PDF2PPt Study

PDF2PPt Study turns scanned / image-only slide PDFs back into editable PowerPoint. Lecture slides often arrive as scanned or image-only PDFs: there's no text to select, so you can't fix a typo, update a title or reuse a bullet. Chat-with-PDF tools only read your slides; PDF2PPt Study rebuilds them. Every page becomes a real .pptx slide with editable text boxes placed where the original text was, on a background with the old text cleanly removed, with figures and diagrams kept as images and speaker notes on every slide. How it works - PyMuPDF renders each page; Meta Muse Spark Open on Baseten reads the page image and returns each text block with its position and style. - OpenCV inpainting / flat fill removes that text from the background; figures stay as images. - python-pptx writes editable text boxes on top, plus speaker notes (summary, key points, key terms). - Runs on Modal behind FastAPI with an access code, rate limits and a daily spend cap. Bonus study tools (from the same analysis): per-slide notes and a course outline, flashcards with Anki export, a quiz that explains each answer, and Q&A with slide citations. Try it - Live demo: https://ansenchen-12--pdf2ppt-study-web.us-east.modal.run (the built-in example needs no code; to upload your own PDF use access code: pixel-ember-orbit-5452) - Code: https://github.com/Ansenchen123/PDF2PPt/tree/hackathon/study