Is this AI generating my cards?

No. Detection marks where text sits on each slide; recognition reads it for cloze cards. Neither writes anything. If the teacher never wrote it, it is not on a card.

Does it work on whiteboard lectures?

For occlusion, no — handwriting on a board is not reliably detectable, and we will not ship a feature that mostly fails. Whiteboard lectures export as full-page cards.

Will re-exporting a lecture duplicate my cards?

No. Cards carry a stable ID derived from the course, the lecture and the slide, so a re-export replaces the old cards and keeps your review history.

Can I edit the cards in Anki afterwards?

Yes. It is an ordinary deck: standard note types, your media, your scheduling. Anki’s own editor can re-edit occlusions on 23.10+.

Full-page cards are live · image occlusion next

Lecture video in,
an Anki deck out

The slides your teacher actually showed, as cards. Every card carries the second of the lecture it came from, so the card you keep getting wrong can send you back to the moment it was explained. Nothing on a card was written by a model: what is on the card is what was on the screen.

Open the converter

Open a recording and pick “Anki deck” in the export menu. Occlusion cards arrive with the detection update.

Three kinds of cards, one run

Free
Full-page cards
Front shows the slide’s title, back shows the whole slide. The simplest thing a deck can be — and already a deck you can review on your phone.
Paid
Image occlusion
Text on the slide is detected and covered; one card per box. Cover everything and guess one, or cover one and name it. The way anatomy diagrams and pathways are studied.
Paid
Cloze deletions
The slide’s text is recognised; you tap the words that matter and the card blanks exactly those. The rest of the page stays as it was shown.

A slide from a real run with detected text boxes covered; one box is open to guess. A chip reads 04:12. — Real output, not a drawing: a slide from one of our demo recordings with the boxes where the detector found text. The header is not covered — it appears on every slide of the talk, so it is left alone automatically. Tap a box to reveal it.

Cards that cite their source

Only a video can give a card a place in time. Each card’s back names the moment of the lecture its slide came from — 23:15 — and tapping it returns to that point in the recording, to hear the explanation again. A tool that starts from a PDF cannot do this: the recording is where the deck comes from.

Found, not written

Detection finds where text sits on the slide; it never composes content. A card cannot contain a sentence that was never on screen, and nothing is summarised, paraphrased or invented. The curation step exists because detection is a machine’s best guess — you can drop a box, merge two, or draw one it missed, in about ten minutes for a whole lecture.

It runs in your browser, like everything else here. The lecture is never uploaded.

From recording to review

In Video2Any

Drop the lecture recording; the slides are found as they are today.
Export “Anki deck”. For occlusion cards, check the suggested boxes first: drop, merge or draw.
Download the .apkg. The deck is named Course :: Lecture N, tagged with the lecture and the date.

In Anki

Double-click the .apkg, or File → Import. No add-ons to install.
Re-exporting the same lecture updates your existing cards in place — scheduling history is kept, nothing is duplicated.
Occlusion cards need Anki 23.10+ on the desktop, AnkiDroid 2.17+ or a current AnkiMobile on the phone; full-page cards work everywhere.

What it costs

Full-page cards are part of the free tier, inside the same free minutes as every other export.

Image occlusion and the curation editor are included in the paid plans — the same plans that remove the length limit.

See pricing

Questions

Is this AI generating my cards?: No. Detection marks where text sits on each slide; recognition reads it for cloze cards. Neither writes anything. If the teacher never wrote it, it is not on a card.
Does it work on whiteboard lectures?: For occlusion, no — handwriting on a board is not reliably detectable, and we will not ship a feature that mostly fails. Whiteboard lectures export as full-page cards.
Will re-exporting a lecture duplicate my cards?: No. Cards carry a stable ID derived from the course, the lecture and the slide, so a re-export replaces the old cards and keeps your review history.
Can I edit the cards in Anki afterwards?: Yes. It is an ordinary deck: standard note types, your media, your scheduling. Anki’s own editor can re-edit occlusions on 23.10+.

Bring your next lecture.