Batik MeanFlow
Generative Flow Matching and Mean Flow applied to manifold learning and Indonesian Batik pattern synthesis
Batik MeanFlow investigates the application of continuous flow matching, Mean Flow objectives, and Riemannian geometry to the generation and synthesis of complex Indonesian Batik cultural motifs.
Private Research Repository
Highlights
- Cultural Heritage & Generative Modeling: Preserving and synthesizing intricate mathematical symmetries and geometric fractals found in traditional Indonesian batik motifs.
- Mean Flow Objectives: Utilizing straight flow paths and velocity field parameterizations to generate high-resolution textile textures in few sampling steps.
- Dataset Curation: Preprocessing, vectorizing, and standardizing diverse regional batik patterns (Kawung, Parang, Megamendung) for deep generative learning.