Skip to content

Add state-driven visuals, approved Velvet plate and overlay art pilot - #1

Merged
astrobyte-dev merged 25 commits into
mainfrom
feature/state-driven-visuals
Sep 18, 2026
Merged

astrobyte-dev merged 25 commits into
mainfrom
feature/state-driven-visuals

Conversation

@astrobyte-dev

@astrobyte-dev astrobyte-dev commented Sep 8, 2026 •

Copy link
Copy Markdown
Owner

Room visuals now derive from existing typed simulation state across all 17 rooms, with separate canonical architecture, scene illustrations and dynamic overlays. Presence, visibility, custody and time stay authoritative; story, parser semantics, relationships, THE PULL and the central mystery are preserved.

The project owner approved the corrected 64-colour Velvet room. The correction and portable evidence were pushed first in fdb3f4b, then the existing promotion workflow bound the exact 320-master / 640-display plate in 288ef7e. The canonical WebP is 52,040 bytes. Source, framing, palette, parent and world-manifest provenance are retained, alongside all previous 48-colour and model experiment evidence. The room's camera, architecture and material identity are locked against casual regeneration. External editing is the provisional hero-room production path; Turbo remains useful for scenes, mood and concepts.

The bounded Velvet overlay pilot supplies one stable Mara candidate, two anonymous figures and five existing-object sprites, with master-grid contacts, counter polygon occlusion, bounded light tint and entity-owned reflections. Actual parser drop/take commands remove the black envelope and its reflection together. The pilot has eleven real LocationVisual review captures, including 320/640, mobile, actual early/late/dawn schedules and separate labelled lighting studies. Draft overlay art is opt-in for review; ordinary feature runtime uses the approved plate without activating these unapproved sprites.

Start with docs/WORKSTATION-HANDOFF.md. Review docs/VELVET-CANONICAL-APPROVAL.md, docs/VELVET-OVERLAY-PILOT.md and docs/visuals/velvet-overlay-pilot/review/index.html. Mara identity, native sprite quality, placement and grounding need human review before activation or expansion. Known art limitations include weak mobile pencil detail, repeated anonymous silhouettes, cleaner prop clusters and approximate reflections.

Validation: 1,008 JavaScript tests; 57 lightweight Python tests; full npm run check including build, narrative, parser, content and embodiment checks; local Pages-path build; formatting; visual asset validation; browser presence/custody/reflection/occlusion/mobile/Off/Reduced/image-failure checks; source/sprite hashes and exact nearest-neighbour master/display verification.

Keep this PR OPEN and DRAFT. Do not merge, update main or deploy public Pages. The existing public blind-playtest deployment is preserved. No new world entities, schedules, gameplay events, portrait library or additional rooms are included in the overlay pilot.

@astrobyte-dev astrobyte-dev changed the title Add state-driven visual preview and local collaborator generator pipeline Add simulation-driven visuals and reference-guided local art tooling Sep 8, 2026
@astrobyte-dev astrobyte-dev changed the title Add simulation-driven visuals and reference-guided local art tooling Add simulation-driven visuals and staged visual production tooling Sep 8, 2026
@astrobyte-dev astrobyte-dev changed the title Add simulation-driven visuals and staged visual production tooling Add state-driven visuals, approved Velvet plate and overlay art pilot Sep 9, 2026
astrobyte-dev and others added 13 commits September 9, 2026 20:30
Introduce commitments, commitment-language, commitment-types, custody and
event-queue modules. World revision 3 adds a social record; older saves
migrate with an empty history and validation rejects malformed agreements.
The parser routes the envelope safekeeping arrangement, keeps a completed
alias introduction from restarting the door scene, and preserves Inez's
non-romantic identity on generic flirting. The journal lists agreements.

Co-Authored-By: Claude Opus 5 (1M context) <[email protected]>
Failed commands stayed in the field after the first human playtest.
Dispatch now owns the text: clear, reset history and completion, refocus,
then call out. Browser playtest expectation updated; add focused command
input browser regression and its fixture.

Co-Authored-By: Claude Opus 5 (1M context) <[email protected]>
New src/trial app with its own engine, state, vessels, interactions,
conversation, save and playtest export. Loaded lazily in dev from
main.tsx when ?trial=sable is present; the original campaign remains the
default. Separate vite.sable.config.ts and sable-trial.html entry, with
dist-sable/ ignored. Includes the Sable test suites, shared interaction
fixture, and the trial browser and evidence scripts.

Co-Authored-By: Claude Opus 5 (1M context) <[email protected]>
Headless and browser review scripts for the human opening pass, the Inez
envelope milestone and the text-opening transition, plus the vestibule
draft setup, component, gallery and review scripts with their Python
layer and validation tools.

Co-Authored-By: Claude Opus 5 (1M context) <[email protected]>
Workstation handoff records the Inez milestone, input correction,
text-led transition, Sable trial and playtest repair. New Sable
foundations, evening and trial handoffs; creator scene guide and story
bible; gameplay direction, new-story and text-led transition notes;
vestibule art brief and next-room review. Visual style records the
future scene-framing preference. Prettier ignores raw playtest
downloads.

Co-Authored-By: Claude Opus 5 (1M context) <[email protected]>
Phaser point-and-click prototype paused at the conversational and
investigative beat, with its evidence captures and licences. Not part of
the active text-led direction; kept for reference.

Co-Authored-By: Claude Opus 5 (1M context) <[email protected]>
Vestibule draft-v1 report, candidate layers, composition, state fixtures
and desktop/mobile review composites, kept unactivated pending owner
visual review. Sable foundations evidence referenced by
docs/SABLE-FOUNDATIONS-HANDOFF.md: gate outputs, hashes, reducer
evidence, preservation and the visible final playtest export.

Co-Authored-By: Claude Opus 5 (1M context) <[email protected]>
Exclude the Inform comparison compiled story, vendored web interpreter
and evidence dumps, the per-run Sable playtest artifact directories
other than the handoff-referenced foundations evidence, and local design
review captures under docs/design/. None of these were tracked.

Co-Authored-By: Claude Opus 5 (1M context) <[email protected]>
Inform 7 source for the Sable comparison, bootstrap, build, audit and
check scripts, web shell, and the README, SPEC, TOOLING and REPORT notes.
Compiled output, vendored interpreter and evidence dumps stay ignored.

Co-Authored-By: Claude Opus 5 (1M context) <[email protected]>
@astrobyte-dev
astrobyte-dev marked this pull request as ready for review September 18, 2026 08:12
@astrobyte-dev
astrobyte-dev merged commit 4366c34 into main Sep 18, 2026
1 check 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