Generative design and automatic optimization are techniques used in architecture and manufacturing industries. It is an iterative design process in which a parametric model is created, which is then analyzed and given a success score. The analysis is then used to drive the next generation of permutations.

Room design was chosen as the subject of the test case, in which a defined set of rules controls the generative design process. A proprietary implementation of the MakeItHome: Automatic Optimization of Furniture Arrangements by Lap-fai yu et al. was used in Houdini.

The database of 3D models has a collection of predefined parameters that influence the placement and relationships between objects. Each object is placed in a random position. Based on the parameters, the objects search for relationships. For example, a table and a chair will have a parent-child relationship. Movements in position and orientation are applied to the objects through an iterative process. Each iteration is scored based on how well the objects stand. With an improvement, the score is used as the basis of the new iteration until an ideal score is obtained.

Generative design is a very useful way to explore programmatic and automatic design and produce an instant overview of measurable scores. Automatic optimization requires a lot of computational power and time to obtain a usable room layout. For complex rooms with many objects, that can easily go up to 20,000 iterations. Scoring is the most intensive process. Without a scoring system, this system can also be used for real-time generation.