1 torsion · 2 bending · 3 axial · 4 direct high energy · T cycles tissue · [ ] impulse · R re-break
The same subject, the same seed, four silhouettes. Sellán et al.
(10.1145/3549540 §6) say their fault is the same regardless of the directionality of
the impact, and name torsion as one of the missing cases. This is that case: a spiral with long
sharp ends under torsion, a butterfly wedge under bending, a transverse break under an axial
load, comminution under a direct blow.
T is the other half. Cortical bone splits along its own grain — osteons run along the shaft and its toughness is far lower for a crack running with them — so it yields long sharp slivers; trabecular bone crushes, tolerating ~30 % strain against cortical's ~2 %, so it is clamped to three pieces however hard it is hit.
Drop the impulse below the greenstick threshold and the bone bends instead of breaking: one fragment, no fault, and a permanent bow. Greenstick is an outcome, not a mode.
[ ] impact energy · , . strain rate · T tissue · H the histogram
The fragment count is not an artist constant. Grady's energy balance
(10.1063/1.329934) gives a characteristic fragment size
s = (24·G_c / (ρ·ε̇²))^(1/3), and the count is the subject's volume over
s³ — so a faster load makes smaller pieces, cubically. The energy delivered is a
second, hard ceiling: creating n fragments costs G_c per unit of new
surface, and a blow cannot buy more than it brought.
A pistol leaves two pieces and a wedge. A rifle comminutes. The histogram is drawn against
Mott's distribution (10.1098/rspa.1947.0042) — many small, few large.
← → impact angle · ↑ ↓ speed · R substrate roughness · B the old four-splat wall
A wall of stains swept over angle and speed. The aspect ratio is
sin θ — the impact-angle relation the whole of bloodstain-pattern analysis
rests on (10.1520/jfs2003224). Spines come from Knock & Davison's
angle-inclusive law, 0.76 · We^0.5 · sin³θ at R² ≈ 0.9
(10.1111/j.1556-4029.2007.00505.x), and satellites appear past Mundo's splash
threshold K = We^0.5 · Re^0.25 ≥ 57.7.
B is the argument, in one keypress. It shows the same wall drawn the old way — four
baked textures picked by seed % 4 — and the arithmetic beside it:
the probability that four consecutive stains contain a repeat is 90.6 %, so the
expected first visible repeat is the fourth stain. Authoring more textures moves the number and
not the shape of the problem.
1 impact spatter · 2 arterial · 3 cast-off · 4 expirated · 5 drip trail · 6 transfer · C clear
The six SWGSTAIN / ASB TR-033 classes, as six mechanisms rather than one cone at six intensities. The arterial one fires once per systole, a handful of large discrete stains along an arc whose reach falls as the body loses pressure — not another isotropic spray.
Cast-off is tangential, and that is the measured fact rather than the folklore
(Williams et al., 10.1111/1556-4029.13855): the drop leaves along the tip's
instantaneous velocity, never radially outward, and its diameter is inversely
proportional to tip speed. Adam (10.1016/j.forsciint.2019.109934) measured the
pendant volume at 150 µL, so a weapon swung repeatedly sheds less each swing — watch the
counter.
The drip trail's spacing encodes walking speed. The transfer wipe spends a conserved budget, so
a dragged body runs out of blood — and the shape of that is worth watching
rather than guessing at: each contact moves min(load, transfer_rate) out of the
budget, so the marks hold their full width while there is blood to give, the last one
is narrower because only a fraction was left, and then there are no more. Not a fade: a supply
running out. That budget is what makes a pattern tell a story.
← → shear rate · ↑ ↓ hematocrit · [ ] age · Space release a rivulet
Blood is a shear-thinning yield-stress fluid, and the two things you can see follow from that and nothing else. Carreau–Yasuda with Cho & Kensey's constants gives the first: a fast rivulet is thin and races, a slow one thickens and beads.
