Figure 1. Amazon Mist (full sunlight)

Figure 2. Amazon Mist (overcast)

Figure 3. River View

Figure 4. 10-m Eucalyptus Tree

Amazon Rain Forest
Proof of Concept

This scene was created with more than 50 images of 12 species of tropical lowland evergreen rain forest trees. Tree heights were stratified in 3 layers to reproduce the giant emergent canopy (~40 meters), main stratum (~30 m), and shade-swelling trees (~10 m). Ground vegetation of smaller trees is apparent along the river banks. The full-sunlight camera view in Figure 1 is 250 meters above the flood plain. Figure 2 shows the same view under overcast lighting.

The camera in Figure 3 is 50 meters above the river. One thing that's lacking in these images is a sense of scale. Most viewers are accustomed to temperate forests and would scale the views accordingly. Rain forest canopies reaching a hundred feet and higher are hard to imagine.

Using 3D models to render many thousands of rain forest trees is not practical. Instead, detailed 3D tree models (>50k polygons) with leaf and bark textures were modeled in tree software and rendered with shadows in LightWave. Lighting was matched to the time of day in the rain forest scene. These tree images were then placed implicitly within the terrain software with heights and densities reflective of their real-world counterparts. The terrain software shades the images (selectively darkens each image) in response to scene lighting. Shadows are cast based on image opacity.

Aside from creating photorealistic scenes economically, this method integrates well with small-scale detailed environments modeled in other 3D software. For example, the foliage models created in Maya can be rendered as high-resolution images and placed in VNS. In this way, VNS can be used for large-scale scenes that match small-scale setups.

Software used to create these images:
Onyx Tree 6 (Onyx Computing)
xfrog 3.5 (Greenworks)
Photoshop 7 (Adobe)
LightWave 3D [7.5] (NewTek)
Visual Nature Studio 2 (3D Nature)

Copyright 2003 R Scott Cherba All Rights Reserved