CHG-0009: Two Effects for the Accents


Date	2026-10-11
Change type	Upgrade
Status	Complete 2026-10-11
Systems	`lp-stand-01`, over USB; `eds-lightd` and `eds-console` on the `services` workload
How	`make -C firmware/lp-stand build flash`; `make -C deploy deploy WHERE=builder`
Risk	Low. Most likely to go wrong: `lightd` asking for an effect the stand does not know, which the stand shows as a steady color
Decisions	ADR-0001
Related	CHG-0008

Summary

The stand’s default effect, breathe, lights a cover’s dominant color and nothing else. The other colors lightd picks, the accents, reach the stand in every scene and are used only by sweep.

After this change the stand knows two more effects, both built on the accents: morph, which turns the whole strip from the dominant color to each accent and back, and chase, which sends the accents around over the dominant color. lightd asks for chase by default, and the console draws both.

Goal

The stand’s firmware knows morph and chase, and reports them by name in scene.applied.
A new album produces a scene with effect=chase, applied by the stand.
The console draws it: the dominant color with the accents moving over it.
morph can be asked for by hand and is applied and drawn.

Scope

In scope: the stand’s firmware; LIGHTD_EFFECT on the workload; the console’s drawing of the two effects.

Out of scope: how colors are picked, which is lightd’s and unchanged; the scene contract’s version, which adding an effect name does not move; the LEDs, still not attached.

Risk and impact

Risk	Where	Guard
`lightd` asks for an effect the stand does not know	the stand	the stand is flashed first. An unknown name is shown as `solid`, by design: the right color, not moving
The console draws something the stand does not do	the page	the console’s effects are a port of the firmware’s, and a test holds them to pixels printed from the firmware’s own code
An effect is unpleasant on real LEDs	the room	no LEDs are attached. `LIGHTD_EFFECT` is one line to change back
The timing fix changes the effects already there	the stand	the existing effect tests pass unchanged; the difference is below a millisecond until uptime is in hours

Procedure

make -C firmware/lp-stand test
make -C firmware/lp-stand build flash PORT=/dev/cu.usbserial-0001
make -C deploy deploy WHERE=builder

Verification

make -C firmware/lp-stand logs shows the stand restart and apply the retained scene.
A new cover: lightd logs effect=chase, the stand reports scene.applied effect=chase.
By hand: a scene with "effect":"morph" posted to lightd is applied and reported.
The console’s strip shows patches moving for chase, and one color turning into another for morph.

Undo

Remove LIGHTD_EFFECT from deploy/lightd.env and run make -C deploy units: breathe again. The firmware can stay; it still knows the old three.

Outcome

When (UTC)	What happened
19:16	Host tests passed, 5421 checks. Built and flashed. The stand restarted as `c35de3a`, rejoined, and applied the retained scene, still `breathe`: `lightd` had not been changed yet.
19:18	Deployed. `lightd` lit the retained cover again with the new default: `effect=chase colors=4 palette="#fddb00 91%, #b87113 4%, #cda712 3%, #7e5f1e 2%"`. The stand applied it 0.2 s later and reported `Showing chase in 4 colours`.
19:18	The console’s picture agreed: `chase`, the same four colors, the stand online and reporting `chase` as applied.
19:18	A scene with `"effect":"morph"` and the same palette, posted to `lightd` by hand, was applied and reported by the stand, and drawn by the console. `chase` was then posted back.

All of the goal’s checks hold. What neither effect has had is a look on real LEDs, or a person’s opinion of it on the console.

Departures from the plan

A fault in the effects already there was found and fixed. Animation time was taken as (float)t_ms. A float holds a millisecond count exactly for about four and a half hours; after that the remainder moves in steps, and after a week breathe and sweep would have advanced in jumps of tens of milliseconds. Time is now divided as a whole number first. The stand had never been up long enough, with LEDs, for anyone to see it.
The cover lit at the time is 91% one yellow, so its accents are small and close to it in color. chase on it is a quiet thing. A cover with real contrast has not been shown yet.

Follow-ups

Look at both on real LEDs, and on the console with a few covers, and say which should be the default. LIGHTD_EFFECT is the one line.
An accent very close to the dominant color barely shows in chase. lightd could prefer accents that differ from it; that is a color decision, and its to make.
morph blends in RGB. Between two colors far apart in hue it passes through a duller one; blending in OKLab would not.