Multi-Camera Angles and Branching in VOB Files

Multi-camera angles and alternative scenes in DVD-Video VOB (Video Object) data are implemented through physical data interleaving and precise navigation metadata embedded directly within the video stream. Rather than storing entire alternate versions of a scene as separate, continuous video files, the DVD-Video standard slices concurrent video streams into small segments and weaves them together into a single sequential file structure. Optical players read this interleaved data using embedded Data Search Information (DSI) tables, allowing the player's laser mechanism to skip unwanted angles and read only the user-selected stream in real time without audio dropouts or playback stutter.

The Challenge of Optical Seek Times

Optical disc drives suffer from significant physical latency when the optical pickup unit moves across the disc surface. If alternative angles or scenes were stored in separate files or distinct disc sectors, switching cameras or branching into an alternate scene would require the laser to jump back and forth continuously to fetch corresponding audio and video. This latency would inevitably exhaust the player's memory buffer, leading to stuttering playback. To prevent this, the DVD specification dictates that all parallel visual streams must be physically co-located on the disc.

Interleaved Video Units (ILVU)

The structural foundation of multi-angle and seamless branching video is the Interleaved Video Unit (ILVU).

  1. Video Object Units (VOBU): At the fundamental level, all MPEG-2 video in a VOB is divided into VOBUs, which typically represent 0.4 to 1.0 seconds of presentation time (a Group of Pictures, or GOP) accompanied by the matching multiplexed audio and sub-picture packets.
  2. ILVU Grouping: When multiple angles or branching paths exist, adjacent VOBUs belonging to the same angle are grouped into an ILVU.
  3. Physical Alternation: The ILVUs for each alternate version are multiplexed chronologically. For example, a three-angle scene is physically laid out on the disc in alternating chunks:
    • [Angle 1 - ILVU A] -> [Angle 2 - ILVU A] -> [Angle 3 - ILVU A] -> [Angle 1 - ILVU B] -> [Angle 2 - ILVU B] -> [Angle 3 - ILVU B]

When a user watches Angle 1, the optical pickup reads [Angle 1 - ILVU A], skips over the sectors containing Angles 2 and 3, and lands directly at the beginning of [Angle 1 - ILVU B].

The player determines where to jump using the Navigation Pack (NV_PCK), which exists at the very beginning of every VOBU. The Navigation Pack contains two primary 980-byte control packets: Presentation Control Information (PCI) and Data Search Information (DSI).

The DSI packet contains the lookup tables that make angle switching possible:

Because the jump distance is strictly constrained by DVD authoring specifications—often capped at a specific maximum sector span depending on drive read speed—the track buffer on the player never fully empties during the hop.

Multi-Camera Angles vs. Alternate Scene Branching

While both features use interleaving, authoring specifications handle them with distinct constraints depending on the creative purpose: