Virtual plants are coming out of the shade
INRIA and CIRAD have joined forces in a Coordinated Research Initiative that bears an evocative name, Soleil (Sun). The goal of Soleil is to grow virtual plants taking into account the amount of light received by each leaf.
 |
An understanding of plant growth would provide precious information about the best way of exploiting the too scarce farmland in tropical and sub-tropical regions. In order to study this problem, Philippe de Reffye's team at Cirad (Center for International Cooperation in Agronomic Research for Development) modeled the architecture of plants and designed a software package called Amap that is capable of simulating plant growth based on light intensity, water and mineral supply and so on.
Up to now, Amap had to suppose that each leaf was receiving the same amount of light energy. However, this hypothesis ignores the play of shadow and light produced by the superposition of leaves and by the presence of nearby trees and buildings. Now, plant matter production is conditioned by the amount of light that can be used for photosynthesis.
. A solution to this problem required a study of complex illuminations. This is precisely one of François Sillion and Cyril Soler's (IMAGIS project, INRIA Rhône-Alpes) specialties. In effect, their work strives to virtually recreate the light ambiance produced by indirect lighting, by simulating the radiative exchanges between lighted objects.
The two researchers devised Bright, a software package based on the radiosity method, which was well suited to the evaluation of the amount of light separately received by each leaf, taking the interaction with neighboring leaves into account. Cirad and INRIA researchers thus coupled Amap and Bright. In doing so, they solved additional difficulties: leaf transparency, leaf wavelength-dependent radiation, leaf superpositions with the effect of screening the light and the mobility of the light source (the Sun).
The problem of the very high number of leaves to be considered also needed to be addressed. It was solved by the development of hierarchical methods that can be used to compute light exchanges at the level of a leaf cluster. Then, a so-called instantiation procedure is defined so that it is never necessary to consider all the leaves at once. The procedure goes through successive zooms, first determining the radiative exchanges on a big branch, then on a smaller branch and so forth down to the leaf level.
Today, researchers are able to simulate the growth of a plant over a period of about ten years and to determine its shape depending on whether access to light is constrained or not. At this stage, the plant is still small but nonetheless already comprises several thousands of leaves. François Sillion now hopes to admire soon a plant a dozen years old on his screen.
