Render scripts

Generate every PNG that lives under assets/images/index/{simcoon,fedoo,microgen}/. The scripts are not run by Jekyll — they are run once locally, and their output PNGs are committed alongside the rest of the site.

Layout

File	What it produces
`render_simcoon.py`	All `assets/images/index/simcoon/*.png` (Chaboche EPCHA hysteresis, Mori-Tanaka homogenization, directional stiffness rose, SMA loop, yield surfaces). Pure matplotlib + simcoon.
`render_fedoo.py`	All `assets/images/index/fedoo/*.png` (plate with hole, Kelvin RVE periodic homog, IPC indentation, tube-compression triptych, beam lattice, shell pressure, hero square). Uses fedoo + simcoon (EPCHA / EPICP UMATs) + pyvista.
`render_microgen.py`	TPMS hero stack: cylinder, sphere, helix sweep, graded gyroid, surfaces grid, warm-cylinder variant. Pure pyvista, reads VTK meshes from `_render_inputs/`.
`render_microgen_hero.py`	Mesh-edge close-up of the graded gyroid (`microgen_hero.png`).
`render_banner_cylinder.py`	Splash banner image (`tpms_cylinder.png` — light-themed cylindrical gyroid with mesh visible).
`render_voronoi.py`	3D Voronoi polycrystal tile (`voronoi_polycrystal.png`). Pure scipy + pyvista, no mesh input.

Inputs

The Microgen and fedoo renders need TPMS / lattice / Kelvin meshes — 14 files totalling ~330 MB. They are too large for git and are kept locally (uncommitted, gitignored) at:

_render_inputs/
├── Kelvin.vtk             # truncated-octahedron RVE for periodic homog
├── cylinder_g5_tpms.vtk   # gyroid wrapped on a cylinder
├── gyroid_graded.vtk      # spatially graded gyroid
├── sphere_g4_tpms.vtk     # gyroid wrapped on a sphere
├── sweep_g7_helix.vtk     # gyroid swept along a helix
└── surfaces_0.vtp … surfaces_8.vtp   # 9 unit-cell TPMS for the surfaces grid

The meshes were generated with Microgen; the recipes are in its example gallery.

If/when these stabilise and become worth archiving, scripts/fetch_inputs.py is ready to download them from a Zenodo deposit (set ZENODO_RECORD in that file once a record is published; the SHA-256 manifest is already pinned). For now, the file is dormant.

Running

Each script picks the right conda env from the comment block; in practice the following works on macOS:

# the simcoon / matplotlib charts
conda activate 3mah && python scripts/render_simcoon.py
conda activate 3mah && python scripts/render_voronoi.py
conda activate 3mah && python scripts/render_microgen.py
conda activate 3mah && python scripts/render_microgen_hero.py
conda activate 3mah && python scripts/render_banner_cylinder.py

# fedoo simulations need its own env
conda activate fedoo && python scripts/render_fedoo.py

render_fedoo.py:render_tube_compression writes ~600 MB of intermediate .fdz archives to a tempfile.mkdtemp(prefix="fedoo_tube_") directory and unconditionally wipes them in a finally block. Stale fedoo_tube_* dirs from earlier crashed runs are also swept on startup so the disk stays sane.

After regenerating

bundle exec jekyll build (in the jekyll conda env) to regenerate _site/.
git status — only the changed PNGs and any source edits should appear; the _render_inputs/, results/, examples/, _site/ and .fdz directories are gitignored and will not show.