“ILM Presents: Making “The Irishman”” by Helman, Estebecorena and Grabli

  • ©Pablo Helman, Leandro Estebecorena, and Stephane Grabli

Conference:


Type:


Title:

    ILM Presents: Making “The Irishman”

Presenter(s)/Author(s):


Project Affiliation:


    Industrial Light & Magic

Abstract:


    This production session will describe the schedule, production, and tech aspects of The Irishman movie. From the very beginning (including some early “markerless” capture work done as early as in 2015) till the end of production showing before and after examples.
    We’ll go over the requirements initially imposed by Scorsese and the actors, who wanted to perform on the set as they always do to deliver their best possible performances while interacting with each other, which made additional performance capture after the shoot definitively out of the question.
    And also the requirements from the DP (Rodrigo Prieto) who wanted total freedom to light, (so we could not control the lighting at the moment of the capture of the performances.)
    To fulfill those demands, we had to come up with a new way of capturing the performances (and lighting), with more precision than ever before, yet in a way that minimizes any interference with all the departments during the shoot.
    This presentation will go over the process of developing a novel markerless method of facial capture, involving the new software developed for this project, and also a new infrared camera rig configuration. And why we departed from the “safe” approach of multiple facial markers (as well as discarding head-mounted cameras), which would not suffice for the level of accuracy (and minimal burden on the actors and crew) required for this project.
    Other aspects that will be discussed:
    • Adapting the pipeline for streaming projects, (with shifting and creating new discipline types to account for further work to be done)
    • Rendering iconic actors entirely in 3D
    • The unprecedented precision and level of subtlety that Mar- tin Scorsese demanded in the rendered images (of these highly recognizable characters for hundred of shots).


PDF:



ACM Digital Library Publication:



Overview Page: