>I disagree that they are easier or faster to produce, just
>different.
> Kevin
I disagree with your disagreement...According to an Apple person
"It took two photographers and a lighting crew around four 12-hour days to take
around 6000 pictures in the Paramount sets. The pictures were stitched into
around 200 panoramas and processed into QTVR movies. I estimate the processing
time is on the average of 1 hour per panorama over a four week period."
That means that 200 panoramas divided by 48 hours = approximatly 5.25
hours for each processed panorama. I defy anyone to create such detailed
environments in a little over 5 hours.
If these environments had been put together using conventional geometrically
based techniques the building and rendering would still be taking place.
Please keep in mind that these are highly detailed environments, not the
usual toy environments we often see in demos.
I stick to my assertion that image based environments while different are
vastly easier to create than geometrically created environements. My only
concern here is that VRML has sufficient hooks to allow for
these types of environments. Simply mapping a texture to the inside of
a cylinder is a useful hack but may not be sufficient for future enhancements.
Work at UNC and CMU (see http://www.cs.unc.edu/~mcmillan/plenoptic.abs.html
and http://www.ius.cmu.edu:80/afs/cs/user/pjn/www/3DStudio.html)
requires the use of range maps to allow travel within the imaged environment.
It would be a shame if future VRML enhancements can not extend to these
types of environments.
Sandy Ressler
------------------------------------------------------------------------------
NAME: Sandy Ressler TELE: (301) 975-3549
USMAIL: National Institute of Standards and Technology FAX: (301) 926-3696
Bldg. 225 Rm B-266 email: [email protected]
Gaithersburg, MD 20899