Advances in autostereoscopy (Photonics Spectra)

[PinkysBrain]

You have one of those Vaio's with optional lenticular 3D screen?

AFAIK those also use the webcam to track the viewer ... these approaches are actually pretty advanced, they slant the lenticular and use an appropriate subset of the pixels based on the position of the viewer's eyes tracked with a webcam (because some pixels will be visible to both the left and the right eye).

Doing something similar with DIY seems a huge technical challenge to me ... using semi-continuous views zones with lenticulars is easier, but you lose a lot more resolution (because you need a lot more views).

[Dinesh]

Gregg, I'm sorry to say that I've only just got to read your article on autostereoscopy. I meant to read it when you first posted. Right now, my desk is covered with unread copies of OPN and Physics Today which, hopefully, I'l get to reading one of these days!

Anyway, it's very informative about the various techniques and organisations working in this field. As a matter of fact, Alioscopy is based here in San Diego. Last year, they invited a few local San Diego companies to a party at their facility and we were amongst the invitees. They had their displays on show and I got to say their displays were pretty impressive!

However, I have two questions and a possible request:
On the second page, about half way down, you say:
"Thus most experimental high-view-count 3-D displays have limited color fidelity, require multiple light modulators and provide significantly fewer than 300 views, which some researchers believe is desirable"
and
"Thus a 100-million-pixel image was created at the expense of color fidelity"

This seems to indicate that there is a relationship between speed of "photon arrival" and colour fidelity. I was wondering what this relationship was. Presumably the colour rendition, ie the colour mixing ability, of the photons, according to the latest CIE chart (CIECAM97?), would depend on the bit depth per colour channel alone? I mean, if a scene contained, say, a particular yellow made up of a particular mixing of specific primaries, then providing the "photon guns" each produced photons of that specific primary and they each fired with a particular intensity, you'd get your desired yellow. The speed of the photon arrival would mitigate the persistence of vision, so that the yellow in question would last just a little bit longer than it ought to, but you'd still see yellow. Am I seeing this wrong?

On page 4, under "What's next", you say:
"Among the hottest topics for 2012 are the benefits of tightly coupling the mathematics of optimization to the functions of pixels and barriers..."
and, on page 5
"The group's analysis also extended the linear algebraic notion of rankto the world of display technology. A rapid sequence of various multilayer frames within the eye's integration period results in 3-D imagery reconstructed with higher fidelity, increasing the system's rank."

I was curious what this optimisation technique was. Does it have anything to do with the latest work on wavelet compression (JPEG 2000)? Also, does "higher fidelity" refer to higher colour rendition or higher resolution? This seems to have a bearing on the previous question regarding colour rendition and photo arrival speed.

The request is: where can I get a hold of your reference 3 by Halle?

[favalora]

Dinesh -

Thanks!

One of the challenges of the article was my need to not terribly over-run the word count limit, while saying things that were useful to newcomers and not weighed down by lots of exotic counter-examples. I recall needing plenty of "often"s and "usually"s and "approximately"s.

Anyhow:

This seems to indicate that there is a relationship between speed of "photon arrival" and colour fidelity. I was wondering what this relationship was. Presumably the colour rendition, ie the colour mixing ability, of the photons, according to the latest CIE chart (CIECAM97?), would depend on the bit depth per colour channel alone?

There are a few ways I could interpret your question. So: what I was trying to say was that, if you assume the spatial light modulator is binary, its switching frequency and number of pixels are important because you can use spatial (e.g. halftoning) and temporal (e.g. pulse-width) modulation to give the appearance of a color gamut. Know what I mean? Unless your question was a diplomatic way of suggesting that I forgot about continuous-range modulators, rather than binary ones, then yes, I forgot about that case. Many of the people working in high-fidelity autostereo use an SLM like a DMD, which is normally run in a binary mode.

I was curious what this optimisation technique was. Does it have anything to do with the latest work on wavelet compression (JPEG 2000)? Also, does "higher fidelity" refer to higher colour rendition or higher resolution? This seems to have a bearing on the previous question regarding colour rendition and photo arrival speed.

It's probably best if you visit this page at the Camera Culture Group for the various papers describing these optimized barrier displays. My phrase "higher fidelty" was intentionally vague because of the complexity of the work. In the simplest sense, the "fidelity" refers to "more than just 2 discrete views."

The request is: where can I get a hold of your reference 3 by Halle?

Google Scholar, dude! Halle's overview of autostereo is here. I think you ought to check out the Plesniak reference that he co-authored, too. (This one.) Thought-provoking stuff...

Come to think of it, this selection of optics projects should keep your mind buzzing for a while.

-Gregg

[Dinesh]

There are a few ways I could interpret your question. So: what I was trying to say was that, if you assume the spatial light modulator is binary, its switching frequency and number of pixels are important because you can use spatial (e.g. halftoning) and temporal (e.g. pulse-width) modulation to give the appearance of a color gamut.

Actually, this is exactly what I was looking for. I was wondering how the colour gamut was created. For the past year or so I've been very interested in how our colour perception has changed as a result of there being far more narrow-band and quasi-mono-chromatic sources in the general environment. The original CIE 1931, based on the experiments of Guild and Wright, used broad band sources and based their primaries on the lamps in the experiment. I think today with all these leds and quasi-laser sources being so predominant our photopic peak may have shifted and the entire photopic curve may have shrunk. When we got our present lcd TV Joy mentioned that the picture "looked 3D". It occured to me that the highly saturated colours in the display, and our different perception of the various colours, may have had the effect of perceiving depth. The greens and yellows appear nearer than the reds and blues because they appear (much?) brighter, and, of course, the brain translates the brighter image as being nearer. I was wondering if this effect carried over into the autostereo world, where depth perception has to be pretty carefully calibrated I'd imagine.

Unless your question was a diplomatic way of suggesting that I forgot about continuous-range modulators, rather than binary ones

Diplomatic? Moi? Joking, no I did not have this in mind.

Thanks for the references also. I had never heard of Google Scholar. Shows what an internet luddite I am!

[edit] By the way, if you or anyone else is interested in the questions arising from the initial assumptions of Guild and Wright, on which the 1931 CIE is based, you may want to look at the works of Stockman and Sharpe.

[Dinesh]

For anyone interested and in the Las Vegas neighbourhood, I just got this email notice from Alioscopy:
http://www.alioscopyusa.com/3d-comm-event

The event referred to is 3D Comm: http://www.insightmedia.info/conferences/3dcomm.php