tanzhouhu/pdf-figure-vector is a Python tool that converts figures in academic paper PDFs into self-contained vector SVGs and editable PPTX files. It redraws the shapes inside the PDF to produce an SVG with fonts and images embedded, and it saves the PPTX in a form where shapes can be edited and drawn directly on slides. It keeps the original element order and overlap order (z-order) as they are.
Standalone PNG/JPG/WebP/BMP/TIFF images can also be converted, through the VTracer vector tracer, into SVG and into PPTX using PowerPoint native freeform shapes. A Skill that can be used in Claude Code and Codex is distributed alongside. Extraction, live preview, and pixel-level comparison with the PDF rendering result can all be checked in a single workflow, and the README explains the font and image embedding method, the extraction and preview procedures, and the pixel comparison and slide verification methods.
That said, photo tracing results stay approximate, and the generated slide files can grow large or processing can become slow. No test suite was found in the repository tree, so the quality assessment is kept conservative. No license is specified and there are no public releases, and it has 52 stars. Its distinguishing feature is a workflow that reconstructs diagrams as editable slide shapes and compares them against the original rendering. It is still too early to judge the use case of simply transferring high-fidelity graphs as they are.