# `bevy_carnage::cross_section`
Anatomical cross-sections for cut faces: the skin, subcutaneous fat, muscle, cortical bone and marrow a cut through a limb, a torso or a head actually crosses, at the depths they were measured at on living adults — baked once into a procedural strip texture and painted onto any cap through a second UV channel.
> **A module of `bevy_carnage` since 0.5.0.** This was a crate of its own — the family below — until 2026-09-04, when the seven leaves were folded back into the one crate a game is meant to depend on. This page is its module documentation, kept whole; the paths in it are spelled the way a consumer reaches them now.
## The family
This module is one of seven kernels `bevy_carnage` composes; the umbrella README lists them all with the path each is reached by. Nothing here is a separate dependency any more.
## Why this exists
Every runtime-fracture system paints its cut faces with one "inside" material, and a limb sliced through reads as a statue: flat pink, edge to centre. A real cut is bands, and the bands have widths. A dermis is two millimetres; the fat under it is seven on a thigh and sixteen on an abdomen; the quadriceps is a centimetre and the hamstring two and a half; then a shell of cortical bone, then marrow. Those numbers exist because needle lengths, insulin injections and body-composition estimates depend on them, so they have been measured with ultrasound on thousands of people. This crate carries them, with the papers, and turns them into pixels.
The whole thing is CPU-side and deterministic: a strip is a pure function of the thickness table, a size and a seed, so two machines bake the same bytes and a test can freeze the digest.
## The model
**Depth below skin is exact and cheap.** A cut face belongs to a convex cell whose supplied faces are the subject's own surface. For a point inside a convex polytope the distance to the boundary is the minimum over its faces of the distance to that face's plane — a handful of dot products per vertex. `depth_below_skin` does that, and `annotate_cap` writes the result into `ATTRIBUTE_UV_1` as `(depth / span, along / tile)` without touching `UV_0`, so nothing that hashed a cap before this crate existed moves.
**The bands are a table with sources.** `Layers::for_region` carries, per region:
| Region | Skin | Fat | Muscle | Cortex | Sources |
|---|---|---|---|---|---|
| Limb | 1.9 mm | 7.2 mm | 18.7 mm | 5.0 mm* | Akkus 2012, Derraik 2014, Abe 2015 |
| Torso | 2.2 mm | 16.0 mm | 7.3 mm | 2.0 mm* | Akkus 2012, Derraik 2014, Abe 2015 |
| Head | 2.0 mm | 2.6 mm | 2.2 mm | 6.0 mm* | Dimitrova 2018 (6.8 mm total to bone; the split is this crate's) |
\* Cortical bone thickness is **not corpus-sourced** — no open-access long-bone cortex table was reachable when this was written — and is stated as this crate's own. Override it through `CrossSectionSettings::layers`.
**The bone is a core, not a floor.** Depth alone puts cortical bone at every point past 27.8 mm on a limb, which is right for a shallow wound and wrong for a severed thigh: the whole middle of the cut face goes to bone however wide the limb is, and the muscle ends up a rim. So `Layers` carries `bone_mm` — the long bone's **outer diameter**, 27 mm on a limb, `0` on the torso and `∞` on the head, and **this crate's own number** like the cortex row — and `annotate_cap` measures the cap's own deepest point `d_core` (the maximum over its vertices *and* its centroid, because the deepest point of a convex cross-section is in its interior) and lays the bands out from it:
| depth | reads as |
|---|---|
| `0 .. fat_end` | skin then fat, at their measured depths — they are measured from the surface |
| `fat_end .. d_core − bone_mm/2` | the **muscle** band, stretched to fill whatever is between the fat and the bone |
| `d_core − bone_mm/2 .. d_core` | **cortex then marrow**, so the bone is exactly `bone_mm` across the deepest point |
The core rule engages only where `bone_mm` is finite and positive **and** the cap reaches the cortex, so a head and a shallow limb cut are `uv1_at` to the bit. A `bone_mm` of `0` — the torso — means *no bone*: past the muscle wall there is a cavity this table cannot name, so the muscle band is drawn to the deepest point and no cortex or marrow ever appears on a trunk. `∞` — the skull — is bone without end: the cortex begins at its measured depth and the interior follows, which is the depth-only model. It is a remapping on the way *in*: what lands in `UV_1.x` is still `depth_equivalent / span`, so the strip texture, its frozen digest and every material sampling it are untouched. `uv1_at_core` is the same rule for a caller building its own geometry, and `uv1_at` stays the depth-only one a wound bed wants.
**The strip is anatomy at physical scale.** Fat is lobules ~2 mm across in a septal net (Worley cells); muscle in cross-section is fascicles ~0.7 mm wrapped in perimysium at ~3 mm; cortex is ivory pierced by ~0.1 mm Haversian canals; the marrow cavity opens through a thinning trabecular lattice into yellow marrow with red patches. Because the strip is parameterised in millimetres, a lobule is 2 mm on a thigh and 2 mm on a finger. The muscle band's colour is not authored: it is a 0.3 mm venous film from `bloodstain::spectral` over a mid-grey substrate, the same optics that colour a pool.
## What it exposes
- `CrossSectionPlugin` — bakes one strip per `Region` on `Startup` into `CrossSectionAtlas`, as two images and a `StandardMaterial` that samples them through `UvChannel::Uv1`. System set: `CrossSectionSystems`.
- `CrossSectionSettings` — the per-region `Layers`, strip size, seed and `Scale`. Insert it before the plugin to override.
- `Layers`, `Layer`, `Region` — the table and the query `Layers::at(depth_mm)`.
- `SkinPlane`, `depth_below_skin`, `annotate_cap`, `uv1_at`, `uv1_at_core` — the geometry side.
- `strip`, `Strip`, `Band` — the bake, engine-free apart from `bevy::math`.
No components. The caller decides which region a subject's cells belong to, hands `annotate_cap` the cell's supplied-face planes, and puts `CrossSectionAtlas::material(region)` on the cap.
## References
- Akkus, Oguz, Uzunlulu & Kizilgul, *Evaluation of skin and subcutaneous adipose tissue thickness for optimal insulin injection*, J. Diabetes Metab. 3:8 (2012). `doi:10.4172/2155-6156.1000216`
- Derraik et al., *Effects of age, gender, BMI, and anatomical site on skin thickness in children and adults with diabetes*, PLoS ONE 9(1) (2014). `doi:10.1371/journal.pone.0086637`
- Abe, Loenneke & Thiebaud, *Morphological and functional relationships with ultrasound measured muscle thickness of the lower extremity*, Ultrasound 23(3) (2015). `doi:10.1177/1742271X14554678`
- Dimitrova et al., *Facial soft tissue thicknesses in Bulgarian adults*, Folia Morphol. 77(3) (2018). `doi:10.5603/fm.a2017.0114`
- Bosschaart et al., *A literature review and novel theoretical approach on the optical properties of whole blood*, Lasers Med Sci 29 (2014). `doi:10.1007/s10103-013-1446-7` — through `bloodstain::spectral`.
## Examples
```sh
cargo run --example cross_section_strip # terminal only — bakes the three strips, prints bands and digests
cargo run --example cross_section # a cut cylinder per region, caps banded (needs a GPU)
```
## Licence
MIT OR Apache-2.0, with the crate.