Need to test dual projector theory... help please.

[GPLavoie]

I use 3D Vision as quad buffered openGL for ArcGIS. I have smashed my head against the wall trying to figure this out so here goes, and please excuse my ignorance level as I am only a lowly photo interpreter.

I need to record my 3D vision desktop to demonstrate how to measure tree heights to a class of college students. It's not going to work with fraps, or taksi, or virtualdub, or any of the other ones I have tried. I have contacted a few people and have gotten nowhere.

I had given up and decided that I would just do all of the the height measurements and z adjustments "live off the floor". No big deal really, but I had a budget constraint that forced me down the passive eyewear road. Dual projection seemed tricky, so I looked for a single projector passive stereo solution. The polarization modulator appeared out of nowhere. I got a quote for one, and well, lets just say I could outfit the class with active eyewear and take less of a hit. So, DLP projection with polarized filters and a whack of matching glasses for the class are my budget friendly option. Now, this is where it starts to get a bit abstract for me, see below:

It came to me out of the blue, and I can't find any documentation on it. If I use the "RGB out" from each projector to a "passthrough" on dual camcorders (one for each eye) to firewire connection back on a separate pc, I should get a separate video file for each eye. Theoretically, could I then replay the files in the 3D vision player or stereoviewer? Would I have to "mux" them? Is there anyone here with the equipment to test this out?

rgrds
G.

[Fredz]

Just curious, where did you find a polarisation modulator and at what price ? For your questions, unfortunately I didn't really understand what you're trying to do...

[GPLavoie]

http://www.depthq.com/modulator.html" onclick="window.open(this.href);return false;
I think they should be the ones to quote you on it. Great gadget, but I think it is too rich for my needs at this time. I think I'll get more brightness out a dual projection passive setup anyways. Theres pros and cons to all of the stereoscopic options.

I haven' seen too many projectors with an HDMI out port, so the RGB out on a lower end projector will do just fine if I utilize the "passthrough" connection of a regular camcorder to link out via firewire to a PC. In a dual projector setup, not only do I have passive stereo up on screen, I also have an analogue signal for each eye through the RGB out port of each projector. I convert this back to digital saved as an avi file on a computer and I'm recording quad buffered stereo. Does this make sense?

G.

[GPLavoie]

Heheh.
What if I were to just use a screen capture software to capture the left and right file separately? ... silly me.

Any comments?

[skrubol]

Most video capture devices only work with composite and S-video. Some work with component and HDMI. Very few work with RGB(HV). There are RGB to component converters out there, but you'll probably need to use a TV resolution and a capture device that supports that res. Other challenges are using 2 capture devices on one PC, and starting/keeping the captures in sync.

The software option is always better if you can make it work. Staying in sync shouldn't be a problem if you can start them in sync (also, as long as there's no drift, it's not too hard to sync them up later if you put in something that's fast enough to sync to.)

[GPLavoie]

What about recording a quad buffered openGL stereo window with software? I haven't found anything out there that will/can do it. I am wondering if that is because it is impossible, or if the process would be too taxing on cpu resources?

I want to be able to record a left and right channel separately on screen.

Anybody know haw to attack this problem?

G.

[Fredz]

If you are a programmer or know one, you can write an OpenGL interceptor or use this library :
http://glintercept.nutty.org/" onclick="window.open(this.href);return false;

[GPLavoie]

I'm a college prof... I want to create a set of quad buffered open GL 3D training videos to help with an applied imagery interpretation class I teach. It will serve as a great way to expose the students to natural resource management/GIS operations, using a stereoscopic projector (and hopefully a class set of the monster vision rf shutter glasses). I could see this having many other applications as well.

There is an applied research program funded through the provincvial government, whereas I could acquire funding to facilitate a project for programming students to put it together... Only, I am not sure as to the complexity of the task.

Would anybody be willing to serve as a bouncing board, if I were to put a project together? I am not a programmer, but I am a photogrammety guy with experience using many different variations of stereoscopic equipment. The end goal of the project is to enable quad buffered desktop recording, the resulting output being a quad buffered stereoscopic video to work in Peter Wimmer's stereoscopic player.

How many hours would a group of programmers need for a project like this? Could it be accomplished in 6 months?

G.

[Fredz]

I'd say a competent programmer could do this in less than one week.

I've already done OpenGL interception in the past in an attempt at creating a stereo 3D driver for Linux and the code for interception was quite simple to write. It's just a matter of intercepting the glXSwapBuffers call, inserting code to alternatively choosing the left and right buffer and then call glReadPixels to save the two images on disk.

Then you'll have to create a video from the images, that can be done with FFMpeg or MEncoder quite easily.

You can get some code to get someone started on this project, this one does what I explained above but for one buffer, it shouldn't be very hard to adapt it to handle two buffers :
http://blog.smr.co.in/linux/graps-captu ... gl-frames/" onclick="window.open(this.href);return false;

[GPLavoie]

Thanks,

I will keep in touch on this. I will talk to the IT department head to see if there is anybody with the skills to do this in the program, or if there will be additional training needed. I am a bit worried that the cpu programming courses don't expose the students to this stuff, but then again, what do I know

Anyways, I'll be in touch.

best regards,

G.