Re: [PanoToolsNG] Re:VTour (Was:Realtime 3d with 360 equirectangular view)
- On Fri, 01 May 2009 07:46:08 +0900, Bernhard Vogl <bvogl@...> wrote:
>Bernard, I find the one-above-the-other display very convenient for
>> I doubt very much if standard Shockwave3d content would work in stereo
>> with the current nvidia (DirectX) stereo driver. I had no luck with
>> Shockwave3d game content a year ago with the Iz3d stereo driver (which
>> makes almost every fullscreen DirectX game work in some sort of stereo
>> -- anaglyph, shutter glasses, Iz3d screen etc). However Shockwave3d is
>> a good display solution for side by side linked stereo panoramas on a
>> double width desktop -- with the monitor outputs going to twin
>> polarized projectors.
> Same here - no stereo with Shockwave.
> I personally find it very inconvinient to watch stereo images side by
> side (my eyes refuse to work that way) but you may also use SPi-V to mix
> the left/right images on the fly for display.
watching through periscope type viewers using mirrors. Admittedly these
are not so good with the increasingly common "wide" displays. Or so I
thought until I realized that mine had an unused option for rotation
through 90 degrees for use in portrait-oriented mode. Perfect!
I will write another note about an article on recent research that may
be relevant to your latest interest in stereo viewing.
I suggest you try to access the article in the April 25 - May 1
edition of the British "Ecoomist" magazine with the unlikely
title "Polyphemus does the hoovering."
This describes a new image-processing technique using carefully
optimised routines for deriving 3D positional information from
moving images. Simultaneous Localisation and Mapping (SLAM) is
primarly designed to process information from sensors using
lasers but Dr. Davidson of Imperial College London has applied
it to digital cameras. He and his colleagues have "recently
been able to show this new form of SLAM working at 200 frames
per second on a camera tossed from hand to hand, using just a
laptop computer to process the images."
This looks off topic, but the technique uses the very rapid
identification of congruent features in contiguous video
frames to trace movements and infer locations, so there is
a real possibility of the techniques being applicable to the
processing of panoramas and/or the adjustment of stereo display
parameters. If you can access it, I'd be interested to hear
your opinion. If you can't, I have access to the article on
the Economist website and might be able to obtain the text
and/or send a link (the latter doubtful as it is largely
restricted to subscribers).