Reproduce “Vector Scaffolding: Inter-Scale Orchestration for Differentiable Image Vectorization” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Jaerin Lee, Kanggeon Lee, Kyoung Mu Lee
Vector Scaffolding stages image vectorization by fitting broad closed Bézier curves before adding smaller curves in high-error regions, while interior gradients support curve positioning. On Kodak, it reaches 28.30 dB PSNR with 1,024 curves, versus 26.91 dB for Bézier Splatting, and the authors report 2.5× faster wall-clock optimization. These results concern closed-curve image reconstruction; the metric does not measure editability.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.