Smart Asset Finder

System Requirements

Hardware and software requirements for Smart Asset Finder.

Minimum Requirements

Hardware

  • CPU: x64 processor with AVX2 instruction set
  • RAM: 4GB minimum
  • Disk Space: 2GB for AI models
  • OS: Windows 10/11 x64

Software

  • Unreal Engine: 5.5 or higher
  • Visual Studio: 2019 or 2022 (if compiling from source)

Hardware

  • CPU: Intel i7 8th gen or AMD Ryzen 5 3600 or better
  • RAM: 8GB or more (16GB for large projects)
  • Disk Space: 5GB+ (for models and cache)
  • GPU: Not required (CPU inference only)

Compatibility

Unreal Engine Versions

VersionStatus
UE 5.7✅ Fully supported
UE 5.6✅ Fully supported
UE 5.5✅ Fully supported
UE 5.4🚧 Coming soon
UE 5.3🚧 Coming soon
UE 5.2 and earlier❌ Not supported

Operating Systems

PlatformStatus
Windows 10/11 x64✅ Fully supported
Windows 7/8❌ Not supported
macOS🚧 Planned for future release
Linux🚧 Planned for future release

Build Configurations

  • ✅ Development builds
  • ✅ Shipping builds (runtime-safe)
  • ✅ Debug builds

Project Size Guidelines

Initial indexing time is dominated by Unreal Engine thumbnail generation (loading assets and rendering 224x224 previews). Times vary significantly based on asset types — textures are faster, meshes and materials are slower due to shader compilation.

Project SizeAssetsEstimated Initial Index TimeRecommended RAM
Small< 1,0003-8 minutes4GB
Medium1,000 - 5,00015-30 minutes8GB
Large5,000 - 10,00030-60 minutes16GB
Very Large10,000+60+ minutes16GB+

After the first index build, the index is cached on disk and loads in seconds. Only new or modified assets need to be re-indexed.

Performance Notes

CPU

  • Inference uses ONNX Runtime with auto-threading (default: 50% of CPU cores)
  • You can configure thread count in plugin settings (0=auto, 1=single, 2-16=fixed)

Memory

  • Per 1000 assets: ~100MB RAM
  • AI Models: 337MB loaded once at startup
  • Thumbnails: ~200KB per asset (224×224 RGBA)
  • Feature vectors: ~2KB per asset (512 floats)

Disk I/O

  • Index saves/loads are asynchronous and don’t block the editor
  • Large projects (10k+ assets) may take a few seconds to load index on startup

Supported Asset Types

Smart Asset Finder is designed for visual asset search:

✅ Fully Supported:

  • Texture2D
  • Material
  • MaterialInstance
  • StaticMesh
  • SkeletalMesh (v2.0+)
  • ParticleSystem (Cascade)
  • PaperSprite

⚠️ Supported with Limitations:

  • NiagaraSystem (v2.0+) - Requires manual thumbnail setup in Niagara Editor

❌ Not Supported:

  • Audio assets (SoundWave, SoundCue)
  • Animation assets (AnimSequence, AnimBlueprint)
  • Blueprint classes (no visual representation)

Known Limitations

  1. Windows Only: Currently Windows x64 only. macOS/Linux support planned.
  2. CPU Inference: No GPU acceleration (uses ONNX Runtime on CPU).
  3. Visual Assets Only: Designed for assets with visual thumbnails.
  4. English UI: User interface is in English only (search supports multiple languages).
  5. Linear Search: Uses brute-force cosine similarity (future updates may add indexing optimizations).

Troubleshooting Requirements

”Plugin failed to load” Error

  • Cause: Missing ONNX Runtime dependencies
  • Solution: Ensure you’re on UE 5.5+ which includes ONNX Runtime 1.20.1

”Out of memory” during index build

  • Cause: Insufficient RAM for large projects
  • Solution:
    • Lower RebuildBatchSize in plugin settings (default: 10, try: 5)
    • Close other applications
    • Upgrade RAM

Slow index building

  • Cause: Too many CPU threads or low-end CPU
  • Solution:
    • Adjust OnnxInferenceThreads in settings (try 1 for slow CPUs)
    • Use smaller batch sizes

Next Steps