Multimedia Compression: Techniques, Standards & Applications

What is Multimedia Compression?

Multimedia compression reduces file sizes while preserving quality. Without compression, a 2-minute 1080p video would require ~1.5 GB of storage. Compression makes streaming (YouTube, Ncell TV) and storage (cloud, phones) feasible.

Why Compress?

• Storage: Save space on devices (phones, hard drives).
• Bandwidth: Faster downloads/uploads (e.g., Daraz product images).
• Cost: Reduce server/storage costs (e.g., NTC’s video archives).

1. Lossless vs. Lossy Compression

mindmap
  root((Compression))
    Lossless
      "No data loss"
      "Used for text, spreadsheets"
      "Examples: ZIP, PNG"
    Lossy
      "Data discarded"
      "Used for audio/video"
      "Examples: JPEG, MP3"

Lossless Compression

• How it works: Removes redundancy (e.g., repeating patterns in text).
• Methods:
  • Run-Length Encoding (RLE): Replaces repeated sequences (e.g., AAAAA → 5A).
  • Dictionary Coding (LZW): Stores repeated phrases once (e.g., ZIP files).
  • Huffman Coding: Assigns shorter codes to frequent symbols (used in PNG, TIFF).

Example: Compressing a black-and-white fax (many repeated pixels).

Lossy Compression

• How it works: Sacrifices quality for smaller files (e.g., discarding less noticeable details).
• Methods:
  • Quantization: Reduces color depth (e.g., 24-bit RGB → 8-bit palette).
  • Discrete Cosine Transform (DCT): Converts images into frequency components (used in JPEG).
  • Psychovisual Models: Exploits human perception (e.g., ignoring high-frequency noise in audio).

Example: A JPEG photo at 80% quality loses some detail but saves 60% space.

2. Key Compression Techniques

A. Audio Compression (MP3, AAC)

flowchart TD
  A["Raw Audio"] --> B["Fourier Transform"]
  B --> C["Remove Inaudible Frequencies"]
  C --> D["Quantize Remaining Data"]
  D --> E["Encode with Huffman"]
  E --> F["MP3 File"]

How MP3 works:

1. Filtering: Removes frequencies humans can’t hear (>16 kHz).
2. Quantization: Reduces bit depth for less critical frequencies.
3. Huffman Coding: Shortens common bit patterns.

Real-world use: Spotify, Ncell Music use MP3/AAC to stream songs efficiently.

B. Video Compression (H.264, H.265)

flowchart TD
  A["Video Frames"] --> B["Keyframe (I-frame)"]
  B --> C["Predictive Frames (P/B-frames)"]
  C --> D["Motion Estimation"]
  D --> DCT --> E["Quantization"]
  E --> F["Entropy Coding"]
  F --> G["H.264 File"]

How H.264 works:

1. Keyframes (I-frames): Full frames (like a snapshot).
2. Predictive Frames (P/B-frames): Store only changes (e.g., a moving car’s delta).
3. Motion Compensation: Predicts movement between frames.

Example: YouTube’s adaptive bitrate switches between H.264/H.265 based on your internet speed.

C. Image Compression (JPEG, PNG)

JPEG vs. PNG:

• JPEG: Best for photos (uses DCT).
• PNG: Best for graphics (uses RLE + LZW).

3. Real-World Applications

A. Streaming (YouTube, Ncell TV)

• Problem: A 1-hour 4K video = ~100 GB uncompressed.
• Solution: H.265 (HEVC) reduces it to ~5 GB while keeping near-original quality.
• How it works:
  1. Server encodes video in multiple bitrates (720p, 1080p, 4K).
  2. Your device picks the best quality based on speed.
  3. Adaptive bitrate streaming (ABR) adjusts in real-time.

B. Mobile Apps (WhatsApp, Pathao)

• WhatsApp: Uses WebP (Google’s format) for images to reduce data usage.
• Pathao: Compresses driver location updates to save battery.

C. E-Commerce (Daraz, NEPSE)

• Daraz: Compresses product images to <100 KB for faster loading.
• NEPSE: Stores stock charts as PNG/SVG (scalable vector graphics) for crisp displays.

4. Worked Example: Compressing a Photo

Scenario: A 5 MB JPEG photo needs to fit in a WhatsApp message (max 16 MB).

Steps:

1. Original: 5 MB, 24-bit color.
2. Apply JPEG Compression:
   • Reduce quality to 70% → 1.2 MB.
   • Use progressive JPEG (loads in stages).
3. Result: Photo fits in WhatsApp and loads faster.

5. Advantages and Disadvantages

6. Common Compression Standards

In the Real World

1. YouTube: Uses H.264/H.265 + AAC to stream videos globally with minimal buffering. A 10-minute video might be 500 MB uncompressed but only 50 MB after compression.
2. Ncell’s Video-on-Demand: Compresses movies to ~1 GB per hour (vs. 10 GB uncompressed) to save mobile data.
3. Daraz Product Images: Compressed to <50 KB using WebP to load instantly on slow 3G networks.

Exam Tip

• Define: Lossless vs. lossy compression (1 mark each).
• Compare: JPEG vs. PNG (2 marks).
• Apply: Calculate compression ratio (e.g., "A 10 MB file becomes 2 MB after compression. What’s the ratio?" → 5:1).
• Diagrams: Draw a block diagram of MP3 compression or H.264 frame types (I/P/B-frames).
• Real-world link: Always relate to streaming (YouTube), mobile apps (WhatsApp), or e-commerce (Daraz).

Key Formulas to Remember

1. Compression Ratio = Original Size / Compressed Size
   • Example: 10 MB → 2 MB → 5:1 ratio.
2. Bitrate = File Size / Duration (in bits per second).
   • Example: 50 MB video in 10 sec → 40 Mbps.

Visual Summary

mindmap
  root((Multimedia Compression))
    Lossless
      "RLE, LZW, Huffman"
      "No quality loss"
    Lossy
      "JPEG, MP3, H.264"
      "Quality trade-off"
    Applications
      "Streaming (YouTube)"
      "Mobile Apps (WhatsApp)"
      "E-commerce (Daraz)"
    Standards
      "JPEG (Images)"
      "MP3 (Audio)"
      "H.264 (Video)"