Grid Architecture for Distributed Rendering
Jose Antonio Fernández-Sorribes, Carlos González-Morcillo, Luis Jiménez
Abstract
Jose Antonio Fernández-Sorribes, Carlos González-Morcillo, Luis Jiménez
Abstract
Rendering is the process by means of which a raster 2D image can be obtained from the definition of a 3D scene. This process is computationally intensive and requires a lot of time to be done when the source scene has certain complexity or when high-quality realistic images are required. YAFRID (Yet Another Free Render grID) is a system that takes advantage of the characteristics of computational grids by distributing the rendering of a scene among a large number of heterogeneous computers connected to the Internet. With that kind of systems, the time a scene takes to be rendered is drastically decreased because the parts in which the complete work has been divided (called workunits) can be processed in parallel and finally joined using an interpolation function to obtain the image result. As we will discuss on the final section of this article, the selection of the workunit size is a key step in this class of work division. With this approach, local optimizations in each workunit could be made to obtain a better rendering time per frame.
OpenAlex reports 4 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
Rendering is the process by means of which a raster 2D image can be obtained from the definition of a 3D scene. This process is computationally intensive and requires a lot of time to be done when the source scene has certain complexity or when high-quality realistic images are required. YAFRID (Yet Another Free Render grID) is a system that takes advantage of the characteristics of computational grids by distributing the rendering of a scene among a large number of heterogeneous computers connected to the Internet. With that kind of systems, the time a scene takes to be rendered is drastically decreased because the parts in which the complete work has been divided (called workunits) can be processed in parallel and finally joined using an interpolation function to obtain the image result. As we will discuss on the final section of this article, the selection of the workunit size is a key step in this class of work division. With this approach, local optimizations in each workunit could be made to obtain a better rendering time per frame.
Key concepts: Computer science, Architecture, Rendering (computer graphics), Computer graphics (images), Grid, Distributed computing, Geology, Art