AI-Powered Processing | pg2b3dm-glb | Tech Maven Geospatial

Mesh Simplification

Intelligent Polygon Reduction

Our AI-powered mesh simplification uses deep learning to reduce polygon count while preserving visual quality. Unlike traditional decimation algorithms that uniformly reduce geometry, our AI understands the semantic importance of different mesh regions.

This means critical features like building edges, architectural details, and distinctive structures are preserved while redundant interior polygons are aggressively reduced.

Up to 90% polygon reduction with 95%+ visual fidelity
Semantic-aware preservation of important features
Boundary and UV preservation options
Configurable target face count
ONNX Runtime for fast inference

                        
# AI mesh simplification

pg2b3dm-glb --data-source ThreeDModel --model-input building.obj \

  --enable-ai --ai-simplify \

  --target-face-count 5000 \

  --preserve-features true \

  --preserve-boundaries true

🔺

Texture Enhancement

ESRGAN Texture Upscaling

Breathe new life into low-resolution textures with ESRGAN (Enhanced Super-Resolution Generative Adversarial Networks) powered upscaling. Transform blurry, pixelated textures into crisp, detailed surfaces.

Our implementation supports 2x, 4x, and 8x upscaling factors with automatic noise reduction and detail enhancement. Perfect for drone mapping data, photogrammetry outputs, and legacy 3D models.

2x, 4x, and 8x upscaling options
Noise reduction and detail enhancement
Automatic seam preservation
Batch processing for large tilesets
Quality threshold controls

                        
# AI texture upscaling

pg2b3dm-glb --data-source SLPK --slpk-file buildings.slpk \

  --enable-ai --ai-upscale-textures \

  --texture-scale 4 \

  --max-texture-size 4096

🖼️

Semantic Processing

Material & Component Classification

Automatically identify building materials and components using computer vision. Our semantic segmentation models can distinguish between roofs, walls, windows, doors, and other architectural elements.

This enables intelligent LOD generation, material-specific optimizations, and automated attribute tagging for downstream GIS and BIM workflows.

Automatic material identification (concrete, glass, brick, etc.)
Building component segmentation
Semantic LOD generation
Attribute tagging and metadata enrichment
PBR material enhancement

                        
# Semantic-aware processing

pg2b3dm-glb -h localhost -U postgres -d mydb -t buildings \

  --enable-ai --ai-semantic \

  --ai-materials --quality-threshold 0.9

🏷️

Pipeline

Complete AI Enhancement Pipeline

                    
# Full AI enhancement pipeline

pg2b3dm-glb --data-source ThreeDModel --model-input city.obj \

  # Enable AI processing

  --enable-ai \

  # Mesh simplification

  --ai-simplify \

  --target-face-count 10000 \

  --preserve-features true \

  # Texture upscaling

  --ai-upscale-textures \

  --texture-scale 4 \

  # Material classification

  --ai-materials \

  --ai-semantic \

  # Quality settings

  --quality-threshold 0.95 \

  --output-format glb

🤖 AI-Powered Processing

Intelligent Polygon Reduction

ESRGAN Texture Upscaling

Material & Component Classification

AI Processing Performance

Complete AI Enhancement Pipeline

Transform Your 3D Data with AI