Add focused NPC portrait and speaking animation to dialogue UI

[JOY]

Summary

Make focused NPC dialogue feel like an RPG conversation instead of debug chat. The player should clearly see which NPC is speaking, with an NPC-side portrait or bust beside the NPC chat text and a simple speaking animation while the NPC response is delivered.

Source Docs

• docs/design/56-focused-npc-dialogue-portrait-lipsync-design.md
• docs/design/141-alpha-focused-npc-portrait-speaking-builder-brief.md
• docs/design/30-alpha-ux-flow.md
• docs/design/37-ai-npc-backend-client-roadmap.md
• docs/setup/play-mode-smoke-checklist.md
• docs/playtests/_templates/focused-dialogue-portrait/README.md

Scope

D1: Focused Dialogue Portrait UI

• Add a portrait or bust anchor to the focused dialogue panel.
• Place it beside the NPC-side message column.
• Resolve portrait presentation from actor profile or local presentation catalog.
• Show a stable missing-portrait fallback.
• Hide or reduce focused NPC overhead nameplate while focused dialogue is active.
• Prove portrait/bust identity matches the focused actor and selected line id.

D2: Text-Timed Speaking Animation

• Add a presentation-only NPC speaking state.
• Start speaking animation when NPC text reveal begins.
• Stop when the line completes, fails, is skipped, times out, becomes stale, is rejected, or dialogue exits.
• Clear old portrait and speaking state when the focused NPC switches.
• Keep this independent from full voice, TTS, or Convai import.

D3: Dialogue State Polish

• Keep player and NPC locked into dialogue state until exit.
• Face the player and NPC toward each other during focused dialogue.
• Disable normal movement input while typing.
• Use non-combat talk or neutral idle animation during dialogue.
• Keep ambient bubbles readable and separate from nameplates.
• Show source and fallback state only in compact debug form.

D4: Ida Faber ARKit Blendshape Readiness

• For Ida-family semi-real bodies, inspect the imported Unity mesh blendshape names before assuming support.
• If the body exposes compatible face blendshapes, map it through an IdaFaberArkit52 lip sync profile.
• Start with mouth and blink targets only; do not require full emotional facial acting for alpha.
• Keep text-timed speaking fallback for Ida bodies without compatible blendshapes.

Future Hook Only

• Leave hook points for audio-amplitude or provider viseme/blendshape lip sync.
• Use Ida Faber's ARKit 52 blendshape documentation as the semi-real body reference, but verify the actual imported Unity mesh before implementation.
• Full voice/TTS/session playback work remains tracked separately in Track scoped NPC voice sessions and playback implementation lane #288.

Acceptance Criteria

• Pressing interact near an NPC enters focused dialogue only when in range.
• NPC messages appear on the NPC side and player messages remain on the opposite side.
• Focused NPC portrait or bust appears beside NPC-side text.
• Portrait actor id matches focused actor id and NPC-side text actor.
• Missing portrait fallback does not break layout and keeps actor name visible.
• NPC speaking animation starts when NPC answer begins revealing or playing.
• Speaking animation stops reliably on complete, skip, exit, fallback, timeout, stale reply, rejected reply, failure, duplicate-send block, or focused NPC switch.
• Typing chat does not trigger movement, jump, attack, or interact input.
• Player and NPC remain facing each other until dialogue exits.
• Ambient bubbles and nameplates do not overlap heavily in the default camera view.
• Play Mode smoke verifies one successful model-backed or scripted-smoke NPC reply and one fallback-visible case.
• Provider output remains presentation-only and cannot mutate memory, relationship, quest, TIME, SECOND, inventory, combat, or body lifecycle state.

Evidence Template

• docs/playtests/_templates/focused-dialogue-portrait/
• docs/playtests/_templates/focused-dialogue-portrait/portrait-actor-parity-note.md

References

• docs/design/56-focused-npc-dialogue-portrait-lipsync-design.md
• docs/design/37-ai-npc-backend-client-roadmap.md
• docs/design/16-npc-society-multi-agent-architecture.md
• PR docs: design focused NPC portrait dialogue #267
• Ida Faber blendshape docs: https://docs.idafaber3d.com/features/blendshapes

[JOY]

Implementation packet added in PR #188: docs/design/45-alpha-backlog-execution-packets.md.

Use that packet as the build contract for this issue: goal, minimum data contract, Unity/backend surface, verification, and cut line. This comment does not close the issue; implementation still needs its own PR and runtime evidence where applicable.

[JOY]

Design updated for the focused NPC dialogue presentation request.

Canonical doc:

• docs/design/56-focused-npc-dialogue-portrait-lipsync-design.md

Implementation direction:

• Ship NPC-side portrait/bust beside NPC chat text first.
• Add a presentation-only speaking state.
• Use text-timed mouth/speaking fallback for alpha so this does not block on voice or Convai import.
• Add hook points for later TTS/audio-amplitude and Convai/provider viseme or blendshape lip sync.
• Provider animation data is presentation only. Nakama/Fusion still own game state, memory, relationship, quest, TIME, SECOND, inventory, and combat authority.

Relevant merged PR: #267