The Crowd Simulation: Creating Variety Within Crowds

DRIDI, Imane (2026) The Crowd Simulation: Creating Variety Within Crowds. Doctoral thesis, Université Mohamed Khider (Biskra - Algérie).

Text Last Version Dridi imane_Thesis_ 2026 (1).pdf Download (4MB)

Abstract

In our daily lives, we interact and meet with many people performing various activities that effortlessly blend into their daily routines. Each individual has a unique appearance and shape and performs specific tasks with different actions. We notice variations in their movement, including differences in walking and running patterns and varying velocities they select for motion. Additionally, we can observe how they navigate around each other and avoid obstacles they may encounter. In this context, we simulate a crowd, focusing on the challenge of representing individual similarities and achieving diversity to execute a simulation realistically and accurately. The objective is to focus on crowd variety while ensuring each character has individual characteristics. The proposed methodology combines two methodologies. The first method focuses on appearance variation, including shape, accessories, color, and texture. The second method uses different types of animations. Applying a locomotion cycle allows for the creation of various animations for each model, using both the animator controller and script. Collision avoidance strategies are used between characters, with a proposed algorithm based on Raycasting, Newton's principles, and the conservation of momentum and kinetic energy in the absence of external forces. This algorithm produces efficient results for determining the exact final velocity values for the characters. The proposed crowd system is simulated under different lighting conditions to render our characters and create visual realism. It includes combining all of these terms to enhance crowd realism in terms of visual quality.

Actions (login required)

View Item