Hi(gh)!

POVghanistan: Paghman Range Panorama
http://www.youtube.com/watch?v=R5pCZdj_VGM

After fixing the triangle gaps problem (see entry in p.b.i), I rendered 
my first small animation of a vista point in POVghanistan... to be 
precise, a mountain saddle in the 4,630-metres Paghman Range west of 
Kabul. Due to RAM limitations, I currently can use only one 
square-degree ASTER data tile at a time, this will probably not change 
in the nearer future (even now, POV-Ray's peak memory usage is 14.2 GB!).

In the coming weeks, I may add smooth triangles and a more elaborate sky 
and atmosphere model, preferably that of Bruno Cabasson.

Later on, I may render animations of low-altitude overflights covering 
more than one data tile, carefully customizing mesh2s to contain as much 
terrain as possible toward the horizon in viewing direction...

See you in Khyberspace!

Yadgar

Post a reply to this message

> In the coming weeks, I may add smooth triangles



I really think that you are getting there, with this. Smooth triangles 
would make it look much, much better IMO. Is it much work?


-- 
Regards
     Stephen

Post a reply to this message

Stephen <mca### [at] aolcom> wrote:

> > In the coming weeks, I may add smooth triangles
>

>
> I really think that you are getting there, with this. Smooth triangles
> would make it look much, much better IMO. Is it much work?
>
>
> --
> Regards
>      Stephen

I observed the same, but I observed even with this very high resolution
height_field I used for Mount Rainier (over 16000x16000 resolution) that in
close ups smooth doesn't work and you have the same as with Joergs animation.

Best regards,
Michael

Post a reply to this message

Hi(gh)!

On 17.07.2013 22:01, Stephen wrote:

>> In the coming weeks, I may add smooth triangles
>

>
> I really think that you are getting there, with this. Smooth triangles
> would make it look much, much better IMO. Is it much work?

Yes, it would be. Because I would have to calculate the normal list 
PRIOR to the face definitions - but as each normal should be the average 
of the normals of the triangles adjacent to each corresponding vertex 
(one at the northwestern and southeastern corner, two at the 
northeastern and southwestern corner, three at the edges and six with 
all others), the normals of the triangles and thus the triangles 
themselves must be defined before!

And, yes, provided I got the correct method for averaging the normals, 
smooth triangles would look indeed much better - but as the smoothness 
is faked, it comes at the cost of producing strange sights when placing 
objects (vegetation, houses etc.) on the mesh...

So instead I look for a method to generate isosurfaces from my 
1-square-degree earth surface slices... I think it would start with a 
pigment function derived from the original elevation heightfield (which 
I also had to generate from DEM data in a non-POV format)... but this 
pigment would be an imagemap constrained to a certain geographical 
region of a sphere rather than simply mapped around the sphere.

I asked the question whether this could be doable with POV-Ray about two 
weeks before on p.b.i., meanwhile also on a German POV-Ray forum, but 
received no useful answers yet.

See you in Khyberspace!

Yadgar

Post a reply to this message

> So instead I look for a method to generate isosurfaces from my
> 1-square-degree earth surface slices... I think it would start with a
> pigment function derived from the original elevation heightfield (which
> I also had to generate from DEM data in a non-POV format)... but this
> pigment would be an imagemap constrained to a certain geographical
> region of a sphere rather than simply mapped around the sphere.

   Maybe my HF2ISO experiments can give you some ideas (except for the 
spherical part):

http://www.ignorancia.org/en/index.php?page=Heightfield_based_Isosurfaces


--
Jaime

Post a reply to this message