This project creates shadow effects with Houdini Digital Assets (HDA). The aim is to generate shadow geometries on given models and eventually consume it entirely. Shadow geometries are created by the first HDA, users could select different patterns and manipulate the parameters to create a variety of looks. The second HDA generates particles that create the consuming effect on the given model and swirl decomposing effects after consumption.
A realtime procedural creature generation system built in Unity. The aim was to create a more open ended generation system than had existed previously. A base body is created using free form deformation, it is rigged and skinned automatically. Limbs are then attached to create one of three creature types swimming, flying or walking. Animations are all driven procedurally and adapt to the creature generated.
This work demonstrates the usage of voronoi fracture extrusions for creating a procedural city blocks. The user is able to control 4 stages of the creation the flora such as trees, bushes, waters like rivers), buildings (city centre, residential area, administrative buildings) and roads with pathways.
This project presents a modular rigging pipeline that enables user to create rigs using either Python API directly or via node-based interface. The system aims to be extensible, allowing rigger to add new custom rig components or extend already existing ones. In order to be tool-oriented this pipeline utilizes metadata concept and component to Python class bindings which leads to consistent functionality for any additional tools.
SDF raymarching algorithms have been typically slow for real-time use until today. And no matter how optimized and efficient they are, real-time and interactive usage of these algorithms, once the CPU-GPU communication occurs, can be quite resource-intensive and expensive, which may lead to low performance and stalling the application. This thesis describes a Vulkan-based implementation for SDF Raymarching in real-time aiming to provide a proof-of-concept, as a mean for creating efficient solutions that simplify runtime evaluation of complex data through the GPU with a low-overhead CPU to GPU abstraction.