prompt(guide): RESUME — re-shoot the ModellerGuide, on the LIVE site this time - #64
Open
red1oon wants to merge 4 commits into
Open
prompt(guide): RESUME — re-shoot the ModellerGuide, on the LIVE site this time#64red1oon wants to merge 4 commits into
red1oon wants to merge 4 commits into
Conversation
…oot the LIVE site this time User: "Good to update Modeller User Guide with some good pics? Fetch the prompts/# on that so we resume in new session." This is that file; the new dated section is the pick-up point. The reason it is worth doing NOW is the reason the old shots were misleading: every existing image was captured on localhost, because localhost was the only place the Modeller worked -- the live site received a 134-byte Git-LFS pointer for its mesh and drew bounding boxes for months. Fixed 2026-07-30. So for the first time a live screenshot and a localhost screenshot show the same thing, and the section says: shoot live, and treat a boxy frame as a deployment regression rather than a framing problem. Names what is newly worth showing, all live-verified today (geoV 6 / sw v43): a wall's real material layers (71 walls, 215 slabs); the party wall as 5 honest slabs at 0.493m of its type's 0.550m, which is the better teaching image because it shows the model telling the truth about a unit boundary; and a window riding its wall now that SampleCastle went 9/74 to 74/74. Carries forward the 7 still-bad frames from THREAD 1 with the instruction to fix the capture harness first rather than re-shoot by hand over a capture bug. Restates the capture discipline explicitly -- prove numerically first, capture the proven state second, never a screenshot as a pass condition -- and the safe_gh_deploy-only ship path. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
…eality
"What a wall is made of" described layer-slicing as a future promise
("once that slicing ships... will render as 7 stacked slabs"). It shipped
(#57/#59/#1096, then the §ROW33-EXCEPTION ruling #62/#1102) — verified live
against https://red1oon.github.io/bim-ootb/modeller/modeller.html today
(§LAYER-GATE armed multiLayer=80 layeredHashes=71 refused=0; party wall
124 tris, bbox thickness 0.493m). Rewrote to present tense with the real
count: 5 slabs (not 7 — 2 belong to the neighbouring unit, correctly
excluded), summing to 493mm of 550mm.
Grid-Stretch section's SampleCastle sentence said "partial... 9/74" — the
void-anchor fix (#1095) closed that gap. Live-verified reach = 74/74
(anchors seeded n=65). Updated the sentence to match.
No new screenshots this pass: per-layer render colour isn't wired yet (a
named, separate gap — MODELLER_MASTER.md row 3), so no camera angle shows
5 visually distinct slabs; and every void-anchor case has an invisible
host by construction, so a before/after ride shot reads as the window
detaching from an unrelated visible neighbour rather than "riding" —
correct math, misleading picture. Text carries the verified numbers
instead of a screenshot that can't yet show them.
Also fixed an unrelated pre-existing mkdocs --strict break (wrong
relative path in MigrateComparisonPaper.md, blocking any deploy).
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
… named unphotographable as conceived Records why: per-layer render colour isn't wired (subject 1) and every void-anchor case has an invisible host by construction (subject 2), so neither is a plain-screenshot problem right now. Text-only update already deployed + live-byte-verified (see docs/ModellerGuide.md this same branch). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Pick-up file for a new session:
prompts/RESUME_MODELLER_GUIDE_SCREENSHOT_FIX.md, new dated section at the end.Why now. Every existing guide image was shot on localhost — because localhost was the only place the Modeller worked. The live site received a 134-byte Git-LFS pointer for its mesh and drew bounding boxes for months. Fixed 2026-07-30. For the first time a live screenshot and a localhost screenshot show the same thing.
Newly worth showing (live-verified,
geoV 6/sw v43):Carried forward: the 7 still-bad frames, with the instruction to fix the capture harness first rather than re-shoot by hand over a capture bug.
Restated: a screenshot is guide content, never proof. Prove numerically first, capture the proven state second. Ship only via
safe_gh_deploy.sh.🤖 Generated with Claude Code
https://claude.ai/code/session_01FZX1USuzGMn9Z88cJXAaWq