vawer-flow-power/luid-dynamics-wind-telemetry-goa

Dual-Stream Fluid Dynamics Dataset for Neural Physics Engine Training

💡 Economic Value & Compute Optimization

This is not an aesthetic nature stock video. This dataset is a highly calibrated testbed designed for training Neural Physics Engines and fine-tuning generative video models.

By providing mathematically synchronized spatial-temporal streams and ground-truth process parameters, this dataset eliminates pre-processing overhead and cuts model convergence/loss optimization machine time (GPU-hours) by an estimated 15% to 70% during the fine-tuning phase.

---

🛠️ Dataset Architecture (Sample: set_0111)

Each data asset represents a single-position, fixed-horizon, non-interrupted sequence captured in dual-phase parameters to decouple fluid kinematics from surface topology.

📂 Repository Structure & Files for set_0111:

• set_0111/video_0111_clip.mov — Stream A (Kinematics & Optical Flow): 10 min, HD @ 100 FPS. Captured at high temporal resolution to allow neural networks to compute precise sub-millisecond velocity fields and wave propagation vectors without motion blur artifacts.
• set_0111/video_0111_frame.mov — Stream B (Structural Topology): 5 min, 4K @ 25 FPS. Captured with ultra-high shutter speed (minimal exposure time) to "freeze" fluid structures, providing crisp, pixel-perfect ground truth for fractal foam geometries, spray dynamics, and water-surface boundary conditions.
• set_0111/audio_0111.wav — Synchronized high-fidelity spatial audio for multimodal video-to-audio correlation training.
• set_0111/METADATA_SET_0111.json — Full technical, environmental, and geographic metadata log for this specific process.
• set_0111/README_0111.md — Original author's documentation and field notes for the sequence.

---

⚖️ Commercial & Research Licensing

This sample dataset is released under the Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) for academic and independent R&D verification.

Commercial Licensing & Full Archive (117+ Calibrated Sets): The complete pipeline includes extended 6K N-RAW / ProRes RAW datasets from contrasting ecosystems (Baltic Sea Coast, Tallinn, Estonia vs. Indian Ocean, Goa). For commercial procurement, custom разметка, or exclusive physics-engine training integration, please contact the author directly via Hugging Face profile or LinkedIn.