that the ghosting is not "my fault" but the monitor's: after all, I can tweak the glasses any way I want and the sync is stable as hell. This means that it shouldn't be a problem when using this system for a projector, for example.
Also, the image is considerably darker with the glasses on - this is of course due to the shuttering. It can be adjusted to be brighter, but some brightness will always be lost since the eyes share brightness coming from the monitor.

I wish I could compare my results so far with those of 3D Vision users! I have never had the chance to use such a system yet. I am pretty sure though, that this system should provide equal or better 3D - at least, if the monitor behaves identically in both cases! I still haven't tried using 100hz or 110hz instead of 120hz, which should also go a long way for reducing ghosting.

As soon as I get bored playing around on my monitor, I will take the time to build a less sensitive sensor for projector use, and adapt the software to different frequencies. Then I can post some results with my 85Hz DLP projector!

Nice job.

Nice work NarcoticV
2 questions about this circuit:
1. is the output to the glasses the same sync signal as used by old 3D TVs?
2. can this circuit be driven by DLP flash instead of Frame marking?

Basically I have a 120hz DLP projector and I want some way to get a standard Sync signal (like on old 3D TVs) out of it.

Hi mAchiNE

ed dongle and glasses will work with my optima hd66 and acer h5360 so should work for you too. Just need a vga output but you can clone the image and use hdmi too. Use the vga to hook up the dongle to the projector and use hdmi\dvi to view the image. This was a little darker then using dlp-link glasses but not to bad. No ghosting. You need the ed activator software to turn on the glasses.

Can someone post a tutorial showing how to make this narcoticv device,please?

I'm pretty certain there IS a hidden display mode, at least on my AW2310. In fact, I managed to trace my occasional ghosting problem to it!

Turns out that when I ever get my ghosting problem (haven't had it for ages) it was because my monitor had lost the 3d signal briefly and turned off this mode. This causes marked ghosting especially on the top and bottom of the screen. You can tell when this 3d mode is on, because in the monitor settings it locks brightness and contrast down.

The difference between running in this mode and not is massive. I've now learn to fix it by disabling 3D vision, turning my monitor off, and disconnecting the dongle. After reversing this process, the problem is resolved. I suspect that it indeed changes the timings for some reason as well, which would account for the fact that it has such increased ghosting.

@ NarcoticV: thats really awesome work! The only problem I can see is that I hear IZ3D is no longer going to be supported. Maybe someone can persuade Tridef to add this functionality to their drivers...

Hi WiredEarp.I have a samsung 3d 2233rz and X3D wired glasses and dongle.Would It be possible to make the X3D glasses and dongle work with my monitor or my 3DTV?


Windows 7 x64
Intel core2duo E8400 3,6hz
Geforce GTX 460 1gb(I also have a gt 720 256mb)
4 gb ram Kingstom
monitor samsung 3d 2233rz
2011 samsung 3DTV pn51D49