Post-graduate theses
Search command : Author="Τέρτικας"
And Author="Αχιλλέας"
Current Record: 2188 of 6624
|
|
| Identifier |
000381057 |
| Title |
Physically-principled character lighting and shading in real-time |
| Alternative Title |
Φωτισμός και σκίαση εικονικών χαρακτήρων σε πραγματικό χρόνο και βάση κανόνων φυσικής |
|
Author
|
Παπανικολάου, Πέτρος
|
|
Thesis advisor
|
Παπαγιαννάκης, Γεώργιος
|
| Abstract |
In many applications, user interacts in a 3D virtual world rendered with physically-based light. In order to enhance user's experience, the graphics community is continuously working on enriching these worlds with realistic depiction of light interaction with objects. The controversy between accuracy and rendering speed is always constant, especially for dynamic surfaces in order to achieve at least 25 frames per second.
In this thesis, we analyze the physically-principled lighting and shading effects that occur on human skin in an area and point light illuminated 3D space. Some of these effects are the following: Ambient Occlusion, Shadow Mapping, Image-Based Lighting, Specular Surface Reflectance, Subsurface Scattering and Environment Mapping. There are systems that support a combination of them, but only in isolated character parts (e.g. human head) in non-hierarchical and non-animated surfaces. The problems that we are tackling are not only limited on how to apply these effects on a fully, skeleton-based, deformable and animated virtual character, but also on how to calculate the most robust physically-based approximations (hence physically-principled) in order to achieve high visual accuracy along with efficient real-time execution in modern commodity graphics h/w.
We have thus created a high-fidelity, physically-principled real-time rendering framework for articulated characters. It features a series of effects that can be applied on the human model either simultaneously or separated. The unification was implemented in screen space (as opposed to standard object-space methods), achieving that way both scalability and efficient real-time execution. Multiple rendering fragment shader passes are created and handled within our system. The output image rendered by any pass can be used as input to the subsequent pass that follows ensuring fast and efficient reusability between the common and essential data for the different effects. For instance, our real-time subsurface scattering method is achieved in just two rendering passes, applying horizontal and vertical convolutions using an intermediate texture as output to the first pass and input to the next. The implementation of our rendering framework is based on widely available open-source tools and techniques such as OpenGL, C++ and GLSL so that it can be easily integrated in modern rendering engines and scene graphs via commodity graphics h/w.
|
| Language |
Greek, English |
| Subject |
Virtual characters |
|
Εικονικά χαρακτήρες |
| Issue date |
2013-07-19 |
| Collection
|
School/Department--School of Sciences and Engineering--Department of Computer Science--Post-graduate theses
|
|
|
Type of Work--Post-graduate theses
|
| Permanent Link |
https://elocus.lib.uoc.gr//dlib/4/9/2/metadata-dlib-1383904715-33796-9618.tkl
|
| Views |
503 |
Collections
