HandWind-Sim++

• 📄 Korea Society of Computer and Information (KSCI) Conference 2026
  ⤷ "A Web-Based Cloth Simulation for Learning-Based Gesture-Controlled Wind Interaction"
  

---

HandWind-Sim++ is a web-based interactive system where cloth and wind fields respond in real time to a user’s hand gestures.

Offline Training (AI part) 🧠

• ① Hand gesture video capture
• ② Asynchronous gesture–cloth data collection
• ③ Vector field extraction from cloth motion
• ④ K-Means clustering → representative wind field templates
• ⑤ LSTM training to predict template weight distributions from hand motion

Real-Time Inference & Simulation (Web part) 🖥️

• ① Continuous hand gesture input
• ② LSTM-based inference → template probability weights
• ③ Weighted blending of precomputed wind field templates
• ④ Application to WebGL-based fluid and cloth simulation
• ⑤ Real-time visualization via WebSocket

Demo video (Click!)

Fluids-2D References

1. Jos Stam. Stable Fluids. In Proceedings of SIGGRAPH. 1999.
2. Jos Stam. Real-Time Fluid Dynamics for Games. Proceedings of the Game Developer Conference. 2003.
3. Mark Harris. Fast Fluid Dynamics Simulation on the GPU. In GPU Gems: Programming Techniques, Tips, and Tricks for Real-Time Graphics (Chapter 38). 2004.

Cloth Simulation References

1. Müller, Matthias, Bruno Heidelberger, Marcus Hennix, and John Ratcliff. "Position based dynamics.“ 2007