This isn't a very interesting picture, but I wanted to present the
technique: one height-field made from a noisy and sharp image (with the
rock texture on it), and another height-field made from a copy of the
first, but blurred considerably (with the snow texture). They are both
scaled exactly the same, but the blurring of the second will make it lower
where the peaks are on the first, and higher _around_ the peaks. Easier
done than said:

--
#local Mountain = height_field {
 jpeg "mountain.jpg" // sharp, noisy image
 smooth
}

#local Snow = height_field {
 jpeg "snow.jpg" // blurred version of mountain.jpg
 smooth
}

union {
 object { Mountain scale <100,10,100> texture { Terrain } }
 object { Snow scale <100,10,100> texture { Snow_tx } }
}
--

I used the same heightfield combination four times in the image, and it was
interesting to see that by declaring the height-fields first (as in the
code above), the parse-time was shorter than just writing the height-fields
into every union of the combined objects. The mountain.jpg image is 1024 by
1024 pixels, but still there are some 'staircase' patterns, especially on
the left. The snow.jpg is only 256 by 256, maybe that is what causes this.

Also, it rendered slower than I expected, as some part of the height-field
(height-field cell I think it was) only had about 6 percent success-rate.

H

Post a reply to this message

Attachments:
Download 'snowscape.jpg' (66 KB)

Preview of image 'snowscape.jpg'