DVD Subpictures vs EIA-608 Closed Captions in VOB

Inside an MPEG-2 Program Stream Video Object (VOB) file, both DVD subpictures and EIA-608 closed captions provide synchronized on-screen text, but they differ fundamentally in their data structure, storage locations, rendering mechanics, and intended purposes. DVD subpictures are pre-rendered bitmap overlays carried within dedicated private elementary streams, allowing rich typography, multilingual support, and DVD menu highlights. Conversely, EIA-608 captions are lightweight, byte-encoded text and control commands embedded directly into the MPEG-2 video user data, designed primarily to meet North American broadcast accessibility standards.

Data Format: Bitmaps vs. Text Data

The primary technical difference lies in how visual information is represented:

• DVD Subpictures: Subpictures are graphical overlays rather than plain text. They consist of 2-bit run-length encoded (RLE) pixel maps. Each pixel can reference one of four colors from a 16-color Color Look-Up Table (CLUT) defined in the DVD's .IFO navigation files. Because they are images, subpictures can display any font, glyph, script (such as East Asian or Arabic typography), or graphic icon without requiring font support in the playback hardware.
• EIA-608 Captions: EIA-608 (also known as Line 21 captions) is an ASCII-based data stream consisting of two-byte words transmitted at a fixed rate (historically 60 fields per second). It carries no graphic raster data; instead, it transmits raw character codes and simple control sequences denoting screen placement, basic colors, italics, and roll-up or pop-on display modes.

Storage and Multiplexing Inside the VOB

Both formats share the VOB container, but they inhabit entirely different multiplex layers:

• DVD Subpictures: Subpictures are stored as independent MPEG private streams (Private Stream 1, standard packet identifier 0x1BD). Up to 32 distinct subpicture streams (sub-stream IDs 0x20 through 0x3F) can coexist in a single VOB, permitting multiple language selections, commentary tracks, and interactive menu button highlights.
• EIA-608 Captions: EIA-608 data is not given its own stream ID. Instead, it is embedded directly into the elementary video stream (0x1E0) as user data packets. Specifically, caption bytes are packed inside the MPEG-2 Picture User Data headers (marked by start code 0x000001B2) on a per-frame or per-field basis, following either ATSC or DVD-Video syntax extensions.

Rendering and Processing Pipeline

The rendering mechanism determines how the viewer sees the captions:

• DVD Subpictures: The decoding and rendering pipeline is fully handled by the DVD player (software or hardware). The player decodes the RLE packets, maps the 2-bit values to the RGB/YUV values from the IFO palette, applies contrast and transparency levels, and composites the final bitmap directly over the decoded video frames before video output.
• EIA-608 Captions: The DVD player traditionally does not render the text. In standard-definition analog setups, the player extracts the byte pairs from the MPEG user data and reconstitutes them onto Line 21 of the vertical blanking interval (VBI) of the analog NTSC signal. The connected television's internal decoder chip is then responsible for generating the on-screen font and rendering the captions. In modern digital playback or software players, the player itself must emulate the television's hardware decoder to render the text.

Styling and Layout Capabilities

Because of their different architectures, visual control varies significantly:

• DVD Subpictures: Authors have pixel-level control over positioning, cropping, and anti-aliasing. Subpictures support arbitrary screen positions, interactive highlights (used for selecting menu items), and synchronized animations via Sub-Picture Control Commands (SP_DCSQ).
• EIA-608 Captions: Formatting is strictly constrained by the 1970s-era NTSC standard. Text is restricted to a fixed 32-column by 15-row grid. Styling is limited to basic monospace character sets, standard pop-on, roll-up, or paint-on animations, and minimal color attributes without arbitrary font choices or anti-aliasing.