RE: Seitz Roundshot and QTVR technique

[Mike Sinclair <sinclair@xxxxxxxxxxxxx>]

I was distinguishing panos made with a continuously rotating lens/slit
camera like the RoundShot from the pan'n'snap ones that Apple's QTVR help
make famous. VideoBrush is more akin to the pan'n'snap as a number of
discrete 2D rectilinear images are warped, stitched together, then unwarped
to form the final interactive image. This makes the pan'n'snap (VideoBrush)
technique more susceptible to perceptual anomalies as object movement,
lighting changes and nodal point violation than the continuous (and gradual)
image generation of the rotating lens/slit camera. I do agree that the
VideoBrush technique makes electronic publishing easier as it's virtually
instant and keeps the images together and in order, sometimes a problem with
the discrete image generation and retrieval as from film and electronic
still cameras. I didn't distinguish between the many-varied ways to
interactively view panoramas - I think each as its own merits. The holy
grail is getting the highest resolution/quality interactive pan, quickest to
the screen and with the least difficulty. Each has its own "warts" here too.

Problems I have with VideoBrush are that it very often has difficulty
accepting video input as it thinks the pan may not be continuous and
prematurely halts and complains (even if the video was generated on a
motorized panning tripod attachment); may try to follow a scene object
thinking it's horizontal (when it's not), resulting in a very skewed pan;
and it doesn't seem to take advantage of all the resolution available in the
digitized video. With the pan'n'snap technique, it is also extremely
difficult to stitch multiple pans together, taken at different tilt angles,
to yield a wider vertical field of view.

I know my opinion(s) isn't necessarily a popular one but putting one's
efforts, money and technology into better tripod attachments (for still
cameras) isn't necessarily progress though it has an obvious market (again
my own opinion). It only raises the barrier to entry of the would-be amateur
panorama photographer. There, I said it.

-Mike-

> -----Original Message-----
> From:	thomas b. kunz [SMTP:tbk@xxxxxx]
> Sent:	Saturday, November 07, 1998 10:14 AM
> To:	panorama-l@xxxxxxxxxxxxxxxxx
> Subject:	Re: Seitz Roundshot and QTVR technique
> 
> I agree with you Mike, If you said - the VideoBrush's technique - It makes
> web-resolution panos VERY easy to make. I'd never used another, because it
> so easy. Results are at www.tbk.de/galerie. Enjoy it!
> 
> Q: Do you, as Panoramists, call the Videobrush- technique a QTVR-technique
> (pan'n'snap) if you don't present it with Apples-QTVR, but instead with
> JAVA-Applets?
> I think this should be seen more different, because there are a lot of
> techniques to produce and show web-resolution panos on the Web. Agree?
> 
> Thomas
> 
> 
> 
> 
> -----Ursprüngliche Nachricht-----
> Von: Mike Sinclair <sinclair@xxxxxxxxxxxxx>
> An: panorama-l@xxxxxxxxxxxxxxxxx <panorama-l@xxxxxxxxxxxxxxxxx>
> Datum: Freitag, 6. November 1998 01:27
> Betreff: FW: Seitz Roundshot and QTVR technique
> 
> 
> Joshua eludes to the beauty of pan-scanning an image with a 1D sensor -
> with
> either a 'slit' of film or a linear CCD. During my 2 yr. ownership of a
> Roundshot, I religiously adjusted the nodal point adjustment rails to what
> the computer told me. After a few weeks and getting lazy, I ignored the
> adjustment and saw no difference in the images. This, in part was because
> of
> the slit and part because I used a 14mm lens. I'd imagine if you were
> using
> a significantly longer fl lens, objects violating the nodal point
> calculation would appear horizontally blurred - their image components
> would
> have a non-zero velocity with respect to the film.
> 
> 
> As a side note:
> 
> A digital video camera's video can be processed to yield a true 720 x 2880
> (lens dependent) x 360 panorama. VideoBrush's technique is more akin to
> the
> pan'n'snap variety and produces inferior results (IMHO). If you violate
> the
> nodal point rule (on purpose) that the pan'n'snap photographers are slave
> to, you can get  truly stunning stereo panorama pairs - with only one
> camera
> and one pass! Through computer processing, you can derive stereo pairs
> with
> as much as a 12" (or more) IPD (distance between viewpoints). I'm hoping
> the
> lab in which I used to work will soon make available or license this
> software. It makes web-resolution panos VERY easy to make - much easier
> than
> the snap'n'pan variety. Also, through multiple pans, pole-to-pole coverage
> (360 x 180 degrees) is attainable. Another advantage of this technique is
> the auto exposure capability as a function of pan angle (similar to what
> the
> new RoundShots give you). This comes with every DV camera! I've seen image
> compressions of 7+ stops while panning an interior next to a window with a
> bright sunlit exterior view - both with good exposure. Almost all of the
> pan'n'snap real estate pans suffer from "window burnout". Another
> advantage
> is that you get about 60 pans per $20 digital tape - hard to beat compared
> to FlashRam or film.
> 
> -Mike-
> 
> 
> 
> 
>  << File: tbk-vcard.vcf >>