Add character production pipeline and video tooling notes

[RollNuts]

Summary

• Add a practical game-character production pipeline for Maya, Meshy, Unity, and Unreal.
• Add divergent hero/enemy character pattern cards so we compare multiple readable directions before polishing one procedural placeholder.
• Add a 2D-to-3D workflow that treats 2D concept sheets as the gate before Meshy preview and Maya cleanup.
• Add free video analysis tooling notes after installing and smoke-testing ffmpeg/ffprobe locally.
• Add a minimal Unity .gitignore so generated Library/, Temp/, Logs/, and UserSettings/ folders do not pollute PRs.

Validation

• git diff --check passed before each commit.
• ffmpeg 8.1.2 and ffprobe 8.1.2 installed via Homebrew and smoke-tested with synthetic 1-second video/audio.
• Verified ffprobe metadata read, frame extraction, and volumedetect audio analysis.

Scope

This PR does not generate or import 3D assets. It defines the spend gate and candidate patterns before using Meshy/Maya.

Stack

Base: #10 plan/market-research-and-task-split.
This should remain separate from #15-#17 Unity implementation stack.

[RollNuts]

Veripsa Core: please review this as a research/planning PR stacked on #10, separate from the Unity implementation stack (#15-#17).

Scope is limited to character production process, divergent character pattern cards, 2D-to-3D/Meshy/Maya spend gates, video analysis tooling, and a minimal Unity .gitignore for generated local folders.

Please flag whether this belongs as one planning PR, whether the character pattern cards should be split from video tooling, and whether the .gitignore should be moved into #10 before merge.

[RollNuts]

Veripsa update: generated concept-sheet variants were added to make the character direction visual, not just textual.

Added sheets: H02, H07, H08, H09, F01, F02, F03, plus a contact sheet and silhouette strip. H09 intentionally abstracts only broad hooded/ritual/relic cues from the user-provided mood image and avoids the key weapon, star charm, black zipper coat, childlike proportions, and recognizable franchise silhouette.

Current recommendation: keep H09, F01, H07, and F03 as primary evaluation candidates. Do not discard H02, H08, or F02; reuse them as elite enemy, alternate class, later trap class, or skin candidates. No Meshy/API spend yet: next gate is max two Meshy previews after 64-128 px game-camera readability review.

[RollNuts]

Veripsa update: this PR now includes a world-definition pivot after multi-agent benchmark review.

Key change: stop weapon-first / rugged-armor character exploration as the active direction. Added docs/product/game-definition-threadlight.md defining Threadlight Pilgrimage: a top-down compact ARPG where a lantern-thread surveyor repairs frayed provinces and brings memories back to a Loom Caravan.

Added docs/research/commercial-expression-benchmark.md to document why successful adjacent games sell role/world/emotional promise before weapons. Updated market-fit, visual/audio benchmark, Maya/Meshy gate, task split, generation log, and character cards so Meshy/Maya spend is blocked until a world contact sheet + gameplay mock prove the Threadlight direction. Existing generated characters are preserved as part library only, not final protagonist candidates.

[RollNuts]

Stack update for Veripsa: follow-up PR #19 is open on top of this branch: #19

Recommended order: review/merge #18 first for market + world definition, then #19 for the visual world-contact-sheet proof. #19 intentionally does not spend Meshy/Maya credits; it provides the screen-facing gate that must pass before any 3D preview.