Skip to content

feat(rigboy): finish v2 with all nine clips on the Spineboy skeleton - #95

Open
firejune wants to merge 3 commits into
mainfrom
jinni/80-rigboy-v2-finish
Open

firejune wants to merge 3 commits into
mainfrom
jinni/80-rigboy-v2-finish

Conversation

@firejune

@firejune firejune commented Nov 1, 2026

Copy link
Copy Markdown
Collaborator

Finishes rigboy/v2: Spineboy's skeleton, constraints and nine human animations on rigboy's own parts, cut at every joint. walk was the only clip that held both bars; now all nine do. No art changed since the cut rounds.

Bar Before After How HQ read it
Rig-play build 23 of 23 23 of 23 rebuilt from rig/ in a copy: all 5 build files byte-identical
Ankle bend within 15° of Spineboy 3 of 9 clips (run 23.98° and 40.91°) 9 of 9, largest 13.02° spine-core, 241 samples per clip rather than the lane's 41
12 joint discs at 0 px 6 of 9 clips (near shoulder 68 px in jump, 43 px in death) 9 of 9 rendered and read jump, death, run; the other six are the lane's figures
Spineboy images in the tree none none 0 of 42 hashes across 206 PNGs
Mesh folds 0 0 lane's figure; v2 has no meshes
@rig-c/play on all nine clips from a clean package pass pass lane's figure, Nova job in its receipts

Causes. The rear foot target sat 18 px forward of the short rear leg, and run kept raw foot translations for a leg shorter than the source's, so the shin folded. The target moved under the hip with walk compensated, and run's translations were scaled to the leg. The shoulders ride torso3 while the torso is one rigid region on torso, so a bending spine carried them off the chest; jump and death key the shoulders back onto it. Three caps sat up to 0.8 px off their pivots; those origins were centred with no rest-pose shift.

Rejected. Replacing v1 in place: it is the documented single-painting fold boundary and consumers pin it. Deleting the round folders: they are the provenance, so they moved under rounds/ to let the final rig read first.

Closes #80

🤖 Generated with Claude Code

firejune and others added 2 commits November 1, 2026 23:23
Every clip now holds the two bars walk already held. All 12 joint discs read
0 px on 9 of 9 clips, and the ankle bend stays within 15 degrees of Spineboy
on 9 of 9. Art is untouched since the cut rounds; the fix is pivots and keys.

The rear foot target sat 18 px forward of the short rear leg, and run kept
raw foot translations for a leg shorter than the source's, which folded the
shin. The target moved under the hip with walk compensated, and run's foot
translations were scaled to the leg. The shoulders ride torso3 while the
torso is one rigid region on torso, so a bending spine carried them off the
chest; jump and death now key the shoulder back onto it. Three caps sat up to
0.8 px off their pivots, and those origins were centred with no rest shift.

HQ measured, not read from the report: rebuild byte-identical on 5 files, 241
samples per clip for bends, disc gate on jump, death and run, 0 of 42 Spineboy
PNGs in the tree.

Rejected: replacing v1 in place, because v1 is the documented single-painting
fold boundary and consumers pin it, and deleting the round folders, because
they are the provenance. They moved under rounds/ so the final rig reads first.

Closes #80

Co-Authored-By: Jinni <342598851+fj-jinni[bot]@users.noreply.github.com>
Co-Authored-By: Roa <342598975+fj-roa[bot]@users.noreply.github.com>
@chatgpt-codex-connector

chatgpt-codex-connector Bot commented Nov 1, 2026 •

Copy link
Copy Markdown

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review Status Commit Review trigger
📝 Code Review ✅ Completed 2026-11-01T14:26:37.213673Z 650dea6 PR opened
ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

  • Open a pull request for review
  • Mark a draft as ready
  • Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

The ankle bars measured bones, and the bones were right: the painted
front foot sat about 33 degrees toe up at idle against the original's
1, because its attachment was rotated. This round measures pixels.

Attachments: front-foot rotation 47.48 to 16.48, rear-foot 110.11 to
108.61, each turned about the bone. Rest targets and the hip rose to
meet the ground.

HQ's own renders and a separate fit over the middle 50 % of the sole:
planted difference from the original at most 4.72 degrees on any foot,
idle lowest pixel 1.47 px off the ground. Joint discs 0 px on walk, run,
jump and death at 1528x1864.

Rejected: redrawing either foot, all 20 part images are unchanged, and
a different measure per foot; one function reads both feet and both
characters.

Co-Authored-By: Roa <342598975+fj-roa[bot]@users.noreply.github.com>
Co-Authored-By: Jinni <342598851+fj-jinni[bot]@users.noreply.github.com>
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.

rigboy v2: Spineboy's skeleton with rigboy parts, cut at every joint

1 participant