This work presents the development of a procedural 2D character generation system, focusing on the application of genetic algorithms to simulate heredity and evolutionary variation within an interactive visual environment. The system enables the creation of characters composed of modular sprites with inheritable visual and functional attributes. It supports the use of various parent selection methods (roulette, tournament, and ranking) and crossover strategies (single-point, two-point, and scattered). The tool was implemented in the Unity engine, with result analysis conducted using Python on the Kaggle platform. Nine simultaneous populations were tested using different algorithm and environment combinations, totaling 450,000 individuals. Logged data enabled performance, convergence, and genetic diversity evaluations. Results show that the crossover method significantly affects convergence speed, and tournament selection tends to yield higher average fitness. The system also includes a visual playback feature to reconstruct characters from logs, making analysis more intuitive and didactic. This proposal integrates learning, visualization, and comparative analysis, contributing to the study of genetic algorithms in game-related procedural content generation.