← Back to Home

Ignis Divinus

Milestone Report

Real-time 3D Fire Simulation Progress Update

Chenyang Zhang
Huachen Qu
Muze Du
Yijun Xiang

Executive Summary

We have successfully implemented a real-time 3D fire simulation system using GPU-accelerated fluid dynamics. Our system achieves physically realistic flame behavior through Navier-Stokes equations, volumetric rendering with blackbody radiation, and advanced post-processing effects. The current implementation runs at 30+ FPS on high-quality settings with a 96³ simulation grid.

96³
Grid Resolution
30+
FPS (High Quality)
256
Ray Marching Steps
800-3000K
Temperature Range

Accomplishments

3D Fluid Simulation

Implemented stable Navier-Stokes solver with Semi-Lagrangian advection and temperature-driven buoyancy

Volume Rendering

Ray marching with physically-based blackbody radiation color mapping (800K-3000K)

Heat Refraction

Screen-space heat distortion using temperature gradients with temporal filtering

Real-time Performance

Achieved 30+ FPS at high quality, 60+ FPS at medium quality on RTX GPUs

Interactive Controls

GUI with real-time parameter adjustment, camera controls, and quality presets

Performance Monitoring

FPS graphs, memory tracking, and detailed frame time analysis

Technical Implementation Details

Fluid Simulation Core

  • Grid Resolution: 96³ voxels with adaptive time stepping
  • Advection: Semi-Lagrangian method with tri-linear interpolation
  • Pressure Solver: Jacobi iteration (40 iterations) for incompressibility
  • Buoyancy Model: Temperature-driven with adjustable α and β parameters
  • Boundary Conditions: No-slip at bottom, open boundaries at top

Rendering Pipeline

  • Volume Ray Marching: Adaptive step size (0.005 base) with early termination
  • Color Model: Blackbody radiation LUT with smooth interpolation
  • Emission: Temperature and density-dependent with adjustable scale
  • Post-processing: Bloom effect, filmic tone mapping, and vignetting
  • Heat Refraction: Screen-space distortion based on temperature gradients

Live Demo

Flame Simulation Demo

3D Volumetric Flame

Real-time flame simulation with natural turbulent motion and realistic color gradients from dark red to bright yellow-white.

Heat Refraction Demo

Heat Distortion Effect

Screen-space refraction creates realistic heat haze, distorting the background based on temperature fields.

Interactive Features

  • Camera Controls: Full 3D orbit and zoom with smooth transitions
  • Flame Source: Adjustable position (X, Z coordinates) in real-time
  • Physical Parameters: Temperature scale, fuel injection rate, buoyancy strength
  • Rendering Options: Absorption coefficient, emission scale, bloom intensity
  • Quality Presets: Low (120 FPS), Medium (60 FPS), High (30 FPS), Ultra (15 FPS)
  • Visualization Modes: 3D volume rendering or 2D slice views (density/temperature/velocity)

Video & Presentation

1-Minute Progress Video

Video demonstrates current progress with all team members presenting different aspects of the project.

Presentation Slides

Download Slides (PDF)

Reflection & Next Steps

Progress vs. Original Plan

We are on track with our proposed timeline. All baseline features have been successfully implemented during Week 1 as planned. The system is stable, performant, and ready for enhancement features in Week 2.

Completed (Week 1)

  • ✓ 3D fluid simulation with stable solver
  • ✓ Volumetric rendering with blackbody radiation
  • ✓ Heat refraction post-processing
  • ✓ Interactive GUI and controls
  • ✓ Performance monitoring system
  • ✓ 30+ FPS at high quality settings

Updated Work Plan (Week 2)

Priority 1: Core Enhancements

  • Turbulence Modeling: Implement vorticity confinement for more detailed flame dynamics
  • Resolution Scaling: Test 128³ grid with performance optimizations
  • Advanced Advection: Compare BFECC vs Semi-Lagrangian methods

Priority 2: Visual Effects

  • Particle System: Add sparks using GPU-based particle simulation
  • Smoke Integration: Density-based smoke when temperature drops
  • Enhanced Materials: PBR shading for environment objects

Priority 3: Polish & Documentation

  • Benchmarking Suite: Automated performance testing across quality levels
  • Video Production: Record high-quality demonstration sequences
  • Code Documentation: Clean up and document for reproducibility

Key Achievements

Our implementation successfully demonstrates that real-time, physically-based fire simulation is achievable on consumer GPUs. The combination of Taichi's GPU acceleration and careful algorithm optimization has resulted in a system that balances visual quality with interactive performance.

Next Milestone: Final presentation on August 12 with enhanced features and polished demo.