By Ken Kiers, EVP Postproduction Services, My Eye Media –
The Interoperable Master Format is an international file-based interchange of multi-version finished audio visual works. SMPTE (Society of Motion Picture Technology Engineers) defines the basic constraints in the SMPTE 2067-2 (Core Constraints) specification. The document defines parameters for video and audio essences as well as the CPL (Composition Playlist), timed text and captions and basic descriptive metadata.
In its most basic form, the components consist of a video essence and an audio essence which are wrapped in an MXF (Material eXchange Format) container. The IMF is driven by the CPL which is an XML (eXtensible Markup Language) metadata document that informs the system of how to put together the audio and video essences to play back a composition. An asset map is an XML document that describes how the assets are distributed across the medium.
The packing list is another XML document that lists all of the files included in the IMP. By using tried and true technologies like MXF, XML, and TTML (Timed Text Markup Language) which have long been a part of other applications, IMF serves to improve overall compatibility within many systems.
If we dig a little deeper than the Core Constraints SMPTE 2067- 2, we find that there are developing applications that define in more detail the image characteristics, resolutions, frame rates and color gamut that are specific to a workflow. Application 2 defines the HD resolution 1080×1920 and JPEG 2000 code stream.
This application is often referred to as studio profile in that it allows for TV and Feature titles to be transformed into multiple versions with multiple dubbed languages, edited versions and special versions. A single IMF can be used as a distribution master across many territories for use on OTT, broadcast or physical media.
Application 2e (Extended) allows for Wide Color Gamut, High Dynamic Range (HDR) using SMPTE 2084 Electro Optical Transfer Function based on PQ (Perceptual Quantization) with a frame size up to 4K resolution. Other applications in progress are Application 3 which allows for MPEG-4 video encoding and linear PCM audio. Application 4 Cinema Mezzanine focuses on creating a framework for preserving / archiving finished source material after post production either sourced from film or from digital elements. Encoding the images at 16-bit with XYZ color primaries and frame size up to 8192 x 6244 (8K).
The QC of IMF
An integral part of any mezzanine level production across all of these developing frameworks is quality control. In an IMF workflow, the QC vendor is uniquely positioned to analyze the individual assets and confirm their integrity by using a combination of automated tools and experienced operators to assemble these assets into the final package. Reviewing materials as they are collected to build a package is the most efficient way to ensure the highest quality assets are used.
Once the main video essence is reviewed and any fixes are made, it is fully encoded. From this point, only the differences from the main video track, such as language inserts or censored shots, will be encoded. While this provides unrivaled efficiency, it increases the possibility for potential issues. Each of these additional elements must be reviewed by the QC vendor for any sizing or color discrepancies. The inserts will be wrapped in MXF just as the main video track and will be implemented in a CPL version.
Thorough reviews of the dubbed languages are needed to determine technical merit like overall loudness of a program (ATSC 85 or EBU-R128), audio peaks and sync, as well as to determine if the audio matches the original version or requires edits. In many cases the international versions of dubbed tracks can have edits that are not documented. Only by comparing tracks against a common video source can some of these edits come to light. Here, again, it is the QC vendor who is best positioned to perform the analysis and recreate the versions as additional Composition Playlists.
Subtitles and captions must receive the same type of technical and quality review. The wide variety of text file types and formats make it difficult to determine timing, positioning and translation quality and the review must be performed by a highly-trained QC operator as the text files must often be converted and re-timed to match picture. IMF looks to solve some of these problems by using ISMC1 – a worldwide subtitle and caption format that is part of TTML which helps to ensure compliance across platforms.
Once all the assets have been reviewed, retimed, edited and aligned on their respective timelines, the QC vendor can move forward to creating a viable IMF package using the approved materials. The same segments of essences can be used in multiple CPLs.
This is where the IMF format is incredibly efficient. An edited or international version can be created via a new CPL by only encoding the video & audio differences from the original version. With the ability to compose numerous versions from a single IMP it can be delivered to many different distributors that can playout the version they need by choosing the corresponding CPL.
More solutions for multiple deliverables
Another IMF implementation that aims to solve the request for multiple deliverables is the OPL (Output Profile List). It contains information on how to transform the composition into a specific deliverable such as an iTunes Package and provides macros for image sizing (pan and scan) as well as audio mixing and routing. An extension of the OPL currently in development – the Semi-Automated Quality Control component — combines two types of automated QC approaches. The first involves comparing the quality of the output to the original essence. The other performs checks for typical audio and video issues such as freeze frames, audio dropouts, black frames and more.
The primary focus of these tools is on transcoded artifacts that are most relevant to human perception. A QC operator can then interpret the automated results and quickly make critical decisions as to whether an artifact is acceptable or requires a new/corrected asset. While these automated QC tools are imperative to increased speed and efficiency of the workflow, they put additional demands on QC vendors to maintain a deep and broad technical expertise.
The growth of worldwide distribution has resulted in an increased demand for multiple localized versions, and IMF provides the most elegant and efficient solution for addressing this demand. As IMF expands its reach by adding more open source tools, it will lower the cost of entry and facilitate higher levels of interoperability. It is vital that all the essences contained within the package are thoroughly reviewed to ensure compatibility, accuracy and quality.
Analyzing the individual assets and confirming their integrity by using a combination of automated tools and experienced operators is a key aspect of an efficient IMF workflow and it is incumbent upon QC vendors to maintain a firm grasp on the ever-changing landscape of standards, tooling, and practices. After having vetted the content, an experienced QC vendor can then create a viable IMF package using approved materials, saving valuable hours and minimizing file deliveries.
As we look forward, additional efficiencies will be recognized with IMF Applications that will allow for the Sidecar Composition Map, which will enable metadata not associated with essences – such as a QC Report — to be related to a composition. Allowing QC vendors to embed this report will provide confidence that the assets being used are accurate and maintain a high level of quality throughout the ecosystem.