Intelligent rendering of dailies: automation, layering and reuse of rendered assets

In visual effects production the amount of digital assets to be created is constantly increasing [2]. These assets can be two dimensional images, three dimensional geometries or animated scenes with multiple characters, large scenarios and complex shading. In order to properly review the creation process for all these contents, it is necessary to rapidly generate accurate previews that show the progress and enable supervisors to provide valuable feedback. Unfortunately, given the complex nature of these assets, it is difficult to manually generate such previews, with the quality required to provide meaningful feedback in time for daily reviews. From a technical standpoint, it is important to consider that MPC's pipeline is designed to work as a multi-layered reference tree of several assets of different types. Then, one asset is defined as a hierarchical structure of interconnected and interdependent assets [1, 3]. Because of this, it is important to ensure that, whatever version of an asset is rendered, all of its dependent assets (characters, animations, cameras, etc.) are properly sourced. In this paper we describe an automated system called \"RenderFlow\" that takes advantage of our digital asset management system to rapidly generate high quality renders of complex scenes by performing an automated gathering of required assets and reusing existing renders to save processing time. The system is integrated within our visual effects production pipeline and takes advantage of several commercially available products (e.g. Autodesk Maya and Pixar's RenderMan).