Trillium

What it can come out as

How each one is actually drawn - region by region, and the layers it is built from - is on the coding tab.

Crossing two of them

Finding one

About 1 in 1,786 wild horses is born showing this. That is the whole of the supply: catch one, or make one with the carrot below. Breeds set their own pools and a breed built around this gene will carry it far more often than the wild population does.

Its gene carrot

The file

common/src/main/resources/horsegenetics/genes/trillium.json — a data-driven gene, not a Java class. It is listed in the index.json beside it and registered by Genes' class initialiser at startup.

Why it is built this way

The d strings are pasted verbatim out of the traced SVG - every command, both cases, all subpaths, fillRule evenodd as the artist's file declared it. Paths were SELECTED (the tracer emits dozens per drawing); none was re-typed or re-pointed.

Originally placed with originU/originV 0.27/0.338 (originU is now the posU knob - see the note below) and sizeU/sizeV 0.38/0.273 in body space, so the viewport is x 11.1-26.7 and y 11.4-20.6 - the full height of BODY (x 7.5-29.5, y 11-21) and about two thirds of its length, with clearance on every side. It is sized to the barrel's HEIGHT because every drawing here except the trillium is portrait and meet fits the tall side first; the spare width is centring room. The first mask is PARTS BODY, so nothing the side-plane extrusion reaches can land on a leg or the neck if the mesh ever moves.

Three layers because the reference has three tones and they have to stay in that order: petals, then the ring, then the eye on top. The petals are the only achromatic mass in the drawing, so they are what the coloured form re-hues.

The ring and the eye are two different hues a fixed distance apart on the wheel, and there is no '$hue + n' in this format (rule 16), so the ring carries its own knob and the eye is a fixed amber. The amber is deliberate: it is the one colour that is the same on every trillium, so it is the gene rather than the line.

At this viewport the marking is about 18 texels tall (a texel is 0.5 body units, and meet fits the tall side first). The silhouette and the tonal split survive that; detail finer than about half a body unit inside the drawing does not, and reads as texture rather than as countable structure. That is rule 15 being accepted deliberately, not overlooked - the barrel is 10 units tall and there is nowhere else for the marking to grow.

Arrived at priority 320. Moved to 291: fields and regions, with the other barrel drawings beside moonwing (283). The arriving number was taken, or sat in a band whose family does not describe this marking - priority is both the paint order and the family the sidebar files a gene under.

Over the SVG point ceiling on arrival - the engine flattens every curve to sixteen segments, so a traced outline costs its command count rather than its size. Brought under it with intake/tools/fit-svg.mjs: flattened the way the engine flattens, Ramer-Douglas-Peucker to a point budget, re-emitted as polylines. The petals 3076 to 700 and the ring 1292 to 700, 0.03% of the viewBox changing side, measured through the engine's own inside test on a 400-sample grid. The whole viewBox is a few dozen texels across on the horse, so none of that reaches a texel. The d strings are therefore no longer the ones the trace produced. Gap 168 is the engine-side fix.

LAYERS MADE DISJOINT on the way in. The pale bloom came out a flat white trefoil and the ring and eye inside it did not show: they were being added to the petals beneath them rather than painted over them. A gene's layers do not stack: each measures its delta against the colour the GENE started from and the deltas are summed (gap 169), so nested marks at full strength add up and clamp. Each layer now multiplies in an inverted copy of the shape that sets every later layer apart from it, which for hard-edged masks is exactly the picture stacking would have given - the order in the file still reads as the order of the marking, it just no longer has to be the mechanism.

PLACED BY THE HORSE, NOT THE FILE (2026-09-11). originU is the posU knob, 0.12-0.40, so the viewport's centre runs from about x 12.7 to x 24.2 and the drawing lands anywhere from shoulder to hip. It used to be a constant 0.27, shared with the other five barrel drawings, and the owner's first look at the intake found six different drawings all standing in the same place on every horse. The range keeps every one of the six on BODY with two units to spare at either end - the widest, candelabra, is 6.2 units - checked on a side-view render at six seeds each. originV stays fixed: these are sized to the barrel's full height and there is no room to move up or down.

What it paints

Each outcome, described the way the gene file describes it, and then the layers that draw it in the order they are painted. The prose is the part a reader checks a bake against; the list underneath is what the engine actually did.

Pale trillium

Three off-white petals on the barrel around a grey ring and a darker eye.

1. petals — PARTS × SVG × SVG inverted × SVG inverted, painted with TOWARD
2. ring — PARTS × SVG × SVG inverted, painted with TOWARD
3. eye — PARTS × SVG, painted with TOWARD

Black trillium

Three near-black petals on the barrel, the ring and the eye stepped darker still.

1. petals — PARTS × SVG × SVG inverted × SVG inverted, painted with TOWARD
2. ring — PARTS × SVG × SVG inverted, painted with TOWARD
3. eye — PARTS × SVG, painted with TOWARD

Coloured trillium

Three petals in the line's own colour around a ring of a second colour and a fixed amber eye.

1. petals — PARTS × SVG × SVG inverted × SVG inverted, painted with TOWARD
2. ring — PARTS × SVG × SVG inverted, painted with TOWARD
3. eye — PARTS × SVG, painted with TOWARD

Wild type

Nothing.

Paints nothing.