Four Labs,
One Brief

"Brighter" is not a hue question. The real one is how bright a field can get before the ember system breaks — and the answer is a computed ceiling: #F0A040 hits 2.14:1 on white, so it can never be the armed colour above field L 0.1135. Above that line armed burns instead of glowing. Three palettes are drawn to that number, and the shipping light scheme turns out to be inverted today — its armed beat is 28% quieter than an idle one.

The shipping screen is a tempo display that also makes noise. It reads BPM well at 1 m and tells you it is running — and it never tells you where you are. The beat dot is a fixed 100 ms flash, so at 120 BPM a glance lands on a row showing nothing 80% of the time. That is an encoding bug, not a sizing bug: make "played" an accumulated area instead of a transient and position reads at 100% duty. Three layouts then argue over what the freed space is for.

Today the pill track asks opacity to answer "where am I", with 1.33:1 of headroom to say it — while a spent verse can render 2.04× brighter than the bar you are playing. Split the question by viewing distance instead: a NOW plate that names the section and counts bars-of-section, an approach gutter that warns two bars out, and a form ribbon that fits the whole tune to width and never scrolls. Brightness then only has to mean spent vs. not-spent.

The tick lane grows a rail. Right-hand strokes rise above it, left-hand strokes fall below it — so sticking gets a second, position-coded channel that is colour-blind safe by construction and survives at 2 m, where the letters cannot. One nullable field (hand: Hand?) lets the 27 accent grids the app already ships render through the same view, unchanged. On PDF import the honest answer is assist, not automate: every suggestion arrives unconfirmed and SAVE stays disabled until the drummer has cleared them.