Publicado em: 06/07/2013

Proposta de Tese em Computação Gráfica

UNIVERSIDADE FEDERAL DO RIO GRANDE DO SUL
INSTITUTO DE INFORMÁTICA
PROGRAMA DE PÓS-GRADUAÇÃO EM COMPUTAÇÃO

DEFESA DE PROPOSTA DE TESE

Aluna: Marilena Maule
Orientador: Prof. Dr. João Luiz Dihl Comba

Título: Order-Independent Transparency Rendering Technique in Bounded Memory for High-Quality Images
Linha de Pesquisa: Computação Gráfica

Data: 12/07/2013
Horário: 13h30min
Local: Sala Sala 215 (videoconferência). Prédio 43412 – Instituto de Informática

Banca Examinadora:
Prof. Dr. Esteban Walter Gonzalez Clua (UFF)
Prof. Dr. Manuel Menezes de Oliveira Neto (UFRGS)
Prof. Dr. Marcelo Walter (UFRGS)

Presidente da Banca:Prof. Dr. João Luiz Dihl Comba

Resumo:

This research consists in developing a practical order-independent transparency rendering algorithm. The algorithm must be prone to be implemented in hardware and used as standard for transparency rendering as the z-buffer is used for opaque rendering. This goal implies an algorithm running in bounded memory, because the hardware has physical limitations. For leveraging the GPUs resources and acquire practical performance, data-level-parallel-friendly approaches will be adopted. The quality must be adequate, meaning no perceptible artifacts and maintenance of the spatial relationship among layers.Transparency is a desired effect in a large set of applications, such as visualization tools (to denote relationship among structures), computer games, animations and simulations, which need transparency to render special effects such as particles (e.g. fire, smoke and rain), foliage and hair. Though the z-buffer can satisfactorily handle opaque primitives, efficient rendering of transparency remains an open problem. The core bottleneck is the depth-sorted order required to correctly evaluate visibility through an unknown number of layers with different opacities. This work presents the order-independent transparency problem, a survey of algorithms proposed in the literature, the author’s proposed techniques, future research paths and goals.