We propose a novel method to obtain a 3D model of a tree based on a single input image by fitting parameters for some procedural plant generator. Unlike other methods, our approach can work with any plant generator, treating it as a black-box function. It is also possible to specify the desired characteristics of the plant, such as the geometric complexity of the model or its size. We propose a similarity function between the given image and generated model, that better catches the significant differences between tree shapes. To find the appropriate parameter set, we use a specific variant of a genetic algorithm designed for this purpose to maximize similarity function. This approach can greatly simplify the artist's work. We demonstrate the results of our algorithm with several procedural generators, from a very simple to a fairly advanced one.

Modeling and rendering large scenes with thousands of plants is still a challenging problem. Geometric models of individual plants consist of millions of triangles each and their complexity must be reduced in advance to make real-time rendering possible. Existing solutions usually implemented as a part of plants generator, make an ecosystem simulator an indivisible all-in-one solution which is hard to modify and integrate. The proposed algorithm performs approximate instancing over a set of plants represented with a specific structure. Groups of structurally similar branches are replaced with instances of one of them during the clustering process. Also, a new fast and universal procedural plants generation method is proposed. This algorithm collects statistics of spatial distribution of branches in the original set of plants and creates new plants trying to imitate parameters from original ones using instances of existing branches. Our generator is able to amplify the amount of plants in the ecosystem with small time and memory overhead. Unlike most existing algorithms the whole process is independent from the original plants generator in our solution.