Hi

I'm using a Digital Elevation Model,converted to a  POV 
heigthfield, to try to produce a more realistic simulation of a 
satellite image than simple shading. I've modified a little the 
following script which was used for other DEM work, however I 
am particularly interested in simulating the effects of changes in 
solar elevation (ie y value of the illuminator). If this angle is low 
then terrain undulations are crisply picked out. Greater than 
20degrees and they become virtually unidentifiable.

In general shaders (and the current script) are unable to cope 
with the low levels of ambient light and, I assume, a greater 
degree of diffusion. I have posted on my web site a couple of 
graphics; the first shows a "real" satellite image 
(http://www.geocities.com/Yosemite/Rapids/5855/satellite_im
age.jpg) which shows small linear landforms clearly depicted. 
The second 
(http://www.geocities.com/Yosemite/Rapids/5855/rendered.jp
g) shows my rendering so far (both using the same solar 
illumination angle). This is much darker however similar linear 
landforms are visible in the centre (although its a different area). 
Does anyone have any suggestions for simulating the results in 
the image??


Cheers

Mike

#include "shapes.inc"
#include "colors.inc"
#include "textures.inc"

camera {
   location <0.0, 1000.0, 0.0>
   //direction <0.0, 200, 4.0>   //4.0 for telephoto effect
   //up <0.0, 1.0, 0.0>        
   //right <4/3, 0.0, 0.0>
   look_at <0.0, 0.0, 0.0>
}

// Define a couple of colors for the light sources.
#declare MainLight = color red 0.9 green 0.6 blue 0.6;
#declare FillLight = color red 0.23 green 0.23 blue 0.25;

// Light source (main)
light_source { <300.0, 154.0, -300.0> White }
//light_source { <500.0, 300.0, 600.0> color FillLight }

height_field  {
   tga "d:\drumlindem.tga"    // x 0-1.0 y 0-1.0 z 0-1.0
   smooth
//   pigment { color red 0.3 green 0.59 blue 0.11 }
   pigment { color red 0.4 green 0.4 blue 0.4 }

 
   finish {
	  //crand 0.025         // dither  - not used often, but 
this image needs it. 
	  ambient 1.5         // Very dark shadows
	  diffuse 0.4         // Whiten the whites
	  //phong 0.1          // shiny
	  //phong_size 100.0    // with tight highlights
	  //specular 0.1
	  //roughness 0.6
	  //brilliance 5.0
   }
   
   translate <-0.5, 0.0, -0.5>  // Center the image by half  
   scale < 1200, 64.0, 1200 >  // scale horiz x 1200 and vert x 
64

}

// end of file

Post a reply to this message

Mike Smith wrote:
> 
> Hi
> 
> I'm using a Digital Elevation Model,converted to a  POV
> heigthfield, to try to produce a more realistic simulation of a
> satellite image than simple shading. I've modified a little the
> following script which was used for other DEM work, however I
> am particularly interested in simulating the effects of changes in
> solar elevation (ie y value of the illuminator). If this angle is low
> then terrain undulations are crisply picked out. Greater than
> 20degrees and they become virtually unidentifiable.
> 
> In general shaders (and the current script) are unable to cope
> with the low levels of ambient light and, I assume, a greater
> degree of diffusion. I have posted on my web site a couple of
> graphics; the first shows a "real" satellite image
> (http://www.geocities.com/Yosemite/Rapids/5855/satellite_im
> age.jpg) which shows small linear landforms clearly depicted.
> The second
> (http://www.geocities.com/Yosemite/Rapids/5855/rendered.jp
> g) shows my rendering so far (both using the same solar
> illumination angle). This is much darker however similar linear
> landforms are visible in the centre (although its a different area).
> Does anyone have any suggestions for simulating the results in
> the image??
> 
> Cheers
> 
> Mike

   A couple of things come to mind. It looks like the sun has the
 advantage of scattering light through the atmosphere where you are
 using a single point light as you source of illumination. The are a
 couple of things you can try. First and easiest would be something
 like a sky sphere, fog with a high offset so it is way above the
 ground, or the simple background color option. Each of these will
 give you a uniform global illumination without causing shadows.
 Where your light source hits the terrain it's intensity will be
 much more than the feeble light the virtual background light sources
 put out and you should still get the contrasting shadows you are
 looking for.
   Another possibility would be to use a cylindrical light source so
 your light location is spread out more than from a single point.
 It also has the advantage of casting light rays in a parallel path
 which the omni directional point lights you are using now do not.

