Gap Analysis

Interoperability, standards, and research gaps in AI-generated virtual environments

Interoperability Gaps

These gaps affect the ability to move content between systems.

Gap IDDescriptionAffected PartiesSeverity
GAP-01Video-based world models produce no exportable 3DGame developers wanting Genie/Oasis contentCritical
GAP-02Gaussian splat format fragmentation across toolsAll neural-native tools, game enginesHigh
GAP-03Generated meshes have topology errors breaking physicsAll geometry generators, game enginesHigh
GAP-04Platform-locked content (Roblox, Horizon, Fortnite)Creators wanting cross-platform portabilityHigh
GAP-05No navmesh generation in most toolsGame developers requiring NPC navigationMedium
GAP-06Physics material properties not transferredSimulation users, robotics applicationsMedium
Critical GAP-01

# Video-Based Outputs Non-Exportable

Video-based world models (Genie 3, Oasis, Odyssey) generate interactive video streams but cannot export 3D geometry. Game developers cannot use this content in engines.

Affected Areas

  • Genie 3
  • Decart Oasis
  • Odyssey

Recommendations

  • Monitor video-to-3D reconstruction research
  • May remain fundamental paradigm limitation
High GAP-02

# Gaussian Splat Format Fragmentation

Each neural-native tool uses different Gaussian splat formats. No standard interchange format exists, and KHR_gaussian_splatting has no ratification timeline.

Affected Areas

  • World Labs Marble
  • Luma AI
  • All neural tools

Recommendations

  • Support KHR_gaussian_splatting ratification
  • Advocate for timeline
High GAP-03

# Topology Errors Breaking Physics

AI-generated meshes commonly have non-manifold geometry, T-junctions, and other errors that break physics engines and collision detection.

Affected Areas

  • All geometry generators
  • Game engines
  • Physics simulations

Recommendations

  • Define quality validation profiles
  • Develop mesh repair standards
High GAP-04

# Platform Lock-In

Content created in Roblox, Meta Horizon, and Fortnite cannot be exported to other platforms. Creators are locked into single ecosystems.

Affected Areas

  • Roblox creators
  • Horizon creators
  • Cross-platform developers

Recommendations

  • Interoperability standards pressure
  • MSF advocacy

Standards Gaps

These are gaps in formal standardization.

Gap IDDescriptionPotential SDOComplexityPriority
SGAP-01No glTF extension for AI generation metadataKhronos (MSF proposal)MediumHIGH
SGAP-02C2PA has no 3D format supportC2PA CoalitionHighHIGH
SGAP-03No OpenUSD AI generation schemasAOUSDMediumMEDIUM
SGAP-04KHR_interactivity lacks semantic vocabularyKhronosHighMEDIUM
SGAP-05No determinism specification for multiplayerMSF/KhronosMediumMEDIUM
SGAP-06No portable world package formatMSF/Khronos/AOUSDHighHIGH
SGAP-07No quality validation profile standardMSF/KhronosMediumHIGH

SGAP-01: AI Generation Metadata Extension

The highest priority standards gap.

No standard exists for capturing:

  • Generation prompt
  • Seed/random state
  • Model version
  • Quality metrics
  • Generation parameters

Proposed Solution: MSF_ai_generation_metadata glTF extension

SGAP-02: C2PA 3D Format Support

C2PA (Content Authenticity) covers:

  • Images (JPEG, PNG, etc.)
  • Video (MP4, etc.)
  • Audio

C2PA does not cover:

  • glTF / GLB
  • USD / USDZ
  • OBJ / FBX
  • Any 3D format

Proposed Solution: MSF advocacy to C2PA Coalition

SGAP-06: World Package Format

No portable format exists for packaging:

  • Scene geometry
  • Textures and materials
  • Physics data (collision, navmesh)
  • AI generation metadata
  • Behavioral scripts

Complexity: High - requires multi-SDO coordination (Khronos, AOUSD)

Research Gaps

These gaps require research before standardization is possible.

Gap IDQuestionWhy It MattersSuggested Approach
RGAP-01Can video-to-3D reconstruction produce game-ready output?Would bridge paradigmsMonitor DeepMind, academic research
RGAP-02What is acceptable topology quality threshold for physics?Defines validation profilesBenchmark study with game studios
RGAP-03Can AI generation be made deterministic across GPU vendors?Enables multiplayerTechnical feasibility study
RGAP-04What generation metadata is needed for reproducibility?Informs schema designCreator workflow research
RGAP-05How do creators actually use AI generation in production?Validates use case assumptionsUser research study

Gap Priority Matrix

Gap Priority Matrix

Impact
High Medium Low
Critical
GAP-01 SGAP-06 GAP-03
High
SGAP-01 SGAP-07
Medium
GAP-05
High
SGAP-04 SGAP-02
Medium
SGAP-05
Low
SGAP-03
Medium
RGAP-01
Low
RGAP-04
Minimal
RGAP-02
High Medium Low
Urgency
Priority Levels:
Critical
High
Medium
Low
Minimal

3 critical gaps requiring immediate attention, 6 high priority items.

Priority Categories

Immediate (2026)

  • SGAP-01: AI metadata extension (Medium effort, High impact)
  • SGAP-07: Validation profiles (Medium effort, High impact)

Near-term (2026-2027)

  • SGAP-06: World package format (High effort, High impact)
  • SGAP-02: C2PA advocacy (High effort, Medium impact)

Ongoing

  • GAP-01: Monitor video-to-3D research
  • GAP-03: Support KHR_gaussian_splatting

Priority Recommendations

1. Immediate Action (2026)

GapActionOwnerTimeline
SGAP-01Draft MSF_ai_generation_metadata extensionMSF AI x Metaverse WGQ2 2026
SGAP-07Define geometry quality validation profilesMSF AI x Metaverse WGQ3 2026

2. Near-term Action (2026-2027)

GapActionOwnerTimeline
SGAP-06Convene world package format discussionMSFQ4 2026
SGAP-02Formal engagement with C2PA CoalitionMSFQ1 2026

3. Ongoing Monitoring

GapActionOwner
GAP-01Track video-to-3D reconstruction researchMSF Research
GAP-02Support KHR_gaussian_splatting ratificationMSF Liaison
GAP-03Develop mesh quality benchmarksMSF + Game Studios