One room instead of two stacked panels

The complaint was that it looked repetitive. The measurement was more specific than the complaint: at 1440 by 900 with a scene open, the rail occupied x 128, width 1184, from y 67 to 188; the scene occupied x 128, width 1184, from y 153 to 876. Same abscissa, same width, one on top of the other, overlapping by thirty-five pixels.

Two boxes of identical footprint each carrying its own frosted surface, each with its own soft edge, and a seam between them. The eye counts the boxes instead of reading the agent.

#One surface, two regions

The fix was structural rather than cosmetic: the two components became one. The glass belongs to the room; the dock and the scene have no background, no border and no shadow of their own. A frosted surface should group an experience, not frame each line inside it.

#Three measurements decided the rest

• The artefact takes 60% of the room and the reading takes 40. The previous split was 49/51 — the application and the text carried the same weight, which is not what the page is selling.
• The dock drops to the rail sprite size when a scene opens. The owl sprite does not compress, so fifteen card-size faces demand 1,072 px in a room that offers 869 at 1024 — 210 px of horizontal overflow, measured.
• The glass closes with two crossed linear gradients rather than a radial mask. A radial gradient never reaches zero on a wide box: at 124% radius on 1,270 px, the left edge is still 79% opaque, which is why a hard vertical edge was visible at x 85 — a jump from RGB 7 to 12 in five pixels.

EvidenceAfter the change, the brightness profile across the left edge rises from 6 to 13 over 160 pixels with no step. That is the whole difference between a plate and a volume.