PDF Processing

Niobium uses PyMuPDF to extract embedded images from PDF files. All PDF processing works without an API key.

Extract images only¶

Export images from a PDF without creating any Anki cards:

niobium -pin /path/to/lecture.pdf -pout /path/to/output

Images are saved as numbered files in a timestamped subdirectory.

Extract from specific pages¶

niobium -pin /path/to/lecture.pdf --page 5-10 -pout /path/to/output

Pages are 1-indexed. --page 1 is the first page. --page 3-7 processes pages 3 through 7 inclusive.

Create cards directly from a PDF¶

Process a PDF and push cards to Anki in one step:

niobium -pin /path/to/lecture.pdf -deck "Lecture 3"

Or export as .apkg:

niobium -pin /path/to/lecture.pdf -apkg ./output

Or from specific pages:

niobium -pin /path/to/lecture.pdf --page 5-10 -apkg ./output

Two-step workflow¶

For large PDFs, extract first to review images before processing:

# Step 1: extract
niobium -pin /path/to/lecture.pdf -pout /path/to/images

# Step 2: review, delete unwanted images, then process
niobium -dir /path/to/images -apkg ./output

This gives you the opportunity to delete irrelevant images (title slides, blank pages) before card creation.

Page selection (--page)¶

The --page flag limits which pages are processed:

# Single page
niobium -pin notes.pdf --page 5 -apkg ./output

# Page range
niobium -pin notes.pdf --page 3-7 -apkg ./output

This works with all PDF workflows: image extraction (-pout), card creation (-deck, -apkg), and AI modes (--smart).

AI-powered PDF processing¶

When combined with --smart --page, Niobium enters a generation-first pipeline where Claude sees the full rendered page and creates cards from scratch (cloze, basic, and image occlusion). See Smart Generation.

niobium -pin lecture.pdf --page 1-10 --smart -apkg