DGOne - DGOne Bounding Box

🔍Overview

DGOne Bounding Box & Tag is an enterprise-grade high-performance screen-space bounding box rendering plugin, optimized for large-scale scenes, providing industrial-grade precision and cinematic-quality visual effects.

🚀 Core Technical Features

1️⃣ Dual-Engine Calculation Architecture

🔷 Auto Bounds Engine

- Zero-Configuration Launch: Automatically extracts 8 corner points from Actor's RootComponent.Bounds

- Intelligent Boundary Recognition: Auto-adapts to collision mesh, physics bounds, visual bounds

- Use Cases: Scene props, vehicles, buildings, static meshes (3000+ objects)

- Performance: ~0.001ms/object (500x faster than manual sampling)

- Precision: Perfectly matches actual object size


🔶 Manual Sampling Engine

- Skeletal-Level Precision: Supports any number of SceneComponents as sample points

- Dynamic Tracking: Real-time tracking of character bones (head, chest, hands, feet, weapons)

- Custom Shapes: Full control over bounding box shape and coverage

- Use Cases: Characters, enemies, complex machinery, multi-part objects

- Performance: ~0.5-1ms/character (with occlusion detection)

- Precision: Pixel-perfect tracking

2️⃣ Triple Performance Modes

⚡ Static Mode (Static Objects)

- Zero Tick Overhead: Completely disables Tick, near-zero CPU usage

- Manual Update Trigger: Only calls `RequestUpdate()` when needed

- Camera-Relative Motion Detection: Auto-updates when camera moves (optional)

- Use Cases: Scene props, furniture, decorations (supports 3000+ objects)

- Performance: ~0.001ms/object, 60+ FPS


🔥 Dynamic Mode (Dynamic Objects)

- Auto Tick Updates: Real-time tracking of moving objects

- Configurable Update Frequency: 0.0s (every frame) to 1.0s (1Hz)

- Intelligent Occlusion Detection: Prevents wallhacks (anti-cheat)

- Use Cases: Characters, enemies, NPCs, moving targets

- Performance: ~0.5-1ms/character, 50-60 FPS


🎯 Motion Triggered Mode (Motion-Triggered)

- Smart Motion Detection: Updates only when movement/rotation exceeds threshold

- Four Advanced Occlusion Strategies: Simple/Spatial/Temporal/Adaptive

- Adaptive Sampling: Intelligently adjusts detection frequency based on visibility history

- Use Cases: Occasionally moving objects (doors, elevators, patrolling enemies)

- Performance: ~0.2ms/object, 60 FPS

3️⃣ Six-Dimensional Optimization System


🎯 Frustum Culling

- Basic Culling: Dot product algorithm, ~0.0001ms/object

- Precise Culling: FConvexVolume algorithm, ~0.0005ms/object

- Auto Strategy Selection: Automatically selects optimal algorithm based on performance mode

- Configurable Edge Tolerance: 5-30° to prevent edge popping

- Performance Gain: CPU reduction 40-60% (zero overhead for objects behind player)


🔍 Sample Points Threshold

- Dual-Check Mechanism: Checks total count at initialization, visible count after projection

- Intelligent Quality Control: Automatically skips low-quality objects with insufficient sample points

- Configurable Threshold: 1-16 points (recommended: 2-4 for characters, 1-2 for small objects)

- Performance Gain: Early culling of invalid calculations, reduces 20-30% overhead


📏 Distance Culling

- Configurable Max Render Distance: 0-infinity (recommended 5000-10000cm)

- Early Exit Optimization: Immediately returns when beyond distance, skips all calculations

- Use Cases: Open world, large scenes


🎥 Active Camera Detection

- Multi-Camera Scene Optimization: Only renders current active/controlled camera view

- Zero Cache Overhead: Direct access to PlayerCameraManager, no cache validation

- Supports All Camera Types: PlayerCameraManager/CameraActor/CineCameraActor

- Performance Gain: Multi-camera scenes CPU reduction 85-90%


📐 Camera-Relative Motion Detection

- Triple Detection Mechanism:

  - World-space camera movement (100cm threshold)

  - Camera rotation (10° threshold)

  - Screen-space object movement (50 pixel threshold)

- Distance-Adaptive Threshold: Near objects more sensitive, far objects more lenient

- Use Cases: Intelligent updates in Static mode


🎲 Advanced Occlusion Culling

- Four Strategies:

  - Simple: Round-robin line tracing

  - Spatial: Distance-based priority (near objects high-frequency detection)

  - Temporal: Time-sliced caching (trust cached results)

  - Adaptive: Intelligent sampling (adaptive based on visibility history)

- Configurable Budget: 1-32 checks/frame (recommended 4-8)

- Performance Gain: Occlusion detection reduced 60-75%

4️⃣ Machine Vision Noise System

🎬 Core Features

- Dual Noise Engines: Random noise (realistic) + Perlin noise (smooth organic)

- Per-Corner Independent Jitter: 4 corners with independent noise, slight box deformation

- Distance Attenuation: Full intensity at close range, auto-weakens at distance

- Smooth Interpolation: Configurable smoothing factor (0-1), avoids sudden changes

- High-Frequency Updates: 0.1-60Hz adjustable, simulates different sensors


🎮 Application Scenarios

- Tactical FPS: Simulate thermal imaging, night vision instability

- Sci-Fi Games: Simulate AI recognition systems, robot vision

- Cinematic Narrative: Enhance immersion and realism

- Competitive Games: Simulate radar interference, signal instability


⚙️ Technical Parameters

- Intensity: 0-20 pixels (recommended 3-5)

- Frequency: 0.1-60Hz (recommended 10-20)

- Performance: +0.0001ms (near-zero overhead)

5️⃣ Modern UI Features

[Rounded Corners System]

- Triangle Fan Algorithm: Reduces 70% trigonometric calls

- Block Fill Optimization: Reduces 50% draw calls

- Independent Control: Box and Tag background separately configurable


[Precise Text Alignment]

- 9-Grid Alignment: Left/Center/Right × Top/Center/Bottom

- Single Measurement: 39% performance optimization

- Auto Edge Anchor: Automatically selects best anchor


[Initial Tag Text]

- Zero-Code Configuration: Set directly in editor

- BeginPlay Auto-Apply: No Blueprint code needed

- Localization Support: FText type, multi-language ready