Skip to content

Put a legend on every animation, and say they are sped up - #67

Merged
ngodzik merged 7 commits into
mainfrom
docs/readme-volume
Nov 1, 2026
Merged

ngodzik merged 7 commits into
mainfrom
docs/readme-volume

Conversation

@ngodzik

@ngodzik ngodzik commented Nov 1, 2026

Copy link
Copy Markdown
Owner

Stacked on #66.

The README's animations had no legend, and nothing said they run faster than the model's clock.

What changes

  • Every animation under docs/media/ carries a legend naming what each colour is.
  • The README says the animations are sped up.

How it was checked

check result
the five animations regenerated from the same runs, a legend on each
tests 209 green

The whole CI was replayed locally on this branch.

🤖 Generated with Claude Code

ngodzik and others added 7 commits November 1, 2026 04:21
A network's [mechanics] carried j_medium, j_self_base and
j_self_from_growth, turned into every cell's mechanics and handed to
nothing: the lattice has always priced a boundary by the kinds of the
cells across it (ContactEnergy, chapter 1). A file that set them looked
as if it changed adhesion and did not.

They are removed from MechanicsSpec, CellMechanics and the example
networks, and [mechanics] now refuses unknown keys, so a network that
still sets one is an error naming the key. Chapter 2 and the changelog
say so. No run changes: eight short examples give main's output byte
for byte.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A held pixel is a boundary condition and never changes, yet every sweep
of the field's relaxation went over the whole grid. A bath holds
everything outside the tissue, and a spheroid starts as a speck in a box
sized for the end of its run, so on its first hundred steps nearly every
pixel swept was held.

Each relaxation now sweeps only the smallest box holding every free
pixel, with the two buffers made to agree outside it first. The values
computed, the largest change per sweep and so the number of sweeps are
those of the full sweep, so every run is the same to the last bit. On one
core, the spheroid's first 120 steps go from 65 seconds to 18.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…lume

Every tissue so far was one slice. A configuration's depth, one when left
out, now stacks slices: a voxel has six neighbours, the first cell is a
ball, a cell divides across a plane whose normal is drawn over the
sphere, a vessel seen end on is a cylinder through the box, the bath
reaches in from all six faces, and radii are those of a ball, or around
a vessel of a cylinder as long as the box. A tumour in a bath or around
a vessel, mutating or under a lethal stress, runs in a volume; a
capillary, a sheet, vessels that grow, immune cells and lesions still
run only in the plane and are refused otherwise.

A plane is a volume one slice deep, run by the same code. The slices
are stacked as rows of the grid, so the first slice is where the plane's
pixels always were, and everything a volume adds, a draw, a neighbour,
a term of a sum, comes after what the plane did and only when there is
a second slice. Every example reruns to the same output, byte for byte.

Dividing no longer scans the whole lattice for each cell: the pixels of
every cell ready to divide are listed in one pass, since splitting one
cell relabels only its own pixels. The most living cells a run lets
divide, 2000, is now a configuration's max_cells, since a tumour in a
volume needs tens of thousands.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
… in a still medium

A spheroid grown in a volume formed its core where the measured
melanoma spheroids do and never stopped. What stops it, each piece
opt-in and refused where it has no meaning (chapter 27):

- debris_loss: a dead cell's debris loses a share of its volume each
  step, from the step it dies, paid where the living tissue closes in
  on the dead, at the rate Greenspan's model asks of the measured end
  structure;
- sweeps_per_step, so that the tissue carries what the core loses out
  to its surface;
- [cycle]: a cycle's duration drawn at birth, a cell growing through it
  from half its target;
- contact_order: contact across edges and corners, weighed so that a
  boundary along an axis costs what it did;
- fluid the tissue shuts in drains into the living tissue around it;
- [waste] and [necrotic_factor], read by the networks as inputs, run
  and kept out of the example;
- [still_medium]: the level next to a spheroid that of a still medium
  around a ball, from the field's inflow.

The example grows in a still medium whose diffusion is set against the
measured end, inside the bracket oxygen's measured diffusions and a
flowing medium leave: core at 242 um (measured 237), arrest at 0.85 of
it (0.87), end at 361 um (348), a cycling rim of 21 um (36), and cores
that grow with size, half as much as the measured ones. Every 2D
example gives main's output byte for byte.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The viewer drew every tissue as one slice. A tissue in a volume is now
drawn as the faces between its voxels and the medium, lit from in front
and from the side, a voxel touching another cell drawn darker so that
the cells show. With the cut on, which is how it opens, the plane
through the middle of the box is drawn as well, so that a spheroid is
seen cut in half, with its core; C opens and closes it. Chapter 26 says
how to watch it.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Every animation under docs/media/ now says in its corner what its
colours mean, and the README says that a few seconds of each stand for
days to months of a tissue's life.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@ngodzik
ngodzik changed the base branch from feat/viewer-3d to main November 1, 2026 13:08
@ngodzik
ngodzik merged commit b4e7da8 into main Nov 1, 2026
9 of 16 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant