How Software Deinterlacers Process Interlaced VOB Video

This article explains how software deinterlacers process interlaced video streams extracted from DVD VOB (Video Object) files. It covers the initial demuxing and parsing of MPEG-2 data, the determination of field order, the mechanics of common deinterlacing algorithms, the distinction between true interlacing and telecine, and the final reconstruction of progressive frames.

1. Demuxing and Stream Parsing

A VOB file is a container format based on the MPEG-2 Program Stream specification. When video is extracted, the deinterlacer or host media framework (such as FFmpeg, HandBrake, or AviSynth) demuxes the container to isolate the raw MPEG-2 elementary video stream.

During decoding, the software reads picture headers to identify whether the video was encoded as field-structured pictures or frame-structured pictures with interlaced field coding. The stream typically consists of 480i (NTSC) or 576i (PAL) resolutions.

2. Identifying Field Order (Parity)

Interlaced video records motion in alternating lines: top fields (even lines) and bottom fields (odd lines), captured at different points in time. Before altering any pixels, the deinterlacer must determine the field cadence:

• Top Field First (TFF): The top field contains the earlier temporal moment.
• Bottom Field First (BFF): The bottom field contains the earlier temporal moment.

This information is typically flagged in the MPEG-2 bitstream metadata. If a deinterlacer misinterprets the field order, temporal sequencing is reversed, resulting in severe jitter and stuttering during playback.

3. Inverse Telecine (IVTC) Check

Before applying destructive deinterlacing, advanced software checks if the video is truly interlaced or merely telecined. Many NTSC VOB files contain 24-frame-per-second film converted to 29.97 fps via a 3:2 pulldown technique.

Software like IVTC filters detects repeating field patterns, drops redundant fields, and weaves the original progressive frames back together without quality loss. If the source is native video (such as recorded live television or camcorder footage), true deinterlacing is required.

4. Processing Fields via Deinterlacing Algorithms

For native interlaced content, each field represents a half-resolution image captured at distinct intervals. A software deinterlacer uses one of several mathematical approaches to eliminate combing artifacts ("mouse teeth") and produce progressive frames:

Basic Methods

• Weave: Combines adjacent top and bottom fields into a single frame. Static areas retain full vertical resolution, but moving areas display heavy interlacing artifacts.
• Blend: Merges adjacent fields together using a blend filter. Combing is eliminated, but the result suffers from ghosting and reduced sharpness.
• Bob (Line Doubling): Takes each individual field and scales it up to a full frame, often using linear or cubic interpolation. This doubles the frame rate (e.g., 29.97i becomes 59.94p), preserving full temporal resolution, but causes vertical jitter ("bobbing") in static areas.

Motion-Adaptive Deinterlacing (e.g., YADIF)

Motion-adaptive deinterlacers analyze differences between successive fields:

• Static areas are processed using the weave algorithm to maintain full spatial resolution.
• Moving areas are processed using spatial interpolation (bobbing) to eliminate combing.

Filters like YADIF (Yet Another Deinterlacing Filter) look at the current field alongside the preceding and succeeding fields to dynamically map motion thresholds across pixel blocks.

Motion-Compensated Deinterlacing (e.g., QTGMC)

State-of-the-art processors use motion vectors to track how objects move across fields over time. Instead of simply interpolating missing lines vertically, the algorithm searches adjacent fields for the missing pixel data along the path of motion.

These algorithms perform noise reduction, temporal consistency checks, and sub-pixel interpolation. This preserves sharp edges, eliminates line shimmer, and smoothly reconstructs full-resolution progressive frames at double the original frame rate.

5. Final Progressive Frame Reconstruction

After applying the spatial and temporal calculations, the deinterlacer outputs a fully progressive pixel grid (such as 720x480p). Depending on the user's settings, the video is output at single frame rate (discarding half the temporal information to produce 25p or 29.97p) or double frame rate (generating 50p or 59.94p to match the original field capture rate), ready for modern progressive displays or re-encoding into formats like H.264 or HEVC.