The news that Warner could not distribute the SMPTE DCP version of the HFR 3D version of The Hobbit because of problems in the field should raise eyebrows. The DCP was in development within SMPTE for over 5 years before the first standards were published in 2006. ISDCF spent much of the past 2 years testing products for SMPTE DCP compliance in its Plugfests. DCI has been updating its specification since the release of version 1.0 in 2005, and more recently has been busy approving the compliance of systems. But none of these organizations blew the whistle ahead of time. Whether or not the problem points to a systemic issue, or whether this is a fluke, will be examined.
The error found with Hobbit is not addressed by SMPTE, DCI, or ISDCF. It was found that several products could not route audio properly. Without audio routing in the media block, certain channels of SMPTE DCP 7.1 audio will appear in the wrong speakers, an untenable situation. Where the ball was dropped is that it is not widely acknowledged that the use of SMPTE DCP requires the media block to route audio for the product to have utility.
The root of the problem begins with DCI. DCI has been aloof about audio throughout the development and updating of its specification, and it carries though into its testing program. Even though DCI recently updated its specification, incorporating around 100 errata, the audio section, Section 3.3, continues to reference an obsolete SMPTE specification for audio channel assignment.
Audio has been so ignored by the entire digital cinema documentation process that one has to piece together how it works by reading several documents managed by different organizations. Interop audio was at one time specified in a document tucked away in a privately held password-protected FTP site, where the entire Interop specification remains today, at ftp://ftp.digicine.com/. (Write if you have reason to obtain the password.) Interop DCP audio maintains a one-to-one relationship between the physical channel that the audio channel occupies, and the output channel assignment of the media block. In other words, if a particular audio track is placed in channel #1 of the DCP, then the media block will output that audio track at audio output #1. The audio track file can contain up to 16 channels, and in time, management of those channels required an organization to oversee it. ISDCF stepped in, producing the audio channel assignment document that now exists on the ISDCF web site, at http://isdcf.com/papers/ISDCF-Doc4-Interop-audio-channel-recommendations.pdf. This document describes not only how Interop audio is to be packaged, but by extrapolation, it also describes how cinemas are wired, due to the one-to-one relationship between channel assignment and audio output.
SMPTE DCP, however, purposely eliminated the one-to-one relationship between channel assignment and audio output, to allow more flexibility in the assignment of the 16 channels that are available. Flexibility comes with a price, however. Audio channel routing must take place in the media block to properly mate with the fixed wiring in the installation. The requirement for routing, however, is not stated in any document from SMPTE, DCI, or ISDCF.
When confronted with this missing link in the documentation system, DCI clams up. It was pointed out earlier this year that DCI compliance is not enough to ensure utility in the cinema, pointing to the audio routing issue. The response varied from “DCI absolutely is enough” to “it’s not our problem.” (I.e., it is the exhibitor’s problem.) To be fair, it was not Warner that had its nose in the air, but perhaps Warner now has a different light to shine on the subject for its colleagues.
ISDCF alone has taken steps to discover problems ahead of the transition to SMPTE DCP. It is aware of the audio routing problem, but only mildly so. After several years of Plugfest evaluations, there is no public document from ISDCF that says “this manufacturer got it right, and this manufacturer needs to improve.” ISDCF is skittish about its public statements, referring to specifications as recommendations, and writing introductory text that makes it appear that the right way to do things is optional. So while ISDCF is in the right position to blow whistles and cause change to happen, its light touch in these matters doesn’t work with every manufacturer. Notably, manufacturers that are publicly traded companies tend to want hard documentation that points to the right way to do something. If hard documentation doesn’t exist to solve a problem that is not well understood, the action taken will invariably be no action. It is no coincidence that three publicly traded manufacturers of digital cinema products missed the boat when it came to audio routing for The Hobbit.
It wouldn’t be fair to point a finger at ISDCF, however, as it’s the only entity trying to do something. SMPTE could have done better by providing tips about routing in its specifications, and DCI most certainly needs to get its act in order when it comes to audio. With the right backup, ISDCF would be encouraged to be stronger in this area.
The conclusion of this report is that it is no fluke that audio routing caused the SMPTE DCP version of The Hobbit to fail in cinemas. The problem is systemic, and it needs to be fixed.