Commit Graph

21 Commits

Author SHA1 Message Date
Z User
bb72baaa3a h3-plan-gate: Step 5 delivery rewritten from verified node pack source (socket vs widget, override rule, one-line plan file) 2026-10-30 22:03:46 +00:00
Z User
7df79bcf09 Restore the strict refmod-isolation validator: token-echo check catches scattered re-descriptions the n-gram check missed 2026-10-30 21:54:01 +00:00
Z User
fb271a8399 Six-section Ref2VA setup: three definition forms, refmod isolation, no length caps 2026-10-30 21:53:22 +00:00
Z User
5ae3aacb73 refmod text lives only in subject_definitions - bodies reference <Subject N> alone, echo checked 2026-10-30 21:07:31 +00:00
Z User
de19ecee44 no length caps and no lean doctrine in subject_definitions - node-author compactness rules removed (different H3 model) 2026-10-30 20:41:57 +00:00
Z User
962c667418 No length cap on the binds-to-image definition form 2026-10-30 20:35:58 +00:00
Z User
c2559f0572 subject_definitions takes exactly three forms: no-image traits, binds-to-image, verbatim refmod 2026-10-30 20:33:24 +00:00
Z User
28f9b80c83 subject_definitions is a lean identity anchor; the look is woven into detailed_description
Rule 6: subject_definitions never carries the full head-to-toe look or a
nude/bare skin state - it is a lean identity anchor (who the label is, tied
to its reference image). The full look is woven into the [Shot 1] body of
detailed_description according to what is going on in the scene: face, hair,
build as the body moves, each garment where the action touches it, footwear
where steps land, uncovered skin named "nude" where the scene exposes it,
and wardrobe/skin changes stated as part of the action that causes them.

- validator: defs lines reject nude/bare-family skin states and anything
  over 25 words (lean-anchor cap); body re-state message updated; two new
  suite FAIL cases (nude-in-defs, look-dump-in-defs); fixtures rewritten
  to lean defs with the look woven into both [Shot 1] bodies
- all three LLM prompts + SKILL.md: defs role flipped everywhere (template,
  SUBJECT ANCHOR, PROMPT CRAFT, REF2VA spec, worked examples, self-checks);
  outline 'subjects' array now documented as the cast inventory the look is
  spent from
2026-10-30 19:42:00 +00:00
Z User
30cbaeb8f6 Switch plan.json output to the native H3 ref2va prompt format
Each scene prompt is now the six native reference sections —
subject_definitions / summary [reference generation] / retention_analysis /
detailed_description (1-2 style sentences then [Shot 1]) /
overall_soundscape / non_diegetic_music — one line per element, headers on
their own line. The full head-to-toe look (nude-not-bare) lives in
subject_definitions and is restated compressed in the [Shot 1] body; both
cast members are defined per scene against the wired subject references.
verify_plan_rules.py auto-detects the format per shot and validates the
ref2va contract (canonical order, directive, definitions, retention,
style prefix, thin floor), so legacy t2va plans keep passing. Both system
prompts, the assembler prompt and SKILL.md carry the new skeleton,
worked examples and self-checks; example_output/README.md marks the old
run as the t2va craft reference.
2026-10-30 19:28:36 +00:00
Z User
e5778f8536 Common-action catalog: show the steps, never just mention the act
Everyday actions are where the LLM compresses hardest. Both prompts get
a worked not-X-but-Y catalog (unbuttoning a shirt, changing a diaper,
turning on a lamp, opening a door, buckling a strap, handing over an
object) under the rule that an action named without its steps is a
report, not a render; self-checks and SKILL.md carry the same catalog.
2026-10-30 06:31:02 +00:00
Z User
baaa533709 Leg-crossing as second canonical example; no motion too minor
'the woman crosses her legs' -> 'raises her left leg and places it
across her right leg' joins the ball example in both prompts, SKILL.md
and the self-checks, with an explicit clause: every movement and motion
of every character, large or small, is written out as its steps.
2026-10-30 06:24:56 +00:00
Z User
7a7321ec82 Spell out every change and movement step by step (H3 drift rule)
Extends the body-mechanics rule beyond the subject's actions: camera
moves, lighting changes, weather, background, wardrobe and any
transformation each get their own ordered sequence of visible stages —
never a summarized change ('the lights dim'). Added to both prompts'
PROMPT CRAFT + CAMERA element + self-checks and SKILL.md.
2026-10-30 06:12:23 +00:00
Z User
fa764ec933 Full look every scene; 'nude' for uncovered body parts, never 'bare'
H3 needs every aspect of the look restated per scene, and uncovered
skin is steered by the skin-state word 'nude' (tank-top with nude
arms), not 'bare'. Added to both prompts' subject anchor + craft
rules + self-checks and SKILL.md; the validator prints a non-blocking
note when a description line uses 'bare'.
2026-10-29 23:18:56 +00:00
Z User
405386a1af Decompose every action into body mechanics (H3 drift rule)
Compressed outcome verbs ('picks up the ball') leave H3 nothing to
animate between start and end poses, so it guesses and drifts off
intent. The craft rules in both prompts and SKILL.md now require the
ordered joint-level sequence with the hand/limb named per step.
2026-10-29 22:46:12 +00:00
Z User
df8df95d39 User-approved scene guide is a mandatory workflow step
Encoded the intended user flow end to end: story + scene count ->
scene guide (outline.json) -> USER APPROVAL (hard stop, changes applied
to the guide first) -> scenes composed one at a time (~10s each) from
the approved guide -> continuity verified across the handoff chain ->
final plan.json.

