MAgArRO Home Page

Summary

Rendering is the process of generating a 2D image from the abstract description of a 3D scene. Despite the discovering of new techniques and algorithms to computational requirements of photo-realistic generated images are such that it is not possible to render them in real time. In addition, the configuration of good render quality parameters is very difficult to be done by non-expert users, and usually are set higher than in fact are needed.

MAgArRO is an approacth to optimize the rendering process in a distributed, non-centralised way by the use of a multi-agent solution that models expert knowledge to achieve local optimizations of rendering variables.

Video Demo

The following flash video shows a demostration of MAgArRO rendering a scene with 12 Agents. In this video de Web interface is used to show the progress of the agents, their internal state and the fuzzy variables used to render each zone. A high-resolution version of the same video (1024x768) is available to download: VideoDemoMagarro.mpeg (MPEG Codec - 17MB)

Get the Flash Player to see this player.

Source Code

The full source code can be downloaded from:

Installation considerations

This prototype has been testing in Debian GNU/Linux Sid, but should run on other operating systems with the following library and tools.

  • ZeroC ICE 3.1.0 or later: To install ZeroC ICE in Debian GNU/Linux, you only need to install de zeroc-ice metapackage. If you use another GNU/Linux distribution, you should visit http://www.zeroc.com/download.html.
  • Apache 2.2 + PHP5: In order to use the remote GUI interface, you need to configure any web server (we've used Apache 2.2.9 successfully) and PHP (with ZeroC ICE support; in Debian distributions it's included on zeroc-ice metapackage.
  • LibGD 5 or later: module for handling graphics directly from PHP scripts.
  • Python Imaging Library (PIL) 1.1 or later
  • Blender 3D Suite 2.41 or later
  • Yafray render engine 0.0.8 or later

Please, read the "readme" file of each main directory for configuration details (variables, config files and relative paths).

Related Publications

Journals

[·] Glez-Morcillo C., Weiss G., Vallejo D., Castro-Schez J. J., Jimenez L. A Multi-Agent Architecture for 3D Rendering Optimisation, Applied Artificial Intelligence, 2010. DOI link.

Conferences

[·] Glez-Morcillo C., Weiss G., Jiménez L., Vallejo D., A MultiAgent Approach to Distributed Rendering Optimization, Innovative Applications of Artificial Intelligence Conference (IAAI 2007)
[·] Glez-Morcillo C., Weiss G., Jiménez L., Vallejo D., Albusac J.A., A MultiAgent System for Physically based Rendering Optimization, Cooperative Information Agents CIA 2007 [Lecture Notes in Artificial Intelligence] [System Innovation Award 2007]

Contact Us

Carlos González Morcillo
Carlos.Gonzalez@uclm.es
University of Castilla-La Mancha
Spain
Gerhard Weiss
Gerhard.Weiss@scch.at
Software Competence Center Hagenberg
Austria
Luis Jiménez Linares
Luis.Jimenez@uclm.es
University of Castilla-La Mancha
Spain
David Vallejo Fernández
David.Vallejo@uclm.es
University of Castilla-La Mancha
Spain
José Jesús Castro Sánchez
JoseJesus.Castro@uclm.es
University of Castilla-La Mancha
Spain
Javier Alonso Albusac Jiménez
JavierAlonso.Albusac@uclm.es
University of Castilla-La Mancha
Spain

Acknowledgements

This work has been funded by the Junta de Comunidades de Castilla-La Mancha under Research Project "E-Pactos" (reference PAC-06-0141) and "Sarasvati" (reference PBC06-0064-4504). Special thanks to Javier Galán for his indoor scene used in this fork for testing the system and to Javier Ayllón for his superb support at the Supercomputation Service of the University of Castilla-La Mancha.