Image occlusion

Anki image occlusion, generated from your PDF

Turn labeled diagrams into Anki image-occlusion cards without drawing a single box. Upload a PDF, GuruOwl finds the figure pages, masks the labels, and exports a real .apkg.

1A labeled figure in your PDF
A labeled heart diagram from a PDF beside the same diagram with every label covered by a solid mask.
2The card you getevery label masked
A labeled heart diagram from a PDF beside the same diagram with every label covered by a solid mask.
Every label on the figure becomes a mask you have to recall. Nothing is left readable — the masks are placed from the label’s own coordinates in the PDF, so they cover the whole word, wrapped lines included.

Image occlusion is the highest-yield card type in anatomy, histology, and pathology — you cover the labels on a plate and recall them one at a time. The reason most people have three of them instead of thirty is the making: Anki’s built-in image occlusion and the classic Image Occlusion Enhanced add-on both hand you a screenshot and a rectangle tool, and every mask is a drag of the mouse. A single labeled plate is five minutes of clicking.

GuruOwl does that pass for you. Upload the PDF, and the figure pages are found by looking at each page directly — how much of it a figure actually covers, and whether the labels are scattered the way a diagram’s are rather than packed like prose. On those pages a vision model reads the labels, and each mask is then placed at the label’s real coordinates from the PDF text layer, not where a model guessed it should go. What comes out is a standard .apkg you review, edit, and study in Anki, GuruOwl, or both.

Finds the diagram pages itself

Point it at a whole chapter or a 300-page textbook. Pages are scored on figure coverage and label scatter, so plates get picked up and prose, contents pages and tables are left alone — you don’t mark anything by hand.

Masks land on the label, not near it

Vision models read labels well and place boxes badly, so only the text is taken from the model — every mask is positioned from the PDF’s own text layer. Labels that wrap onto two or three lines are covered as one mask instead of leaking their last word.

Real Anki image-occlusion export

Export a standard .apkg and open it in Anki Desktop, AnkiMobile, AnkiDroid or AnkiWeb — or study the same deck in GuruOwl with FSRS scheduling on web and iOS.

Editable before export

Every draft card is yours to tighten, merge, or delete against the source — so the final deck is what you actually need to review.

Frequently asked questions

Keep exploring

Turn your PDF into a deck you’ll actually review

50 free credits on signup — about one full PDF, no card required.