The second is the interesting one. A yield-stress fluid moves only while the driving stress exceeds its own yield stress, so a rivulet arrests mid-wall the moment the rising yield stress overtakes the falling drive. That is what clotting is — the same comparison, at a different age — so the clot, the arrested rivulet and the pool that stopped creeping are one mechanism rather than three special cases. The plot shows both curves crossing.
← → scrub the age · H humidity · Space play
Scrub the age axis and watch four channels move at once. The colour walks
oxyhaemoglobin → methaemoglobin → hemichrome (Bremmer et al.,
10.1016/j.forsciint.2011.07.027) — three stops, not two, which is why old blood is
a brown rather than a dark red. The gloss collapses; the rim dries first, matte
at the edge while the centre is still wet (Smith, Nicloux & Brutin,
10.1038/s41598-020-65465-4); and the crust cracks late into a craquelure.
Push the humidity past 50 % and a serum halo appears outside the pool,
where the serum phase-separates and spreads (Laan et al.,
10.1016/j.forsciint.2016.08.005). Below the threshold there is none at all, at any
age — a refusal rather than a very small number.
The channel that matters most is not a colour: moistness, not hue, is what drives
disgust (Oum, Lieberman & Aylward, 10.1080/02699931.2010.496997), so
wetness is a specular channel here rather than a tint.
1–4 walk the tier · Space replay the kill · F spam flashes · A the aim cone
The same kill, replayed at four tiers. Reduction is substitution, never deletion. At Stylised the emitter still fires, on the same tick, in the same direction, at the same magnitude — it swaps the blood palette for a spark. Vermintide 2's gore-off deleted the channel and made the game harder to read; Gears of War 4 replaced blood with sparks and kept the hit confirmation.
The stops map onto rating descriptors rather than taste: ESRB's Animated Blood, Blood and Blood and Gore (defined as mutilation of body parts), and PEGI's gross-violence criterion, which turns on emphasis and persistence.
F holds down the flash button, and the meter shows the gate refusing the fourth flash inside any one second — WCAG 2.1 SC 2.3.1, technique G19, whose safe harbour is 3 Hz. A gore framework is mechanically a saturated-red flash generator, so the limiter lives in it and cannot be configured upward.
A draws the aim-exclusion cone: decals that would land on the reticle are refused, because gaze concentrates at screen centre while aiming. Ten degrees, which is the visual-field unit WCAG itself uses — not thirty, which is folklore.
click shoot the body · Space a burst · G gravity direction · D the digest
Blood lands where the subject was hit in UV space, runs down the actual geometry, pools where the surface turns, and dries in place. Texture-space blood accumulation is AAA-only — The Last of Us Part II's blood and wound render targets — and everywhere it exists it is a GPU render target.
Which is why nobody can hash one. This one is CPU-authoritative, so the digest on screen is a fact about the canvas: paint the same sequence twice and it matches, move one hit by a single texel and it does not. That makes it the first wetmap that can be a golden.
Dry paint does not move. Only wet paint drips or spreads, which is what makes a run stop where it stopped.
Space spill · G grab and drag · T tear the mesentery · R reset
XPBD Cosserat-style strands: guts spill with weight, coil on the floor, stay tethered by the
mesenteric membrane, and then tear. The formulation is Deul et al.'s direct position-based
solver for stiff rods (10.1111/cgf.13326) over Bergou et al.'s discrete elastic rods
(10.1145/1399504.1360662).
Deterministic, and that is a design constraint rather than a happy accident: a fixed substep count, a fixed iteration count, a fixed constraint order and no early-out. A tear is monotone — once a mesentery link parts it is skipped forever — so it cannot heal and the solver cannot oscillate.
click shoot · 1–4 gore tier · T tissue · H the numbers · R reset
The flagship: one body, one shot, everything at once. The bone breaks the way that blow loaded it, the blood pattern that fits the wound is thrown, the wetmap it lands on records it in UV space, the guts it opened spill and tether, and the drying it leaves walks the colour and the gloss over the next half minute.
Four crates in one frame — bevy_carnage, bloodstain,
bevy_wetmap, bevy_viscera — which is also why this demo can only be
built from the monorepo: each crate's own mirror contains that crate alone.