Why text barely shrinks but scans collapse
The single most useful thing to understand about compression is where the weight lives. Text and vector graphics are stored as compact instructions, effectively drawn on demand, so they are already tiny and there is little to squeeze. Images are stored as pixels, and pixels are heavy, so a scanned page or a photo-rich report is almost entirely image data. Compression works by resampling those images to a sensible resolution and re-encoding them efficiently, then discarding duplicate and unused objects. That is why a 40-page contract of pure text hardly changes while a 5-page scan can shrink by most of its size.