We provide code to synthesize handwriting (bottom) in a target style (top) via learned spaces of style and content, and can exploit any available reference samples (middle) to improve output quality.