Why Can't You Select Text in Some PDFs?
You've probably encountered this frustration: you open a PDF, try to select some text to copy it, and nothing happens. The cursor doesn't highlight anything. The text looks perfectly normal, but it behaves like an image.
This happens because many PDFs — especially those exported from Google NotebookLM, Canva, Gamma, and other AI slide generators — render text as rasterized images rather than selectable text layers. The PDF contains pixel data, not character data. Your PDF viewer sees a picture of text, not actual text.
Common scenarios where this occurs include: NotebookLM slide exports, scanned documents, screenshot-based presentations, infographics with embedded text, and AI-generated visual content.
The Solution: Local AI OCR
Optical Character Recognition (OCR) can read the pixels in your PDF and reconstruct the text content. DeckEdit runs OCR entirely in your browser using state-of-the-art AI models that process your files locally without uploading them to any server.
Step 1: Upload Your PDF or Image
Open DeckEdit and drag your PDF or image file into the upload zone. DeckEdit supports PDF, PNG, JPEG, GIF, and WebP files up to 30MB and 50 pages. No account or registration is required.
DeckEdit offers two modes depending on your goal. PPTX Conversion mode creates an editable PowerPoint file where every detected text element becomes a real text box you can click and edit. Image PDF Edit mode lets you click directly on detected text regions to modify or delete them in-place, then export in the original format.
After processing, download your file. In PPTX mode, you get a PowerPoint presentation with editable text boxes and the original slide background preserved. In Edit mode, you get a clean PDF or image with your modifications applied.
Why Text Becomes Unselectable
Understanding why this happens helps you avoid it in the future. PDF files can contain text in two fundamentally different ways:
When NotebookLM generates slides, it creates beautiful layouts but exports them as rasterized images. The text you see is actually rendered into the image layer, making it impossible to select or edit without OCR.
Tips for Best Results
- Export from NotebookLM as PDF rather than taking screenshots. PDFs preserve higher resolution, which improves OCR accuracy.
- Enable hardware acceleration in your browser (Settings → System → Hardware Acceleration). This enables faster AI processing.
- For multi-page documents, use DeckEdit's page selector to convert specific pages rather than processing the entire document at once.
- After conversion, review the text in PowerPoint. OCR achieves 95%+ accuracy on clean slides, but unusual fonts or very small text may need minor corrections.
No Software Required
DeckEdit runs entirely in your browser: there is nothing to install, no account to create, and no files leave your device. Whether you are on a Chromebook, a managed work laptop with restricted installs, or simply want to avoid yet another desktop app, you can start editing PDF slides immediately at deckedit.com. AI models are cached after first use for offline operation.
More Guides
NotebookLM Makes the Slides. Here Is How to Actually Edit Them.
NotebookLM hands back a PDF with the text baked into the image. Here is the honest way to turn it into a truly editable .pptx, in your browser.
Convert an Image to Editable PowerPoint: JPG or PNG to a Real PPTX
Convert an image to PowerPoint where every heading and bullet is a real editable text box, not a flat picture. Free, in browser, no upload.
How to Edit NotebookLM Slides: The Complete Step-by-Step Guide
Gemini Notebook (formerly NotebookLM) creates beautiful slides, but the text is locked inside images. This guide shows you exactly how to make every word editable — for free, in under 3 minutes.