10 ML
Main Components
Read processed CSV datasets back into Grasshopper. Supports mag_U and all spatial features.
Download an ONNX wind-prediction model from HuggingFace for the Wind Predictor component. Yel 2.0 is public; Esen 1.0 and Poyraz 1.0 need a HuggingFace token. All are 8-channel Wind Predictor models. (Yel 1.0 is a different architecture — the GAN image model used by GAN Predict via its API — and cannot be loaded here.) Models cache in ~/SUS_LAB/ and are reused on subsequent runs.
Run ONNX wind-field prediction end-to-end. Computes SDF, building height, Zrelative, U/Uref, direction features from geometry, assembles the 8-channel input tensor, runs ONNX inference, and outputs predicted wind speeds. Supports legacy 1ch (U), 2ch (U + k) and new 4ch (U + k + Uroof + kroof) models.
Calculate Pedestrian Wind Comfort using predicted wind fields from the ONNX model.
Predict a pedestrian wind-speed field from buildings using the Eddy3D GAN (no CFD run). Sends the geometry to the GAN API and returns wind speeds + a colored result mesh.
Resample per-direction wind-magnitude fields onto a new point grid (nearest-neighbour average, with per-direction rotation). Prepares grids for GAN applications.