Rendering of Scene Data in Digital Cinema Workflows

Abstract

Several years ago, the Academy of Motion Picture Arts and Sciences formed a committee to study the use of a scene-referred image format for image encoding. The goal was to offer an alternative to the widely used printing density encoding usually stored in the DPX file format. There were many factors motivating people to examine a scene-referred encoding, including making compositing real images with computer graphics easier to accomplish, fulfilling a desire for increased dynamic range to facilitate higher quality manipulation of images in post-production, and adopting an encoding that is better suited for material originated on digital cameras. Since the inception of this committee, Kodak has been an active supporter of the authors' participation, allowing significant contributions to the underlying design considerations for workflows using scene encoding. One of the central elements of the Academy's workflow is the concept of a “reference rendering” transform that takes scene image data and prepares it for the anticipated digital cinema viewing environment. The concept of a reference rendering is controversial because the final “look” of most movie and television productions is often considered to be central to the creative aspect of each production. But it will be shown that defining a reference rendering transform can offer several advantages to creating a scene-referred image encoding. This paper discusses the basic premise behind a “reference rendering transform” and explores some of the technical considerations involved in designing such a transform.