Video Fundamentals: Frame Rates, Codecs, Formats, Streaming & Compression

Core Concepts of Video Fundamentals

What is Video?

Video is a time-based medium that combines sequential images (frames) with audio to create motion. It relies on three fundamental principles:

1. Persistence of vision: The brain perceives rapid image changes as continuous motion.
2. Frame rate (fps): Number of frames displayed per second (e.g., 24fps for films, 60fps for gaming).
3. Resolution: Pixel dimensions (e.g., 1080p = 1920×1080 pixels).

graph LR
    A["Video"] --> B["Frames"]
    A --> C["Audio"]
    B --> D["Frame Rate (fps)"]
    B --> E["Resolution (pixels)"]
    C --> F["Sample Rate (Hz)"]
    C --> G["Bit Depth (bits)"]

Frame Rate and Motion Perception

Frame rate determines smoothness and realism of motion. Higher fps reduces motion blur but increases file size.

Worked Example: YouTube Upload

• A 60fps 1080p video requires ~100 Mbps bitrate (vs. 24fps at ~20 Mbps).
• Trade-off: Higher fps improves quality but increases storage/bandwidth costs.

Video Resolution and Aspect Ratio

Resolution defines pixel count, while aspect ratio defines shape (width:height).

Real-World Tie-In: Daraz Live

• Daraz Live streams in 720p at 30fps to balance quality and mobile bandwidth.
• Why? Most Nepali users have limited data (~500MB/month), so 4K would be impractical.

Video Codecs: Compression Algorithms

Codecs (codecer-decoder) compress raw video data to reduce file size. Key types:

How Codecs Work:

1. Intra-frame compression: Compresses each frame independently (e.g., JPEG for video).
2. Inter-frame compression: Exploits similarities between frames (e.g., H.264 predicts motion).
   • Keyframes (I-frames): Full frames for reference.
   • P-frames: Predicted from previous frames.
   • B-frames: Bidirectional (from past/future frames).

sequenceDiagram
    participant I as I-Frame (Keyframe)
    participant P as P-Frame
    participant B as B-Frame
    I->>P: Reference
    P->>B: Forward Prediction
    B->>P: Backward Prediction

Worked Example: Ncell’s Video Calls

• Uses H.264 at 30fps, 720p to minimize latency.
• Why? Low bitrate (~1 Mbps) ensures smooth calls on 2G/3G networks.

Video Container Formats

Containers hold video + audio + metadata in a single file. Common formats:

Video Streaming Protocols

Streaming delivers video without full download. Key protocols:

Worked Example: YouTube Streaming

1. HLS splits video into small chunks (2-10s).
2. Adaptive bitrate: Switches between 720p/1080p based on user bandwidth.
3. Buffering: Preloads chunks to avoid playback stutter.

timeline
    title YouTube Buffering Process
    YouTube Server: Requests chunk 1 (720p) | Updates: Chunk 1 loaded
    User Device: Plays chunk 1 | Requests chunk 2
    YouTube Server: Sends chunk 2 (adjusts to 1080p if bandwidth allows)

Video Compression Techniques

1. Spatial Compression

• Reduces pixel redundancy in a single frame (e.g., JPEG for images).
• DCT (Discrete Cosine Transform): Converts pixels to frequency domains.

2. Temporal Compression

• Exploits similarities between frames (motion estimation).
• Example: A talking head changes little between frames → store only differences.

3. Quantization

• Reduces precision of data (e.g., storing 8-bit color instead of 16-bit).
• Trade-off: Higher quantization = smaller file but more artifacts.

In the Real World

1. YouTube (Google)

   • Uses H.264/VP9 + HLS/DASH for adaptive streaming.
   • Why? Balances quality and bandwidth for global users (including Nepal’s slow connections).

2. Daraz Live (Daraz)

   • Streams in 720p at 30fps with RTMP for low-latency shopping events.
   • Why? Most Nepali users have limited data (~500MB/month), so 4K is impractical.

3. Ncell’s Video Calls

   • Uses H.264 at 30fps, 720p (~1 Mbps) to work on 2G/3G networks.
   • Why? Older networks can’t handle 4K, but low latency is critical for calls.

4. Nepal Stock Exchange (NEPSE) Live Feeds

   • Uses low-bitrate H.264 (480p, 15fps) to show stock prices in real-time.
   • Why? Traders need speed over quality—high fps/4K would delay updates.

Exam Tip

1. Frame Rate vs. Resolution:

   • 24fps = cinematic, 30fps = standard, 60fps+ = gaming/VR.
   • 1080p is most common in exams; know its pixel count (1920×1080).

2. Codecs and Containers:

   • H.264 = YouTube, H.265 = 4K, MP4 = web, MKV = lossless.
   • RTMP = live, HLS/DASH = on-demand.

3. Compression Trade-offs:

   • Lossy (H.264) = smaller files but artifacts.
   • Lossless (ProRes) = no quality loss but huge files.

4. Real-World Applications:

   • YouTube = H.264 + HLS.
   • Daraz Live = 720p, 30fps, RTMP.
   • Ncell calls = H.264, low bitrate.

5. Common Pitfalls:

   • Don’t confuse frame rate (fps) with bitrate (Mbps).
   • Aspect ratio is width:height (e.g., 16:9), not pixels.
   • Keyframes (I-frames) are critical for seeking in videos.

Final Checklist for Exams: ✅ Can you list 3 frame rates and their uses? ✅ What’s the difference between H.264 and H.265? ✅ How does YouTube’s adaptive streaming work? ✅ What’s the pixel count of 1080p? ✅ Why does Daraz Live use 720p instead of 4K?