stationary projector onto visor

This is for discussion and development of non-commercial open source VR/AR projects (e.g. Kickstarter applicable, etc). Contact MTBS admins at customerservice@mtbs3d.com if you are unsure if your efforts qualify.
Post Reply
rbairos
One Eyed Hopeful
Posts: 9
Joined: Fri Oct 30, 2015 7:23 am

stationary projector onto visor

Post by rbairos »

Has anyone ever researched projecting from a stationary projector, directly onto a visor?

Advantages:
-remove the need to physically carry the active electronics on your face, allowing for much lighter, potentially wider visor. (google cardboard without the phone)
-potentially cheaper, (high resolution stationary DLP projectors vs lightweight high resolution oled display)

Disadvantage:
-Inefficient use of resolution, as the usable projector image would only be the subset of pixels that land on the visor rear-screen at various angles/positions.
-Brightness loss from rear-projection at a distance.
-Would have to stay in focus over the defined range of the user's head.

Just curious,
Rob.
User avatar
cadcoke5
Binocular Vision CONFIRMED!
Posts: 210
Joined: Mon May 24, 2010 8:43 pm
Location: near Lancaster, PA USA

Re: stationary projector onto visor

Post by cadcoke5 »

Another advantage, is that you can have some powerful computers generating the images, since they dont have to be on the user.

It may be possible to make the resolution better by zooming the projector, and tracking the user's head. I have seen a system that projected a computer generated face onto an actor's own face. However, the demo I saw certainly did not keep the projected image aligned with the actor's face good enough. I would think that for VR, the alignment would have to be much, much better.

-Joe
rbairos
One Eyed Hopeful
Posts: 9
Joined: Fri Oct 30, 2015 7:23 am

Re: stationary projector onto visor

Post by rbairos »

Yes, I mean for motorized tracking, it would only be required to move coarsely at a slower speed.
Pixel-resolution tracking would still be done by projecting a sub-set of the pixels onto the goggles.
Focus control could be avoided by laser projector.

Even without the motorized tracking, I envision a setup like this costing roughly <$100 for hobbyists. The bulk of which would be a pocket projector, the rest for LED tracking, two lenses, and goggles.
Practically speaking, such a system would require the users head to remain in a roughly cubic area, 2 feet per side or so.
User avatar
cadcoke5
Binocular Vision CONFIRMED!
Posts: 210
Joined: Mon May 24, 2010 8:43 pm
Location: near Lancaster, PA USA

Re: stationary projector onto visor

Post by cadcoke5 »

I didn't realize you were thinking of such a small area. I suspect the coming high resolution, flexible OLED displays, will soon eliminate the advantages of your idea.

I looked up the actor projection idea I recalled from a few years ago, and came across a more recent project. The tracking is certainly better, though I still don't think it is good enough for VR. And note how slowly the models move their face. I am guessing that this is necessary for their system to keep up with the movements.
http://weburbanist.com/2015/01/21/face- ... n-mapping/
http://www.mirror.co.uk/news/technology ... en-4943458

Here is the original projector actor that I saw;
http://www.research.ibm.com/people/p/pi ... hics01.pdf

-Joe
rbairos
One Eyed Hopeful
Posts: 9
Joined: Fri Oct 30, 2015 7:23 am

Re: stationary projector onto visor

Post by rbairos »

Thats a great application of projection mapping.
Still though, if wearing the display with external optical tracking for position, gives reasonable results now, the same quality should be had, by projecting externally I would guess. (assuming no motorized movement).

Even with cheap highres oleds, it would still require tethering for power or data at minimum, but likely the better choice in the end for the same price.
[Edit: I meant upcoming cheap light oleds likely the better choice, as you point out.]


Cheers
User avatar
cadcoke5
Binocular Vision CONFIRMED!
Posts: 210
Joined: Mon May 24, 2010 8:43 pm
Location: near Lancaster, PA USA

Re: stationary projector onto visor

Post by cadcoke5 »

Note that the face mapping had targets attached to the actor's face. I suspect they were simply retro-reflective targets. These are like the reflectors you see on bicycles, and reflect the light coming from any direction, back to its source. If you use these types of targets on the faceplate, and put an IR light source next to the lens of the tracking camera, then you will not even need to use battery powered LEDs for your targets.

You said, " if wearing the display with external optical tracking for position, gives reasonable results now". But, if you look closely at the video in the first link I gave, the image certainly jitters some on the actor's face. This is the equivalent of having your virtual world experience an earthquake. Also, latency is a BIG challenge with VR. There are many articles discussing that. The time it takes to track the user's face, calculate any movements in the game or world, and then update the image for the projection, all add up for latency.

Over 10 years ago, I experimented with a head tracking program called "Cam to Pan" and used it to drive a stereoscopic display monitor. The lag that occurred was acceptable then, because the virtual object apparently existed inside the monitor. This type of system is sometimes called a "fishtank VR" system. The delays were quite acceptable, even with the processing capability at the time. because in a "fishtank VR" system it is acceptable for the object to not be anchored and to float around a bit. In a truly immersive VR world, that same level of floating would be a BIG problem.

There are a couple of tests for feasibility that you might do. Of course, if the test takes a lot of effort, it might be better to test your actual idea. The actor face mapping set-up is already designed, and if you can replicate that system, you could test the process with a still image. If the university where the project was done is near you, and the set-up still exists, you might even pay them a visit.

If that is not possible, you might try to see if you can simulate it using a Google Cardboard type of set-up. Calculate what the delay your system will have, and try to introduce the same amount of delay in the Google Cardboard system.

The other item to test, is the amount of "jitter" the image may have. I,e, the error in tracking. Introduce the same amount of jitter in your Google Cardboard setup, and see if it is acceptable. Another way to test is to setup your projection system, without any tracking, but play back an image that jitters a bit. If your VR world undergoes a non-stop magnitude 8 earthquake, then your idea will not work.

-joe
rbairos
One Eyed Hopeful
Posts: 9
Joined: Fri Oct 30, 2015 7:23 am

Re: stationary projector onto visor

Post by rbairos »

Sorry, I understand that latency and tracking jitter is a major concern for VR applications, but I'm proposing that the optical face tracking employed in the video you showed, is not the same type of optical tracking employed in current VR positional tracking systems, such as in the Oculus, or even Vive light house, which produce reasonable results.
I understand their positional tracking relies on either active LEDs, or active sensors respectively, taking away from the advantage of fully powerless goggles. (in addition to parallel systems tracking acceleration, orientation, etc).

Your feasibility tests are very reasonable.

[edit:] Anyone know what the best latency / accuracy available for passive IR targets is?
User avatar
cadcoke5
Binocular Vision CONFIRMED!
Posts: 210
Joined: Mon May 24, 2010 8:43 pm
Location: near Lancaster, PA USA

Re: stationary projector onto visor

Post by cadcoke5 »

Passive IR targets, if they are retro-reflective, have no different latency than LED ones. The retro-reflective targets can be quite bright, and do very well. I am puzzled why they aren't used more often.

-Joe
WiredEarp
Golden Eyed Wiseman! (or woman!)
Posts: 1498
Joined: Fri Jul 08, 2011 11:47 pm

Re: stationary projector onto visor

Post by WiredEarp »

Yeah, I actually prefer the retroreflective trackers for my TrackIR over my TrackClipPro. They seem to be detected better at extreme angles.
Post Reply

Return to “VR/AR Research & Development”