SKILL.md: goal line states the five moves; new Step 2 'Present the
guide for approval (do not skip)' — plain-language presentation, STOP
and wait, approved guide is the contract; Step 3 works only from the
approved guide; Step 4 renamed 'Verify the continuity, put it all
together'; revising section split into guide-level vs single-scene.

README: workflow reframed as 4 stages with Stage 2 = approval (human in
the loop, never skipped); example gains a real guide-presentation +
'Approved — go' exchange between outline and scene writing; steps
renumbered.

Scene-agent prompt: stage table gains the Approval row and a 'Stage 2 is
a hard stop' note; assembler prompt re-labeled Stage 4. Suite + --check
green.
2026-10-29 19:28:18 +00:00
Z User
f5575fa678 Scene composition is the gate, not word counts
The skill exists so the LLM writes/rewrites each scene one at a time
with full attention on making it a complete scene composition — not so
it can hit a word target.

Prompts: the word-budget craft bullets are replaced by an 8-element
composition checklist (subject anchor, style anchor, continuity anchor,
blocking, camera, performance, physical texture, exit state) in both the
scene-writer PROMPT B (new SCENE COMPOSITION section) and the single-pass
prompt. Chase the elements, never a number; padding is as much a failure
as a thin one-liner.

Validator: hard word bands (<10s: 150-200, >=10s: 200-300) removed from
plan and fragment checks. Replaced by a thin-scene floor (100/120 words)
that only trips lazy output, plus a new continuity-anchor check — chained
fragments must open inside the previous end_state. No target band, no
upper limit; output lines say so.

SKILL.md/README: goal, Step 2, craft rules, table rows, and example
outputs updated to match. Suite, --check, chained fragment checks all
green.
2026-10-29 19:13:27 +00:00
Z User
83cc5145bf Realign: one plan JSON, written one scene at a time
The deliverable is the single merged plan.json — the same shape the
single-pass system prompt emits — and the multi-clip video is what the
node itself produces from it: each shot renders as its own clip, chained
through the H3 context loop into one continuous video. Removed the
--clips per-scene export (external stitching is NOT the design).

Workflow restated everywhere: outline turn -> the LLM writes each scene
one at a time (one focused turn per scene) -> build_plan.py puts it all
together at the end into plan.json -> paste once, queue once.

build_plan.py: --clips option and export block removed; docstring now
describes the node-side chaining. README: intro/example/validating/
wiring rewritten around the single deliverable. SKILL.md: goal, Step 3
(Put it all together at the end), Step 4 (the node makes the multi-clip
video), and the revising loop updated. Suite + --check green.
2026-10-29 19:00:11 +00:00
Z User
54a8271072 Multi-clip videos as the primary deliverable: per-scene clip export
build_plan.py --clips [DIR] (default clips/) now exports one validated
single-shot plan JSON per scene (clip_01.json, ...) alongside the merged
plan. Each clip renders through plan_json_input as its own video at FULL
duration (no chain overlap trim), then the clips are stitched in order;
the mid-motion open/close craft rules make the cuts read as one take.
plan.json stays as the optional single continuous-take render.

SKILL.md Step 3/4 rewritten (merge exports clips + plan; deliver =
render-per-clip then stitch), README intro/example/wiring/validating
updated to multi-clip-first with real clip output shown. Verified from a
simulated project root: export green, every clip re-checks PLAN OK.
2026-10-29 18:40:18 +00:00
Z User
b308dd77e1 SKILL.md: rewrite for project-root layout and Windows venv
The <skill_dir> placeholder and python3 commands confused the LLM and fail
on Windows. Now: all commands are concrete and relative to the project root
(the folder containing h3-plan-gate/), invoked as
.venv\\Scripts\\python.exe (Windows venv layout) with python/.venv/bin/python
as fallbacks; python3 removed entirely. Added a start-of-task smoke test,
explicit file-output locations, one-file-at-a-time scene loop with immediate
fragment validation, and a revise-one-scene loop. All command paths verified
from a simulated project root (suite, --check, merge, fragment --prev all
green).
2026-10-29 18:27:44 +00:00
Z User
b9b0e41469 Add SKILL.md: Agent Skills entry point for the h3-plan-gate folder
The h3-plan-gate directory is now a drop-in skill for Cline (.cline/skills/
or ~/.cline/skills/) and any Agent-Skills-compatible agent. SKILL.md carries
the full agent procedure (outline -> one fragment per turn with per-turn
validation -> build_plan.py merge -> delivery notes) plus the condensed
craft rules; the prompt .md files remain the deep reference and worked
examples stay bundled. README files table updated.
2026-10-29 17:55:52 +00:00
Z User
bd50b1ba51 Add H3 plan-JSON gate: agent workflow prompts, validator, assembler, examples
- h3_scene_agent_system_prompt.md: outline turn + one-scene-per-turn writer
  with a truthful _handoff chain (style_and_cast frozen at outline)
- h3_plan_assembler_system_prompt.md: verbatim-copy LLM merge step
  (deterministic alternative: build_plan.py)
- h3_plan_json_system_prompt.md: single-pass variant for short 2-4 scene
  stories; prompt_prefix fixed to [""], native H3 three-line shot template
- verify_plan_rules.py: node-replica validator (plan + fragment modes,
  handoff-chain checks, embedded pass/fail suite)
- build_plan.py: deterministic fragment merger with --check for existing plans
- h3_agent_workflow_examples/: worked 3-scene greenhouse courier run
  (outline -> 3 fragments -> validated plan.json, 1042 frames = 43.42s)
2026-10-29 17:18:28 +00:00