Yadif: A Fast, High-Quality Deinterlacing Filter for Video

Yadif: A Fast, High-Quality Deinterlacing Filter for VideoInterlaced video—once the standard for broadcast and many consumer formats—stores each frame as two fields captured at slightly different times. While efficient for legacy transmission and display on CRTs, interlacing creates combing, judder, and motion artifacts on progressive displays (modern LCD/OLED/LED screens) and in digital workflows. Deinterlacing converts interlaced content into progressive frames suitable for editing, encoding, and viewing. Among available tools, Yadif (Yet Another DeInterlacing Filter) is widely used for its balance of speed and quality.

This article explains what Yadif is, how it works, when to use it, how to apply it (with examples using FFmpeg), its strengths and limitations, tips for best results, and comparisons with other deinterlacing methods.

What is Yadif?

Yadif is a software deinterlacing filter originally developed for the libav/FFmpeg projects. It performs temporal and spatial analysis to reconstruct progressive frames from interlaced fields. Yadif is designed to be computationally efficient while producing visually pleasing results, making it a common choice for real-time playback, batch transcoding, and video processing pipelines.

Key facts:

• Type: Single-rate temporal deinterlacer (can also operate in double-rate mode).
• Integration: Available as a filter in FFmpeg/libavfilter and in media players/builds that include libavfilter.
• Primary goal: Fast operation with good motion handling and minimal artifacts.

How Yadif works (overview)

Yadif’s approach combines motion detection with spatial interpolation:

• It analyzes adjacent fields (previous and next) in addition to the current field.
• For stationary areas, it merges fields to produce a full-resolution progressive frame.
• For moving regions, it uses motion-adaptive interpolation to reduce combing and temporal artifacts.
• Yadif can operate in:
  • Single-rate mode: outputs one progressive frame per input frame (field blending or interpolation).
  • Double-rate mode: outputs two progressive frames per input frame (restores original temporal resolution), often producing smoother motion but doubling frame count.

The filter is relatively simple compared to advanced motion-compensated deinterlacers, which strive to estimate precise motion vectors; Yadif uses heuristics to determine motion and choose interpolation strategies, which keeps CPU/GPU usage low.

When to use Yadif

Use Yadif when you need:

• Fast deinterlacing for batch transcoding or live playback.
• Good visual results for mixed content (talk shows, sports, news) without heavy computational cost.
• Simple integration via FFmpeg or libavfilter-based tooling.

Avoid Yadif when:

• You require the very best quality deinterlacing for archival restoration or finishing workflows—advanced motion-compensated (MC) deinterlacers may produce better results.
• The footage contains very complex or fast motion where motion-compensated approaches would reduce artifacts more effectively.

Using Yadif with FFmpeg — practical examples

Yadif is built into FFmpeg as the yadif filter. Below are common command examples.

1. Basic single-rate deinterlace (fast, reduces combing):

   ffmpeg -i input_interlaced.mp4 -vf yadif -c:v libx264 -crf 18 -preset veryslow -c:a copy output_deinterlaced.mp4 

   This produces one progressive frame per original frame (not doubling frame rate). Use when you want simpler output size and smoothness.

2. Double-rate deinterlace (restores temporal resolution; smoother motion):

   ffmpeg -i input_interlaced.mp4 -vf "yadif=mode=1" -c:v libx264 -crf 18 -preset veryslow -c:a copy output_deinterlaced_2x.mp4 

   mode=1 enables double-rate (also known as field doubling), so a 25 fps interlaced source can become 50 progressive frames per second.

3. Force field order (if source field order is known or detected incorrectly):

   ffmpeg -i input_interlaced.mp4 -vf "yadif=deint=interlaced:parity=1" -c:v libx264 -crf 18 -c:a copy out.mp4 

   Note: many FFmpeg builds auto-detect field order; parity=0 or 1 forces top-field-first or bottom-field-first handling respectively.

4. Batch processing multiple files (Linux shell example):

   for f in *.mkv; do ffmpeg -i "$f" -vf yadif -c:v libx265 -crf 22 -c:a copy "deint_${f%.*}.mkv" done 

Strengths of Yadif

• Speed: Lightweight algorithm suitable for real-time or fast batch processing.
• Simplicity: Easy to apply with FFmpeg; configurable modes.
• Balanced quality: Good compromise between artifact reduction and sharpness preservation for many types of content.
• Availability: Included in standard FFmpeg/libavfilter distributions; no external dependencies.

Limitations and common artifacts

• Not motion-compensated: Yadif uses motion-adaptive heuristics rather than computing motion vectors, so it may produce artifacts on very complex or high-speed motion.
• Haloing or soft edges: In some cases, edges may appear slightly soft or show haloing around high-contrast transitions.
• Field order errors: Incorrect field-order detection can cause combing or judder until parity is corrected.
• Doubling frame rate increases bitrate and file size and may require frame-rate-aware filters (like fps or decimate) later in the pipeline.

Tips for best results

• Verify field order: If you see persistent combing, try forcing parity=0 or parity=1.
• Use double-rate only when you want smoother motion and are prepared for larger file sizes or higher frame-rate outputs.
• Combine with denoising before deinterlacing if source is noisy—noise can confuse motion detection.
• For final mastering or archival: test motion-compensated deinterlacers (e.g., mvdeint, nnedi3 with interlacing-aware workflows) and compare results.
• If using hardware-accelerated encoders, match deinterlacing output frame rate and resolution to avoid reprocessing.

Yadif vs. other deinterlacers (brief comparison)

When to choose an alternative

• Choose motion-compensated deinterlacers for archival restoration, cinematic material with complex motion, or when minimizing artifacts is critical.
• Choose field blending only when smooth, film-like motion is acceptable and motion artifacts are minimal—note that blending introduces ghosting.
• Consider neural-network-based filters (nnedi3, RIFE-based temporal upscalers) for advanced pipelines where quality is paramount and compute budget allows.

Summary

Yadif remains a practical and popular deinterlacing choice because it delivers a strong balance of speed, simplicity, and visual quality. It’s especially useful for real-time playback, batch transcoding, and everyday video processing where computational resources or turnaround time are limited. For the highest-quality restoration work, evaluate motion-compensated or neural approaches, but for most common use cases, Yadif is a fast, effective, and easy-to-use deinterlacer.