I'm using PovRay to visualise the distribution of galaxies from a
recently completed survey. The starting point is the 3D position
of 1/4 million galaxies. The survey is in 2 wedges, see
   http://astronomy.swin.edu.au/~pbourke/tmp/2df/set4/plan.gif
Each dot is a galaxy.

I take a 100Mpc cube (in white in the above diagram) and map some
variable onto the galaxy position. I also derive a density function
and create a volumetric dataset which I can represent as such, see
   http://astronomy.swin.edu.au/~pbourke/tmp/2df/set2/points_gas_1.jpg
for example, or I can compute isosurfaces (marching cubes), see
   http://astronomy.swin.edu.au/~pbourke/tmp/2df/set4/hi_transp.jpg
and recently I've been making fishnet stockings
   http://astronomy.swin.edu.au/~pbourke/tmp/2df/set4/mesh2.jpg
(Sorry for the big picture).

So finally to my question. I would like to experiment with an isosurface
(current a huge bucket of triangles in a mesh) where the transparency
was proprotional (or some other function) to the angle between the
camera and the triangle normal. In particular, where the dot product
is low I want high transparency, where the dot product is large I
want a high opacity. I'm interested in representing the borders of the
isosurface while being able to see the objects inside.

Ideas?
-- 
Paul Bourke
pdb_NOSPAMswin.edu.au

Post a reply to this message

On Wed, 04 Jun 2003 20:08:58 +1000, Paul Bourke <pdb### [at] swineduau> wrote:
> So finally to my question. I would like to experiment with an isosurface
> (current a huge bucket of triangles in a mesh) where the transparency
> was proprotional (or some other function) to the angle between the
> camera and the triangle normal. In particular, where the dot product
> is low I want high transparency, where the dot product is large I
> want a high opacity. I'm interested in representing the borders of the
> isosurface while being able to see the objects inside.
>
> Ideas?

AOI pattern ? Available in MLPOV and will be in next MegaPOV.

http://news.povray.org/povray.binaries.images/30297/

This can be faked with official 3.5 by applying different slope pattern to each
triangle in mesh. 

http://news.povray.org/povray.binaries.images/31718/

Both ways produce border with different width. To get even border you can post
process your rendering. Render first frame without isosurface. Then render
second frame with isosurface but without any other object in quick color mode
(Quality=1). Then render third frame with locating first frame on the plane
before camera with macros included in screen.inc official include file and using
eval_pattern over second frame locate set of spheres making border. Hope I wrote
it in uderstable way.

ABX

Post a reply to this message

> This can be faked with official 3.5 by applying different slope pattern to 
> each triangle in mesh. 

Sure but I was hoping to seperate the data generation (for a particular
isosurface at least) from the camera animation in PovRay.

> http://news.povray.org/povray.binaries.images/31718/
> 
> Both ways produce border with different width. To get even border you can post
> process your rendering. Render first frame without isosurface. Then render
> second frame with isosurface but without any other object in quick color mode
> (Quality=1). Then render third frame with locating first frame on the plane
> before camera with macros included in screen.inc official include file and using
> eval_pattern over second frame locate set of spheres making border. Hope I wrote
> it in uderstable way.

I think I'm a bit lazy for this one.
-- 
Paul Bourke
pdb_NOSPAMswin.edu.au

Post a reply to this message

On Wed, 04 Jun 2003 20:38:34 +1000, Paul Bourke <pdb### [at] swineduau> wrote:
> > Both ways produce border with different width. To get even border you can post
> > process your rendering. Render first frame without isosurface. Then render
> > second frame with isosurface but without any other object in quick color mode
> > (Quality=1). Then render third frame with locating first frame on the plane
> > before camera with macros included in screen.inc official include file and using
> > eval_pattern over second frame locate set of spheres making border. Hope I wrote
> > it in uderstable way.
>
> I think I'm a bit lazy for this one.

Lazy ? So that's why you choosed general group for advanced question ;-)

Personally I think it would be the fastest way: first image rendered without
probably slowest isosurfaces, second image rendered with isosurface without any
shading or lightning, third image just set of spheres over plane.

ABX

Post a reply to this message

Paul Bourke <pdb### [at] swineduau> wrote in news:pdb_NOSPAM-
157### [at] netplexaussieorg:

> I'm using PovRay to visualise the distribution of galaxies from a
> recently completed survey. The starting point is the 3D position
> of 1/4 million galaxies. The survey is in 2 wedges, see
>    http://astronomy.swin.edu.au/~pbourke/tmp/2df/set4/plan.gif
> Each dot is a galaxy.
> 
> I take a 100Mpc cube (in white in the above diagram) and map some
> variable onto the galaxy position. I also derive a density function
> and create a volumetric dataset which I can represent as such, see
>    http://astronomy.swin.edu.au/~pbourke/tmp/2df/set2/points_gas_1.jpg
> for example, or I can compute isosurfaces (marching cubes), see
>    http://astronomy.swin.edu.au/~pbourke/tmp/2df/set4/hi_transp.jpg
> and recently I've been making fishnet stockings
>    http://astronomy.swin.edu.au/~pbourke/tmp/2df/set4/mesh2.jpg
> (Sorry for the big picture).
> 
> So finally to my question. I would like to experiment with an isosurface
> (current a huge bucket of triangles in a mesh) where the transparency
> was proprotional (or some other function) to the angle between the
> camera and the triangle normal. In particular, where the dot product
> is low I want high transparency, where the dot product is large I
> want a high opacity. I'm interested in representing the borders of the
> isosurface while being able to see the objects inside.
> 
> Ideas?


If you use your density function/data to render an isosurface in POV-Ray,
then you can define another function that calculates the dot product
stuff.

This second function can then be used to control a pigment_map or a
texture_map so that you get an "Angle Of Incidence" controlled texture
(e.g. with varying opacity).


For some sample code, see my recent post:

'Source code for "AOI textures for isosurfaces"'
-> povray.text.scene-files


Tor Olav

Post a reply to this message