-- 
Ken Tyler

mailto://tylereng@pacbell.net

Post a reply to this message

Ken

Thanks for the advice (and also the previous advice on heigth 
field editors a while back!). I tinkered with some of the settings 
you suggested; alas to no avail! I tried using the sky sphere and 
fog separately and together and they appeared to make little 
difference. I assume I can't be using them correctly, particularly 
the fog option. I also tried using cylindrical light but this totally 
washed out the image. As you can see I'm not particularly au 
fait with all the settings and POV itself. One thing I forgot to 
mention, but is obvious when you look at the images is that it is 
rendered looking vertically down onto the landscape.

As an aside, how can I find the bounds of the coordinates 
currently being used for the imported heigthfield??

Thanks

mike

Post a reply to this message

Mike Smith wrote:
> 
> Ken
> 
> Thanks for the advice (and also the previous advice on heigth
> field editors a while back!). I tinkered with some of the settings
> you suggested; alas to no avail! I tried using the sky sphere and
> fog separately and together and they appeared to make little
> difference. I assume I can't be using them correctly, particularly
> the fog option. I also tried using cylindrical light but this totally
> washed out the image. As you can see I'm not particularly au
> fait with all the settings and POV itself. One thing I forgot to
> mention, but is obvious when you look at the images is that it is
> rendered looking vertically down onto the landscape.
> 
> As an aside, how can I find the bounds of the coordinates
> currently being used for the imported heigthfield??
> 
> Thanks
> 
> mike

I'm not sure what you mean by the bounds. If you mean the scaling of
the HF it will always be x=1 z=1 and y = 0-1 depending on the height
of the pixels for that location. The overall scale of the HF will
be determined by the final scale used in the HF statement. If you need
a net list of all elevations on the HF that is dependent upon the
software producing it and it's capabilities.


  Try the code below which is a stripped down version of the original
with some changes made. I lowered the ambient from 1.5 to .12. the
ambient when set that high will take over almost complete control
of the shadows which in your case is something yu don't want to do.
A higher diffuse value will make the pigment used more reactive to
the light tht falls on it and will produce cleaner shadow lines.

  I removed your HF image name to include my own so you will have to
replace it. I also scaled the HF by half because the Hf image I used
was to small to gieve me any definition with the scale you had and
I lowered the camera and cut halfed the distance of the lights to
reflect this. You can change it back by doubling the camera height
and the light distances to have what you had before. The effects
should not change but you be the judge of that.
  I think it is closer to what you wanted but without the original Hf
image to play with and the reference photo here to compare to it is
all subjective. I added a gradient texture to give me an idea of what
the elevations were but you could easily change that to anything you
want even back to the grey you were using before.

  camera {location <0.0, 600.0, -0.1>look_at <0.0, 0.0, 0.0>}

  // Try different settings below to see difference

  //light_source { <100.0, 54.0, -200.0>rgb 1}
  //light_source { <200.0, 54.0, -300.0>rgb 1}
  //light_source { <300.0, 54.0, -400.0>rgb 1}
  light_source { <400.0, 54.0, -500.0>rgb 1}
  //light_source { <500.0, 54.0, -600.0>rgb 1}
  //light_source { <600.0, 54.0, -700.0>rgb 1}

  height_field  {
   tga "mt-h.tga"
    smooth
     texture{
      pigment{
       gradient y
        color_map{
         [  0 rgb<0,.6,0>]
         [.30 rgb<.9,.87,.72>*.75]
         [.30 rgb 1]}
        scale 1
        }
       }

   finish {
          ambient 0.2 // Can take orver control of brightness keep low
          diffuse 1.0 // Makes the pigment more reactive
                      // to the light that strikes it
   }

    translate <-0.5, 0.0, -0.5>
   scale<1200, 64,1200 >*0.5
 }

-- 
Ken Tyler

mailto://tylereng@pacbell.net

Post a reply to this message

Ken


Thanks very much for the suggestions. I've done some more 
tinkering around with the settings and the results are much 
more encouraging. However the settings are want to control in 
relation to the real sun are its vertical and horizontal position in 
the sky (ie x,y,z). Your settings gave a range of 5-15degrees for 
the angle of elevation, which were a touch on the dark side; I 
changed this to give an elevation of 35degrees (300,350,-500) 
and the results look quite close to a real image given an 
elevation angle of ~10degrees. Do you have any suggestions 
for "correcting" this?

If it is useful I have put the tga file on my website 
(http://www.geocities.com/Yosemite/Rapids/5855/tga.zip); 
both this and the sample image should have a pixel resolution 
of ~30m, so they are comparable.

Cheers

Mike

PS If the image is automatically scaled between 0 and 1 on all 
three axes, sun position y=54 makes it 54 times higher than the 
maximum height??

Post a reply to this message