Post produced feature film and episodic content for the home is typically delivered to distribution channels in a mezzanine format such as Apple ProRes or Interoperable Master Format (IMF). The tools and methods for editing mezzanine files to add logos, dub cards, or other picture inserts are well known, but extra care must be taken to handle Dolby Vision metadata correctly when combining Dolby Vision assets. Conceptually, it is helpful to think about editing Dolby Vision assets as a two-step process. The first step is to normalize the Dolby Vision metadata across all assets, and the second step is to actually perform the necessary edits.  

 

Step 1: Dolby Vision Metadata Uniformity 

When combining multiple Dolby Vision assets into a single Dolby Vision master, the following video and metadata (Dolby Vision XML) parameters must be the same across all assets: 

• Dolby Vision Metadata Version: CMv2.9 (v2.0.5) or CMv4.0 (v4.0.2, v5.1.0) 
• Mastering Display 
• Frame Rate 
• Aspect Ratio (Canvas) 
• All Color Encoding Details (Primaries, White Point, Color Space, Signal Range) 

It is possible to combine assets with different aspect ratios into a single Dolby Vision master provided that the overall image resolution, aka the “canvas,” is the same for all assets.  

 

Before beginning the process of asset assembly, make sure the Dolby Vision metadata is normalized across all assets to be combined and that the Dolby Vision metadata accurately reflects the true frame rate and color encoding properties of the corresponding video essence. It is also necessary to normalize the video properties between the various assets, but that process should already be familiar to most mastering engineers and is outside the scope of this document.  

 

 

Dolby Vision Metadata Version 

If the assets do not all share a common Dolby Vision metadata version, any asset with CMv4.0 (v4.0.2, v5.1.0) metadata can be downgraded to CMv2.9 (v2.0.5) using Metafier:
 

 
Refer to: Dolby Vision Professional Tools User Manual

It is not possible to upgrade CMv2.9 metadata to CMv4.0; only version downgrades are possible at this time. If the finished master must be delivered in CMv4.0, the segment or segments with v2.0.5 metadata must be re-analyzed and trimmed with Dolby Vision CMv4.0 to generate v4.0.2 or v5.1.0 metadata. The process of re-analyzing and trimming should be supervised by an experienced Dolby Vision colorist.  

 

Mastering Display 

Changing the Dolby Vision mastering display changes the way the content is mapped by the Dolby Vision algorithm. If making such a change is absolutely necessary to allow for concatenating Dolby Vision assets, it’s generally preferable to alter the mastering display properties of short elements such as logos or picture inserts so that they match the mastering display properties of the main program material. This approach leaves the Dolby Vision mastering display – and therefore the Dolby Vision mapping – of the main program unchanged.
 

Refer to: Mastering and Target Displays
 

The Dolby Vision mapping should be visually checked following any mastering display edits to make sure the original artistic intent isn’t significantly altered. 

 

Mastering display changes can be made using Metafier from the Dolby Vision Professional Tools. In this example, the mastering display is being changed to 1000 nits with Rec. 2020 color primaries (mastering display ID 21): 

 

 

 

Step 2: Dolby Vision Asset Assembly 

A Dolby Vision master can be assembled from multiple assets either by concatenating the separate assets into a single, larger mezzanine file or by joining the assets by means of a composition playlist (CPL) in the Interoperable Master Format (IMF). Preparing an IMF package with multiple video segments rather than concatenating multiple assets into a larger file allows the individual assets to be maintained as separate, modular components that can be easily swapped out with just a playlist update and supplemental video file.

 

Note: 

Most studios and distribution platforms still require a single larger concatenated video, so it won’t always be possible to deliver a multi-segment IMF package to every distribution partner, as such it is important to contact and verify the approved delivery requirement with the distribution partner.

Depending on the video properties and JPEG 2000 encoding parameters used, IMF video masters can be very large, on the order of terabytes. For this reason, it may be beneficial to keep logos and inserts separate rather than rewriting most of the very large video file for the sake of a relatively minor change. Again, this may not be an option depending on the intended recipient of the mezzanine. Please refer to your specific customer’s delivery requirements.

Option 1: Concatenation of Assets 

Assets are combined into larger ProRes or MXF files that contain all program material in a single file wrapper. Many third-party mastering systems are able to handle this type of assembly on a timeline while easily keeping the Dolby Vision metadata in sync with its corresponding video.  

The Mezzinator tool from the Dolby Vision Professional Tools can also handle asset concatenation, but some care must be taken to maintain synchronization between video and metadata. In this example, Mezzinator is combining Dolby Vision metadata and video segments from multiple sources into a single longplay asset. The metadata edits are described in a user-generated “reel” file, while the video edits are described in a user-generated “playlist” file. The tool executes the edits it finds in the reel and playlist files and outputs the single concatenated asset.  

 

 

 

Refer to Appendix B of the Dolby Vision Professional Tools User Manual for complete instructions on reel file creation. Refer to Appendix C for complete instructions on playlist file creation.  

Option 2: Assembly via IMF CPL 

Segments are kept as separately wrapped essence files – each with its own Dolby Vision metadata – and linked together by a composition playlist (CPL), which can be used by some encoding tools to create a single longplay output stream. The CPL is a human-readable XML file that is often described as the “recipe” for a media deliverable. The audio and video assets listed in each segment of the CPL are the “ingredients” necessary for preparing the recipe.  Leaving logos or inserts as separate components allows for maximum flexibility. Because the original main program video and metadata remain unchanged, the QC burden can potentially be reduced. 

 

Asset Assembly Tools 

Please refer to this list for tools supporting Dolby Vision XML import and asset assembly.

 

Assembly at Encoding 

Some over-the-top (OTT) distribution platforms and Blu-ray authoring facilities are equipped to encode from multiple video and metadata segments, whether those are supplied as an IMF package or in another form. In these cases, delivering a pre-assembled Dolby Vision master may not be necessary.  

 

Assembly at Playback 

Some OTT platforms don’t assemble video segments at all, but rather encode the segments as separate pieces that are streamed to consumer playback devices in the correct sequence.