Re: Re^2: 3D-LCD Software??

[P3D Larry Berlin <lberlin@xxxxxxxxx>]

At 10:50 AM 1/9/97 CET, you wrote:
>................
>I was referring to computer animated scenes, but I guess it's also true
>for real scenes. You should understand that there is a difference between
>intra-frame and inter-frame motion blur. Intra-frame motion blur occurs
>if an objects moves fast enough in vertical direction to be scanned in
>different subsequent lines. Inter-frame motion blur happens if the scanning
>device is too slow to keep up with object motion (after-image effects
>caused for example by too slow shutter speed).
>   In computer graphics, an animation sequence looks much more realistic
>if you simulate motion blur. Many rendering systems neglect this, because
>it's an expensive operation. My personal experience is that it is sufficient
>to take into account the advancement of an object during each time step,
>i. e. each half frame, regardless if it is perceived by the left or right
>eye only. Obviously the brain doesn't separate here.
>
>Joerg

Hi Joerg,

I understand your point but I maintain that there is one point on which the
brain usually relies, with both eyes open, they see the *same instant* of
movement at the *same time*. For LCD viewing, we use short time intervals
and persistence of vision to present each eye with the part of the scene
appropriate to it, an artificial but successful strategy. The only way to
improve on this timing loss is to maintain perfectly coincident information.
Regardless of intra or inter frame movement or movement at any velocity, the
left and right view should be considered as representing the same identical
moment and provide extremely coincident information to each eye. ANY change
from this, such as your half frame method, may work in the oveall scheme of
things, but degrades the quality of the stereo experience to a measurable
degree.

We need to *not stop* when we've discovered a level of *sufficient*! We need
to go beyond that level for best optimized results. If you plan to market
your work such a step will almost guarantee you're ahead of the competition
who may still be using the half frame method! If you want to use half frame,
then double the sample rate so both eyes benefit from all changes.

In your example you are cheating the mind of accurate information by
squeezing more action into a given number of frames or length of animation
and sacrificing some of the stereo effect. It may work but it's not what I
would recommend for quality stereo content. Maybe it's a good technique for
internet use where anything to reduce the file size is important. Try it
both ways with something you really like and see if you don't appreciate the
more complete version better, given a choice! It's also good to maintain
such quality elements as far as possible since other necessary steps such as
interlacing, when applicable, will degrade the image as well.

Consider the drawbacks, any movement in the horizontal direction will
provide inaccurate depth perception for the item in question (blurred or
not), and the mind is very sensitive to this factor. Any vertical movement
results in vertical disparity in which case the mind relies more on one
eye's version of the scene than the other, again reducing the stereo value
of the information and even encouraging the mind to not seek or rely on the
depth information. Motion blur is minimal in reality except for the fastest
motion. In graphics it's sufficient to suggest motion with a bit of blurring
carefully done. I'd prefer the accurate scene with maybe a bit of blurring
included equally for both eyes, rather than scrambling the stereo
information with object motion. There is a difference.

Larry Berlin

Email: lberlin@xxxxxxxxx
http://www.sonic.net/~lberlin/
http://3dzine.simplenet.com/


------------------------------