Skip to content

docs: the rig guarantees the seams, and the composing is the consumer's - #459

Merged
firejune merged 1 commit into
mainfrom
docs/the-object-and-the-composer
Sep 7, 2026
Merged

docs: the rig guarantees the seams, and the composing is the consumer's#459
firejune merged 1 commit into
mainfrom
docs/the-object-and-the-composer

Conversation

@firejune

@firejune firejune commented Sep 7, 2026

Copy link
Copy Markdown
Owner

Closes #445. bun run selftest 568 PASS / 0 FAIL / 0 SKIP / 0 HOLE, exit 0 — unchanged, as a text-only change should leave it.

The caption

Scene direction of this kind is authorable on plain Spine 4.3

The rig guarantees the seams — idle loops while gaze and turn return to rest, so the hand-offs meet at 0 differing pixels — and the composing is the consumer's.

This is an upgrade rather than a hedge, and the reason is measurability. "Scene direction" cannot be measured. The seam can, and films/three-scene already measured it: "That is a property of the rig rather than of the assembler… Step 7 measures exactly that and gets 0 differing pixels." The film proved the object-side claim in its own body; only the wrapper mislabelled it.

<code> rather than backticks, following the dominant form in the file's other two captions — a caption inside raw <em> is not a place to gamble on inline markdown.

The bone name

docs/FACE.md illustrated a fix as "splitting a bone into a stack (headroll_idle under headroll_scene)".

⚠️ A previous pass called this a socket name and therefore ours. That was wrong, and reading the line is what corrected it — it is not a socket, it is a hypothetical bone in prose, and the sentence containing it says outright that "both are runtime or rig decisions the motion spec cannot express". So the thing being named is consumer-side.

headroll_layer, because the same sentence already says "on the layered track". The name now says what the sentence says, and nothing is borrowed.

Two sites the card lists, deliberately unchanged

  • films/three-scene/ keeps its name. A demo is the one place a scene belongs — "there we are the consumer" — and the name is honest. Renaming it would make the tree less accurate, not more.
  • assets/rigc-scene.gif keeps its name. README.md links it by a raw githubusercontent URL on main, so renaming breaks that link for every cached and forked copy of a 3.11 MiB asset. A filename carries the link; the caption carries the claim. The card says a rename is not required by the rule, and it is not.

Two further sites left alone on the card's own instruction — checked, not assumed

docs/FACE.md:841 ("scene work exercises a renderer where game-part work does not") and :903 ("a scene-direction budget"). Both are about the art's fidelity and the renderer, not about who composes.

🔁 A previous application of this rule listed four violations and three fell away on inspection. Sweep the claim, not the word — a grep-and-replace here is the over-application that makes a good rule annoying and then ignored.

The caption under the portrait film claimed scene direction, which the object
and scene clause says is not ours. The replacement is not a hedge: scene
direction cannot be measured and the seam can, and films/three-scene already
measured it at 0 differing pixels across every hand-off. The film proved the
object-side claim in its own body and only the wrapper mislabelled it.

FACE.md illustrated a bone stack as headroll_idle under headroll_scene. The
sentence around it already says both fixes are runtime or rig decisions the
motion spec cannot express, and already calls the thing a layered track, so the
name now says what the sentence says.

Two sites the card lists are deliberately unchanged. films/three-scene keeps its
name because a demo is the one place a scene belongs, and there the name is
honest. assets/rigc-scene.gif keeps its name because README links it by a raw
URL on main, renaming breaks that link for every cached and forked copy of a
3.11 MiB asset, and a filename carries the link rather than the claim.

Two further sites are left alone on the card's own instruction, and checked
rather than assumed: scene work exercises a renderer and a scene-direction
budget are both about the art's fidelity, not about who composes. Sweep the
claim, not the word.

Closes #445.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
@firejune
firejune enabled auto-merge (squash) September 7, 2026 05:45
@firejune
firejune merged commit 8d13735 into main Sep 7, 2026
1 check passed
@firejune
firejune deleted the docs/the-object-and-the-composer branch September 7, 2026 05:48
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.

the README caption presents a composition we made as something rigc does

1 participant