A Framework for Cluster-based Rendering and Postprocessing

Despite the ever increasing performance of computer hardware, single machines are not always capable of processing complex tasks in reasonable time, if at all. At the same time, the amount of data to be generated or processed by computer systems increases at a similar rate. Generating images using a...

Full description

Saved in:
Bibliographic Details
Published in:Workshop on Software Engineering and Architectures for Realtime Interactive Systems (Print) pp. 1 - 6
Main Authors: Frericks, Philipp, Roth, Thorsten, Hinkenjann, Andre, Kruijff, Ernst
Format: Conference Proceeding
Language:English
Published: IEEE 01.03.2017
Subjects:
Clustering algorithms
Complexity theory
I.3.2 [Computer Graphics]: Graphics Systems-Distributed/network graphics
Image resolution
Pipelines
Rendering (computer graphics)
Scheduling algorithms
Task analysis
ISSN:2328-7829
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Despite the ever increasing performance of computer hardware, single machines are not always capable of processing complex tasks in reasonable time, if at all. At the same time, the amount of data to be generated or processed by computer systems increases at a similar rate. Generating images using algorithms from computer graphics is a good example for such scenarios. Because visual analysis of data can greatly benefit from the usage of large, tiled display systems, and techniques to render images become more and more complex in order to generate visually pleasing results, using a single machine is often not sufficient to enable reasonable workflows.In this paper, we introduce a framework for cluster-based rendering and postprocessing. Distributing the generation and postprocessing of images to a cluster of machines allows to speed up these tasks. In particular, we aim at rendering images at high resolutions, e.g. for large display walls. To our knowledge, no other framework for distributed rendering is capable of postprocessing images in a distributed fashion without being limited to frame tiles.We present a software design that provides a simple interface for creating render sessions, enabling arbitrary applications to delegate the generation of images to a cluster. In addition to a simple configuration process, we intend to support platform interoperability. We focus on the design of a flexible and modular architecture in order to make the framework independent of rendering techniques and scheduling algorithms. Thus, these components should be interchangeable to enable a variety of use cases. The same applies to the integration of postprocessing methods.
ISSN:2328-7829
DOI:10.1109/SEARIS41720.2017.9183510