Currently, only PDFs with raw text in them get properly OCRed and added to the search index. We should support PDFs that are just images or contain images.
Václav Vančura
Are you using standard OCR? Have you thought about using Docling for more effective transcription?
