there are a few beautiful models (pointed out by ken in
an earlier post) at:

http://www-graphics.stanford.edu/data/3Dscanrep/

does anybody know what their "range" models are?  they
talk about the range models as opposed to 3D models, but
i can't make any sense of it.

the reason i'm asking is cuz i really liked the buddha and
dragon they have, but i don't know if the zipped "range"
files they have would fly in pov.

thank you, miker

Post a reply to this message

In article <3cacae79$1@news.povray.org>, "ZSpider" <zsp### [at] gtenet> 
wrote:

> the reason i'm asking is cuz i really liked the buddha and
> dragon they have, but i don't know if the zipped "range"
> files they have would fly in pov.

I'm guessing a "range scanner" is a device that determines the distance 
to the surface of an object from a certain point or points, allowing it 
to get points on the surface of the object...sort of like real-world 
raytracing. The range image is like a depth buffer from a camera that 
wraps around the scene.
It looks like these files are provided for people doing research in 
surface reconstruction algorithms who don't have access to a scanner. In 
other words, these are not models, but raw data to be used to generate 
models. They would be fairly useless for POV.

Most of them have reconstructions available, these look like some kind 
of mesh format, maybe VRML (the tool they used is called vrip). Some of 
them have reconstructions in other formats...the Happy Buddha and Dragon 
are in Inventor, VRML, and QSplat.

-- 
Christopher James Huff <chr### [at] maccom>
POV-Ray TAG e-mail: chr### [at] tagpovrayorg
TAG web site: http://tag.povray.org/

Post a reply to this message

ZSpider wrote:
>
> http://www-graphics.stanford.edu/data/3Dscanrep/

range data=height_field
ply=mesh

I made ply2mesh2 converter. Tested it on the buddha.
http://luxlab.com/tmp/buddha.zip

Use like this:

global_settings{assumed_gamma 1}

camera {
  location  <0,0.5,-4>
  direction 1.8*z
  right     x*image_width/image_height
  look_at   <0,0,0>
}

light_source{<30,-30,-10> rgb<0.3,0.5,1>*0.4} 

light_source{<-30,30,-30> rgb<1,0.7,0.2>}

#include "buddha.inc"

object{Buddha pigment{rgb 1} finish{diffuse 0.7 ambient 0}
rotate 190*y scale 10 translate -1.5*y}


_____________
Kari Kivisalo

Post a reply to this message

Hi Kari
One has to be as fast as light to download your mesh2 as you put it in a
temp folder :-)
It is no more available at the URL you gave. :-(
May we get it anywhere else?
Many thanks
Marc

Post a reply to this message

Marc Jacquier wrote:
>
> It is no more available at the URL you gave. :-(

It's still there. What error did you get?


_____________
Kari Kivisalo

Post a reply to this message

> I'm guessing a "range scanner" is a device that determines the distance
> to the surface of an object from a certain point or points, allowing it
> to get points on the surface of the object...sort of like real-world
> raytracing. The range image is like a depth buffer from a camera that
> wraps around the scene.

The range scanners I am familiar with use a laser to determine distance.
The basic concept can be constructed using a simple range finder (I believe
hunters use these) mounted on a 2 axis mount.  The more expensive versions
are automated and more precise, they are capable of collecting millions of
points in minutes.  The resulting data is like having a sphere with a grid
on it - each grid cell has a range.  Now you have spherical coordinates for
all points seen by the scanner - easily converted to rectangular coordinates
and aligned with other scans.  The data for each cell can include intensity
(even for seperate channels of RGB for color!)

The data collected is like have a completly dark scene with a single point
light (no radiosity please :-).  The areas that are *lit* are the areas that
will be scanned up to the range limit and object reflectivity requirements
of the of the scanner.

> It looks like these files are provided for people doing research in
> surface reconstruction algorithms who don't have access to a scanner. In
> other words, these are not models, but raw data to be used to generate
> models. They would be fairly useless for POV.

You are right - just points have limited use directly in POV, but the number
of things that can be done with the points is limitless.  Most don't require
special knowledge, just creativity:

Using spheres for each of the points one can re-create the scene.  By
controling the size of the spheres the *looks* of the scene can be easily
adjusted.  This method can produce from fairly decent images.  If available,
intensity or color could be added to each shpere giving a better image.

The data could be used to generate a height field for terrain generation
(I've done this with an entire open pit coal mine).

A mesh can be created from the points giving an actual surface.

This is just a small spattering of what I have managed to do with point
clouds and POV.

Post a reply to this message

3cacae79$1@news.povray.org...
> there are a few beautiful models (pointed out by ken in
> an earlier post) at:
>
> http://www-graphics.stanford.edu/data/3Dscanrep/

Note that these models were converted by Mael to povray not so long ago:
http://news.povray.org/3ba8aae0@news.povray.org
http://news.povray.org/3ba8ae27@news.povray.org
He also did some interesting tests of subsurface scattering and patina with
them.

G.






--

**********************
http://www.oyonale.com
**********************
- Graphic experiments
- POV-Ray and Poser computer images
- Posters

Post a reply to this message