Video Compression Basics

What is Video Compression?

**Video compression** is the process of reducing the file size of a video by removing redundant or unnecessary data while trying to maintain visual quality. This is essential because uncompressed video files are enormous and impractical for storage or transmission.

Raw Video Reality

• 1080p @ 30fps = ~3GB per minute
• 4K @ 30fps = ~12GB per minute
• Impossible to stream or share easily
• Requires massive storage capacity

Compressed Video Benefits

• 1080p @ 30fps = ~50MB per minute
• 90-95% size reduction possible
• Suitable for streaming and sharing
• Maintains acceptable visual quality

How Video Compression Works

Types of Compression

🔄 Lossless Compression

• **Perfect quality:** No data loss
• **Moderate compression:** ~50% size reduction
• **Use cases:** Professional editing, archival
• **Examples:** FFV1, ProRes, DNxHD

📉 Lossy Compression

• **Quality trade-off:** Some data permanently lost
• **High compression:** 80-95% size reduction
• **Use cases:** Streaming, web, social media
• **Examples:** H.264, H.265, VP9, AV1

Compression Techniques

🖼️ Spatial Compression (Intraframe)

Reduces redundancy within individual frames, similar to JPEG compression for images.

• Removes high-frequency details less visible to human eye
• Uses mathematical transforms (DCT - Discrete Cosine Transform)
• Quantizes color and brightness information
• Each frame compressed independently

⏱️ Temporal Compression (Interframe)

Exploits similarities between consecutive frames to achieve higher compression.

• **I-frames:** Complete frames (keyframes)
• **P-frames:** Predicted frames (reference previous frames)
• **B-frames:** Bidirectional frames (reference past and future)
• Motion vectors track object movement between frames

Understanding Video Codecs

A **codec** (coder-decoder) is the algorithm that performs the compression and decompression. The choice of codec significantly impacts file size, quality, and compatibility.

Codec	Compression	Quality	Compatibility	Best For
H.264/AVC	Good	Very Good	Excellent	General purpose, streaming
H.265/HEVC	Excellent	Excellent	Good	4K content, modern devices
VP9	Very Good	Very Good	Good	YouTube, web streaming
AV1	Excellent	Excellent	Limited	Future-proofing, Netflix

Key Compression Parameters

Bitrate (Most Important)

**Bitrate** determines how much data is used per second of video. Higher bitrate = better quality but larger file size.

Bitrate Types

• **CBR (Constant):** Fixed bitrate throughout
• **VBR (Variable):** Adjusts based on content complexity
• **ABR (Average):** Targets average bitrate

Typical Bitrates

• **4K:** 35-45 Mbps
• **1080p:** 8-12 Mbps
• **720p:** 5-8 Mbps
• **480p:** 2.5-4 Mbps

Quality Settings (CRF/QP)

**CRF (Constant Rate Factor)** provides better quality control than bitrate by maintaining consistent perceived quality.

High Quality

• CRF 17-20
• Professional use
• Large file sizes
• Minimal quality loss

Balanced

• CRF 21-25
• General purpose
• Good quality/size ratio
• Most common choice

High Compression

• CRF 26-32
• Small file sizes
• Visible quality loss
• Web/mobile optimized

Resolution and Frame Rate Impact

Resolution Effects

• **4K (2160p):** 4x the pixels of 1080p
• **1080p:** Sweet spot for most content
• **720p:** 50% reduction in file size vs 1080p
• **480p:** 75% reduction, mobile-friendly

Frame Rate Considerations

• **60fps:** Smooth motion, ~40% larger files
• **30fps:** Standard for most content
• **24fps:** Cinematic look, smaller files
• **15fps:** Significant compression, choppy motion

Advanced Compression Concepts

GOP Structure (Group of Pictures)

GOP defines the pattern of I, P, and B frames, affecting compression efficiency and seek performance.

• **GOP Size:** Number of frames between I-frames (keyframes)
• **Shorter GOP:** Better seek performance, larger files
• **Longer GOP:** Better compression, harder to seek
• **Typical:** GOP of 60-120 frames for streaming

B-Frame Configuration

B-frames provide the best compression but require more processing power and introduce encoding delay.

• **No B-frames:** Fastest encoding, larger files
• **2-3 B-frames:** Good balance of compression and speed
• **4+ B-frames:** Maximum compression, slower encoding

Psychovisual Optimization

Human Visual System Considerations

Modern codecs use psychovisual models to optimize compression based on how humans perceive video:

What We Notice Less

• High frequency details in motion
• Color differences in dark areas
• Slight blurring in background areas
• Compression artifacts in complex textures

What We Notice More

• Blocking artifacts in smooth areas
• Facial detail degradation
• Text and graphics quality
• Color banding in gradients

Compression Artifacts

🚫 Common Artifacts

• **Blocking:** Visible 8×8 pixel blocks
• **Ringing:** Oscillations around sharp edges
• **Mosquito noise:** Flickering around objects
• **Color banding:** Smooth gradients become stepped
• **Motion blur:** Fast movement becomes smeared

✅ Prevention Tips

• **Appropriate bitrate:** Don't over-compress
• **Deblocking filters:** Enable in encoder settings
• **Two-pass encoding:** Better bit allocation
• **Proper preprocessing:** Noise reduction before encoding
• **Test different presets:** Find optimal settings

Encoding Presets

Understanding Encoder Presets (Speed vs Quality)

Preset	Speed	Compression	Quality	Use Case
ultrafast	⚡⚡⚡	⭐	⭐	Live streaming
fast	⚡⚡	⭐⭐	⭐⭐	Real-time encoding
medium	⚡	⭐⭐⭐	⭐⭐⭐	General purpose
slow	🐌	⭐⭐⭐⭐	⭐⭐⭐⭐	High quality archival
placebo	🐌🐌	⭐⭐⭐⭐⭐	⭐⭐⭐⭐⭐	Maximum quality