POV-Ray : Newsgroups : povray.binaries.images : Weekend landscape : Re: Weekend landscape Server Time
8 Aug 2024 22:14:46 EDT (-0400)
  Re: Weekend landscape  
From: Abe
Date: 26 Apr 2005 07:45:00
Message: <web.426e2934cf19ddd8f9b565770@news.povray.org>
"Thomas de Groot" <t.d### [at] internlnet> wrote:
> I love it!
>
> What about some (more) technical info?   :-)
>
> Thomas

Ooof! I was afraid somebody would ask.

The scene structure was not quite as clean and robust as I wanted it. That
is, when I plugged in a different height field later I found that I did not
quite have the control over appearance which I had hoped for. However, the
rough outline is as follows:

Lighting: Used an area light for the landscape and a point light for media
(atmosphere and cloud illumination). Pre-calculated the radiosity data for
a faster render.

Atmosphere: Three independent media containers (ala mediasky.pov with
minimum altitude set to zero). 1)Blue scattering and red absorption:
(extinction set to zero and replaced with an explicit absorption media for
better control over color). 2)Haze: White isotropic scattering for
atmosphere "thickness". 3)Sun glow: strongly forward scattering media.

Cloud: as previously described. Needed to be set behind the atmosphere
"shell" in order to avoid media container artifacts and also to get decent
shadow color (i.e. not gray). In my opinion the most successful CG clouds
are those that acknowledge color not only in the (directly) illuminated
regions but also in the shadows. Furthermore, there is often a color
variation with distance due to atmospheric effects. Fluffy white (rgb
<1,1,1> and gray rgb <.3,.3,.3> clouds in a blue (rgb <0,0,.75>) sky only
have limited appeal. That said, I'm probably as guilty as anybody, for
rendering aesthetically challenged clouds. :)

Terrain: 1024 square height_field. I used a procedural texture to try to
distinguish the rough areas from the smooth in the terrain. This is
actually what the main point of the scene was. Since steeper slope does not
exactly equal rougher terrain, I was trying for some other measurable
criteria, including 2nd derivative of the gradient, local difference in
gradient as calculated with two different point spacings, and sample point
difference from the surrounding local average. I think I used the second
method in this instance. I have not really nailed this down yet, and
sometimes I am not even sure I _really_ know what I am doing!

As far a camera placement and view orientation I used Terragen as a preview
tool (very practical) and translated the parameters to POV.

I hope this answers some questions. Sorry for the lack of any actual code.

Abe


Post a reply to this message

Copyright 2003-2023 Persistence of Vision Raytracer Pty. Ltd.