Scene resource anatomy

The First Contact scene is a native .scene CR2W resource rooted at scnSceneResource. Its root records more than the visible node graph. Actor acquisition, entry and exit names, screenplay data, localization storage, referenced animation sets, interruption policy, debug symbols, and graph metadata all belong beside the graph.

The outer resource

This is the focused shape used by the completed checkpoint. Handle wrappers and unrelated serialization metadata are omitted:

scnSceneResource
├── version: 5
├── cookingPlatform: PLATFORM_PC
├── sceneCategoryTag: minorQuests
├── actors: [contact actor 0]
├── playerActors: [V actor 1]
├── entryPoints: [start -> node 1]
├── exitPoints: [contact_done -> node 3]
├── interruptionScenarios: [Default scenario 0]
├── screenplayStore: scnscreenplayStore
├── locStore: scnlocLocStoreEmbedded
├── resouresReferences: scnSRRefCollection
├── sceneGraph: scnSceneGraph
├── sceneSolutionHash: scnSceneSolutionHash
└── debugSymbols: scnDebugSymbols

version: 5 is the scene-root version. It is not the outer WolvenKit JSON format version or the CR2W header’s game-version number. Likewise, PLATFORM_PC is the cooking target and minorQuests is the category tag; they do not register or launch the resource.

The serialized property is spelled exactly resouresReferences. The missing second c is part of the native schema name. Correcting it to natural English creates a different, unrecognized property in a hand-edited source.

Typed stores and collections

Empty typed containers are still part of a sound minimal resource. This focused WolvenKit-shaped excerpt shows the decisive types:

{
  "locStore": {
    "$type": "scnlocLocStoreEmbedded",
    "vdEntries": [],
    "vpEntries": []
  },
  "resouresReferences": {
    "$type": "scnSRRefCollection",
    "lipsyncAnimSets": [
      {
        "$type": "scnLipsyncAnimSetSRRef",
        "asyncRefLipsyncAnimSet": {
          "DepotPath": "base\\animations\\facial\\generic\\interactive_scene\\generic_facial_lipsync_gestures.anims"
        }
      }
    ],
    "cinematicAnimNames": [],
    "cinematicAnimSets": [],
    "dynamicAnimNames": [],
    "dynamicAnimSets": [],
    "gameplayAnimNames": [],
    "gameplayAnimSets": [],
    "ridAnimations": []
  }
}

The full scnSRRefCollection in this checkpoint also retains typed empty arrays for RID animation containers, animation sets, camera animations, cyberware animation sets, deformation animation sets, and facial animation sets. The root keeps empty ridResources, effects, props, markers, notable points, reference points, VO-info, and workspot arrays. These values describe the intentionally small scope; they are not invitations to delete the owning properties indiscriminately.

The spoken line is externally localized, so scnlocLocStoreEmbedded remains empty. That store becomes relevant for scene choices, not for the subtitle and voice path taught here. See Localization paths.

Solution and debug data

The checkpoint retains the nested solution-hash types:

sceneSolutionHash: scnSceneSolutionHash
└── sceneSolutionHash: scnSceneSolutionHashHash
    └── sceneSolutionHashDate: <unsigned value>

Treat sceneSolutionHashDate as scene metadata that must survive the authoring and round-trip workflow. Do not use it as a screenplay, event, or content identity, and do not claim that matching it proves runtime behavior.

scnDebugSymbols is also structured data. First Contact supplies two performersDebugSymbols rows and explicit empty arrays for:

• sceneEventsDebugSymbols;
• sceneNodesDebugSymbols;
• workspotsDebugSymbols.

Debug rows do not acquire actors or connect graph nodes. They map editor-facing performer information onto the native scene identities explained in Actors and performers.

Default interruption scenario

First Contact includes one scnInterruptionScenario, not an omitted or null policy:

The unequal thresholds create a one-unit hysteresis band: the distance must rise above 6 to interrupt and fall below 5 to return. Do not rewrite them as a single equality test. The scenario’s interrupt/return line policy is scene behavior; it does not set a completion fact, choose the contact_done exit, or deactivate a community. Those are separate handoff and cleanup operations.

The scenario shape and distance-condition families are Observed in vanilla. The exact First Contact serialization is Structurally validated. The supplied Lab 5 tests do not interrupt an active line or exercise CutDestination. If your scene supports those paths, add explicit walk-out, return, and reload cases rather than inferring their behavior from normal playback.

What to preserve in a focused comparison

When comparing a WolvenKit round trip, review semantic structure rather than incidental handle allocation:

1. confirm the root type, version, platform, and category;
2. confirm every typed store and collection still has its native $type;
3. compare actor, performer, graph-node, entry, exit, screenplay, event, and locstring IDs in their own domains;
4. compare the lipsync array length with every actor’s referenced slot;
5. compare both interruption conditions and their thresholds;
6. compare graph edges, output and input stamps, startNodes, and endNodes;
7. confirm the solution-hash and debug-symbol containers remain present.

A successful comparison is Structurally validated evidence. It does not show that the marker resolves, the contact is ready, subtitles merge, audio plays, or the exit returns to the questphase.

Previous: Scenes. Next: Actors and performers.