When photo import is the right tool
There is a whole category of study material that never exists as a file you can import: the library book you can't take home, the slide already on the screen, the whiteboard about to be wiped, the printed handout, the scanned past paper with no text layer. That's what this is for.
If your material is a proper document, use PDF import instead — it reads the text directly and is more accurate than any camera.
Getting a clean read
OCR quality decides card quality, and OCR quality is mostly about the photo. Four habits fix nearly every bad read:
- Flatten the page. Curved pages near the spine distort the text lines; press the book open or shoot one page at a time.
- Shoot square to the page, not at an angle. Hold the phone parallel to the paper rather than leaning over it.
- Watch your own shadow. The most common bad photo is a page lit by a lamp behind you, with your head shading half of it.
- Fill the frame with the text you care about. Cropping tight to one column or one section is better than capturing a whole spread from a distance.
One generation a day, free
The free tier gives you 1AI generation per day, so pick the page that matters. If you're photographing your way through a whole module in one evening, Premium removes the cap. Reviewing decks you already have is never limited, online or off.
Handwriting, diagrams and the honest limits
Tidy handwriting usually reads fine. Rushed lecture scrawl often doesn't, and no app on the market genuinely fixes that — expect to correct a card or two.
Diagrams are a real limitation worth stating plainly. A labelled diagram carries its meaning in the layout, and text extraction flattens that into a list of labels. For anatomy, circuits, or anything where the spatial relationship is the fact, write those cards yourself and let the app handle the surrounding prose.
Equations are the other weak spot: symbols survive inconsistently through OCR. For maths-heavy pages, cards are the wrong format anyway — work problems instead and save flashcards for definitions, conditions and when-to-use-what.
A workflow that actually fits a lecture
The version of this that students stick with looks like: photograph the slides during the lecture without trying to write anything down, then that evening pick the single most important capture and generate its deck. Ten minutes, one deck, reviewed the same day. Repeat tomorrow. Over a semester that accumulates into a syllabus-wide deck you never sat down to build — see the full method for lecture notes.