Introduction to Digital Intermediates
By Dominic Case, Atlab Australia

Definition
Originally, the term Digital Intermediate was coined to describe a process whereby digital technology was substituted for the established photochemical methods of film duplication – Interpositive and Duplicate Negative. This made the already well-developed tools of digital image manipulation - from advanced colour correction to morphing - as available to the filmmaker as they were in video postproduction and the graphic arts. Like the photochemical process of duplication, the DI process can be used for short sections of a production (for example to obtain a special “look” for a sequence) or, increasingly, for the entire production.

By transferring material from a range of sources (film, video, CGI, etc) to a common digital format, it is possible to mix and match to an unprecedented degree. As a result, the term Digital Intermediate tends now to be used in a wider sense: the source material doesn’t have to be film, and the final distribution format can be for DVDs and Digital Cinema as well as for traditional 35mm film prints.

However, it would be wrong to regard the Digital Intermediate as simply video technology extended to film. There are a number of fundamental differences that set the Digital Intermediate process apart from the broader term of Digital Postproduction.

Three Stages
First however, it is useful to define the process. Put simply there are three stages: input, output and “the middle bit”.

Input
Refers to the capture or ingestion process. Typically, this is scanning the negative in a film scanner or datacine and storing the image data in a digital file – one for each frame. Note the significant difference: data is stored as files, not as a stream of information like video. It is possible, however, to ingest other media (conventional video, HD, etc) into the same digital file format.

The middle bit
Is a generic term for all of the image manipulation that can be done on the digital files. Colour correction, with all the power that digital technology brings, is the most prominent. As well, dirt removal, image repair, wire removal (and other unwanted image elements), background replacement, crowd extension, morphing - the list is unlimited – can all be added to the basic set of titles and optical transitions that were commonplace in optical printing and duplication departments.

Additionally and importantly, the entire production can be conformed in the digital realm: physically cutting and splicing the original negative to the exact frame is unnecessary.

The volume of data involved is massive.

Output
is the converse of input: a film recorder is used to transfer the final digital files to a film negative, which takes the place of the “duplicate negative” of traditional processes (except that no final cut “original” negative exists except in a digital form). The digital files can also be output to suitable digital formats for Digital Cinema distribution, DVD mastering, and so on.
The lack of a physical final negative means that archival requirements must be given special consideration: there is, as yet, no long-term evidence that digital files can be preserved reliably.

3 fundamental rules
Different styles and genres of production have different requirements and different priorities, while different approaches to the Digital Intermediate process can have quite different budget implications. However, three rules underlie all true D I processes:

Preserve information
Film carries an astonishing amount of visual information: there is a wider tonal range and greater resolution in the original negative than will be displayed in almost any format – cinema, television, DVD or any other. The traditional approach on telecine has been to make selections – colour grading – and reduce the resolution to the output format, at the initial transfer session. The Digital Intermediate process ideally takes the opposite view: the entire tonal range is preserved at transfer, and the highest practicable resolution is used (typically 2K or 4K when coming from 35mm film), so that decisions later on, particularly in colour correction, are not compromised. In particular, data is stored uncompressed, or in a format with as close to lossless compression as possible. This naturally results in massive files and data storage and network speeds become major issues. A single frame scanned at 4K resolution (4096 pixels wide) is larger than 55Megabytes.

Nul transfer
The ideal test of a good film-to-digital-to-film process is that the final negative is capable of producing a print indistinguishable from a print from the original negative (except for the work done in “the middle bit” of the DI of course). Note that as with traditional film duplication, this is not the same as producing a negative that is identical to the original negative.

This means that the image transfer characteristics of the input and the output processes must be ideal in terms of mapping film values to digital values, and must be perfectly complementary to each other. The test of this is that if any set of digital tonal values (e.g. a grey scale) is generated and output to negative, and that negative is then scanned, it should produce exactly the same digital values. Maintaining this calibration extends to the chemical process of the negative exposed in the film recorder: the complete cycle is interlinked.

While any one facility may achieve this, extra care must be taken if any part of the work is being done elsewhere: output in one facility may not necessarily complement the input in another facility.

WISYWIG
Partly bound up with the previous principle is the idea of matching a digital display to the final film display. The exact colour of the red, green and blue primaries in various digital displays and the spectral characteristics of film emulsion dyes, all result in different colour gamuts – or the range of colours that can be reproduced. Much care and regular calibration is essential if the colorist is to see the same colours while grading that she or he will eventually see on the final film prints. Even with this calibration, it is common for short tests to be shot and processed on a production to gain confidence in the final result before committing compete rolls of a production to final film output.

What does it cost and where do I get it?
A full Digital Intermediate is, taken on its own, likely to cost a little more than straightforward photochemical duplication. However, it can offer cost savings in production (some time-consuming lighting issues can for example be addressed in the postproduction process) as well as in distribution (a digital master eliminates the need for the lengthy telecine transfer of the interpositive for video mastering). Furthermore, the DI provides far greater control over the look of the final product.

A number of ACVL member facilities offer the Digital Intermediate process. Their track record and experience in the style and format of your production are all useful indicators – as are their approach to the criteria described above.

Links to Additional Information About Digital Intermediates
from ACVL Members: