robbyt/source-compare

Video Source Comparison

Compare multiple video sources to determine which has better quality. Higher bitrate or resolution does not automatically mean better quality.

Core Principle

Quality = how closely video resembles original master.

When comparing sources:

• Both derive from some common master
• The better source is closer to that master
• Technical specs (bitrate, resolution) are secondary to actual visual quality

Comparison Workflow

Step 1: Gather Metadata

Run MediaInfo on both sources:

mediainfo source_a.mkv > source_a_info.txt
mediainfo source_b.mkv > source_b_info.txt

Compare:

• Resolution
• Bitrate
• Encoder used
• Color space parameters
• HDR metadata (if applicable)

Note: Higher numbers don't guarantee better quality.

Step 2: Extract Comparison Frames

Choose frames that reveal quality differences:

Good comparison frames:

• Dark scenes with gradients (reveals banding)
• Scenes with fine texture (hair, fabric, foliage)
• High motion scenes (reveals compression)
• Scenes with text or sharp edges (reveals filtering)
• Areas with strong colors (reveals color handling)

Extract frames with ffmpeg:

# Extract frame at specific timestamp
ffmpeg -ss 00:15:30 -i source.mkv -frames:v 1 frame_15m30s.png

# Extract multiple frames
ffmpeg -ss 00:10:00 -i source.mkv -frames:v 1 frame1.png
ffmpeg -ss 00:25:00 -i source.mkv -frames:v 1 frame2.png
ffmpeg -ss 00:45:00 -i source.mkv -frames:v 1 frame3.png

Step 3: Visual Comparison

Using SlowPics (https://slow.pics):

1. Upload comparison frames from each source
2. Uncheck "Show border" and "Smooth scaling"
3. Use clicker mode (not slider)
4. Press number keys (1/2/3) to switch rapidly between sources

What to look for:

| Area | Better Source Shows | |------|---------------------| | Dark gradients | Smoother transitions, less banding | | Edges | Clean edges, no haloing/ringing | | Textures | Preserved grain/detail, not smeared | | Colors | Natural, not oversaturated or shifted | | Compression | Less blocking, less mosquito noise |

Step 4: Check for Filtering

Signs of harmful filtering:

Lowpassing (blur):

• Soft, smeared appearance
• Fine details less defined than other source
• Some Blu-ray authoring studios apply this

Sharpening:

• Haloing around edges (bright/dark outlines)
• Unnatural crispness
• Line warping on straight edges

Color manipulation:

• Oversaturated colors
• Crushed blacks or blown highlights
• Different color grading than other source

Upscaling:

• Details too sharp for claimed resolution
• AI hallucinations (invented details)
• Native resolution lower than file resolution

Common Scenarios

Blu-ray vs Web

Blu-ray usually better when:

• Same master, Blu-ray has more bitrate
• No filtering applied during authoring
• Proper color space handling

Web can be better when:

• Blu-ray has lowpass filter applied
• Web has newer/better master
• Blu-ray is poor upscale from SD source
• Web version has better color grading

Multiple Blu-ray Releases

Different authoring studios, regions, or editions can have different quality:

• Check authoring studio (some are known for lowpassing)
• Compare frame-by-frame for filtering
• Look for remastered vs original master

SD vs HD

HD version better when:

• Truly native HD content
• Good upscale from clean SD master

SD version better when:

• HD is poorly upscaled from SD source
• HD has additional filtering damage
• SD is closer to original DVD master

Red Flags

Higher bitrate source might be worse if:

• Encoded with inferior encoder (NVENC vs x264)
• Has filtering applied (sharpening, lowpass)
• Is upscaled from lower resolution
• Has wrong color space applied

Higher resolution source might be worse if:

• Upscaled from lower resolution original
• Has AI upscaling artifacts
• Native production resolution was lower

Quick Decision Guide

| Comparison | Check First | Likely Winner | |------------|-------------|---------------| | Blu-ray vs Web | Filtering, authoring | Depends on studio | | 4K vs 1080p | Native resolution | If native, 4K | | HEVC vs AVC | Encoder, settings | Neither inherently | | High vs low bitrate | Encoder efficiency | Higher if same encoder | | HDR vs SDR | HDR legitimacy | SDR if HDR is fake |

Tools Reference

• SlowPics (slow.pics) - Frame comparison hosting
• vs-preview - VapourSynth comparison tool
• mpv - Frame extraction, playback comparison
• MediaInfo - Metadata comparison

Additional Resources

• JET Guide: https://jaded-encoding-thaumaturgy.github.io/JET-guide/
• ${CLAUDE_PLUGIN_ROOT}/references/quality-myths.md - Why specs don't equal quality
• ${CLAUDE_PLUGIN_ROOT}/references/artifacts.md - What quality issues look like