Image occlusion

Image occlusion turns an image into cards — perfect for anatomy, maps, diagrams, and UI screenshots. There are two ways to make them, and they work side by side: draw boxes interactively, or write a numbered list.

Interactive (recommended)

Run the command “Create image occlusion at cursor”, pick an image, and draw boxes right on it in the studio:

Drag on the image to draw a box; drag or resize existing boxes.
Give each box a Markdown / LaTeX answer, or leave it empty to simply hide a label that is baked into the image.
Delete a box with its red ×.

It saves as a self-contained decks-occlusion block in your note.

Each mask becomes a card. On the front the box covers the label; on the back it is revealed, along with whatever answer you typed for it:

Front

Anatomy of the heart — the box over the lower-left chamber is hidden.

Back

Left ventricle

One mask, one card. An answer can be markdown or LaTeX; leave it empty to simply uncover a label already on the image.

What gets stored

```decks-occlusion
image: "[[anatomy/heart.png]]"
version: 2
masks:
  - id: m1
    x: 12.5
    y: 30
    w: 18
    h: 9.5
    answer: "Left **ventricle**"
  - id: m2
    x: 55
    y: 22
    w: 14
    h: 8
    answer: ""
```

Worth knowing about that block:

image accepts a [[wikilink]], an ![[embed]], a ![alt](path), or a bare path.
Coordinates are percentages of the rendered image, not pixels — so a diagram occluded on a laptop is masked correctly on a phone.
An empty answer makes a deletion-only box: it hides the region and reveals it again with no text panel, which is what you want when the label is already printed on the image.
Each box keeps a stable id. Moving a box, resizing it, or rewriting its answer all preserve that card’s review history, and editing one box never disturbs its siblings.

While reviewing

The active box is hidden on the front with an accent outline; Show answer reveals it in place and renders its answer below. Whether the other boxes stay hidden follows the cloze context setting on the deck’s profile — “hidden” masks them all, “open” shows the rest.

In reading view you see the fully answered diagram, with an Edit button to reopen the studio. You can also re-edit from the flashcard manager, which opens the studio rather than a text editor.

Numbered list (simple)

You can also provide an image plus a numbered list; the numbers on the image map to the list items.

## Bones of the arm

![[arm_bones.png]]

1. ==Humerus==
2. ==Radius==
3. ==Ulna==

Each list item is one card. The image is the front of every card; the list is the back, with the active item blanked.
All highlights in one item are blanked and revealed together, so an item can hold more than one word.
An item without highlights still makes a card — the whole item becomes the answer.
Empty items are skipped.

For this to be recognised, the first non-empty line under the heading must be the image embed and every line after it must be a numbered list item. If either is untrue the block falls through to ordinary cloze or header + paragraph parsing — which is the usual reason a list “isn’t working”.

Requirements

Image occlusion is built on top of cloze deletions. Because of that, cloze must be enabled on the profile the deck uses (it is on by default), and the block must sit under a heading at the profile’s configured header level, like any other card.

Reverse cards are not generated for image-occlusion cards in either style.