🔮 Mapping Photography Portfolios by Visual Similarity With Machine Learning

January 2026

#Machine Learning #Visualization #Three.js

I scraped the profile of @withluke and created these visualizations optimized to load around 2250 images. Click to see the live demo.

The methods used were inspired by:

3D View — Google Arts Experiments TSNE Viewer
Spherical View — IsoMatch: Creating Informative Grid Layouts
Grid View — IsoMatch: Creating Informative Grid Layouts

Prior to this, I built a Python implementation, which you can find here: Grid-Layout-Image-Artwork. This pipeline can be replicated for any image dataset (including any Instagram profile) to create a grid layout using PCA or non-linear dimensionality reduction methods such as t-SNE or UMAP.