Same data, different order — see why JPEG compression works so well
8×8 DCT basis patterns. Low frequency at top-left, high at bottom-right. Active coefficients highlighted in cyan.
Both views use the exact same amount of data at every point. The only difference is the order in which information is revealed.
Pixel basis sends pixels left-to-right, top-to-bottom. You see a growing strip of the image.
DCT basis sends low-frequency coefficients first, then progressively higher frequencies. You see a blurry-but-complete image that sharpens over time.
Natural images concentrate most energy in low frequencies, so the DCT representation captures the important structure first. This is why JPEG works: discard high-frequency coefficients with minimal visible impact.
Try it: Compare Campus vs. Checkerboard at 10%. Which basis wins for each?