Citations, figures and tables

Citations#

Type @ and start a name — the autocomplete searches your whole library plus the project’s own bibliography. Picking one inserts a citation chip: rendered as (Author, Year) or as collapsed numerals like [3,5,9–14], your choice — toggle Au–Yr / [1] in the References pane (it’s stored per document). Numbers assign themselves in order of appearance and stay correct as you edit.

The bibliography maintains itself. Citing a paper from your library copies its entry into the project’s references/library.bib automatically, so the project stays self-contained. Even quicker: paste a DOI (or a doi.org link) straight into the text — Flux fetches the metadata, adds the reference, and leaves a citation chip where you pasted.

Double-click a citation chip to edit its group — add, remove, and reorder the references cited at that spot as one unit. Alt+R opens Reference Search in the margin (mark rows with Space, insert them as one citation with Enter); Alt+C edits the citation under the caret.

Figures#

Insert a figure with /figure (or Ctrl+K → Insert figure…). The picker shows every figure at full size. Scroll or use the arrow keys through long lists, and search by name or caption. When the project has more than one canvas, the dropdown next to the search box limits the list, and your search, to a single canvas.

In the text, the embed is a compact ⌗ name chip with the live figure rendered beneath it: the real figure from Figure, with its caption. Put the caret on the chip’s line to see its source, or double-click the chip to open the figure. Hover the figure for width controls (50% · 75% · 100% · Auto, or drag the grip); Open in Figure jumps to editing it.

Captions live on the figure itself and are edited in Figure-Meta (Alt+M, available in Paper and Figure), so a figure and its caption cannot drift apart. Exports carry them automatically.

Reference a figure with @fig-...: it becomes a live “Fig 3” chip whose number always matches the figure’s current designation in Figure, even after figures are added, removed or renumbered there. @tbl- and @eq- work the same way for tables and equations, with their own autocomplete. Type @@ for the figure picker, which can also select specific panels (Fig 3a,c) and has the same canvas dropdown as the insert picker.

Double-click any reference chip to jump to the thing it points to; a table or equation reference scrolls the editor to its source. Ctrl+Enter follows the reference under the caret.

Slides in documents#

Use /slide or Ctrl+K → Insert slide…, choose a deck, then choose a slide. The picker previews the completed slide. The block in your document starts at step 0. In the text, the embed line shows as a compact ▷ Deck 3 · Slide 4 chip with the deck’s title and the slide’s name, as under the slide itself. Hover the chip for the slide’s position in its deck. Put the caret on the line to see its source. Long values such as the IDs and the image path show as … until the caret reaches them. Double-click the chip to open that slide in Slides. A dimmed chip means its deck or slide is missing.

Click the artwork or › underneath it to advance one animation step. Tracks within a step play together with their authored timing. Automatic and “with previous” steps still wait for your next click here. Clicking during an animation finishes that step; the final step stops. ‹ goes back and ↺ resets to step 0.

While the artwork has focus, Right, Enter, or Space advances, Left goes back, Home resets, and Escape returns to writing. The Animation toggle lets you snap between steps; your system’s reduced-motion preference supplies its initial setting. Width buttons and the right-edge grip resize the display without changing the source slide. Use ↗ to open the original in Slides, or Replace… to choose another source.

Each occurrence plays independently. Typing and scrolling retain its position; reopening a document starts again at step 0. Saved source edits appear in the block. If its deck or slide disappears, the block shows an unavailable message and retains its replacement control. If a slide with a 3D model cannot play here, the block keeps its still image and says 3D preview unavailable — showing a still underneath; hover that line for the reason. Flux identifies the source by deck and slide IDs, so renaming or reordering it is safe.

The editable source is an ordinary Quarto image with a generated SVG fallback:

![Optional caption](../slides/talk/renders/results-step-0.svg){#slide-example .flux-slide deck="talk" slide="results" width=100%}

The caption belongs to this document. Speaker notes stay in the deck. Reading preview and exported HTML retain playback; PDF and Word show the step-0 image, regardless of the step currently visible in the editor. Word includes a raster fallback. Generated SVGs can be rebuilt by exporting or inserting the slide again.

Tables#

Insert a table with /table — it arrives labeled and numbered, with the first header cell selected so you can just start typing. Everything in the cells renders properly: bold, code, math like $x^2$, citations, even references to other tables.

While you are reading or writing elsewhere, a table’s pipe-text source is out of the way: it shows as a small blue ▦ Table 3 pill, with the rendered table below it — the same idea as the pill on a figure’s source line. Put the caret in the table and the source comes straight back, exactly as you typed it. There are three ways in:

• Click a rendered cell — the caret lands in that cell’s source.
• Click the rendered caption — the caret lands in the caption text.
• Click the pill, or arrow into it from the line above or below.

Move the caret out and it folds away again. Nothing about the .qmd on disk changes — this is only what the editor shows you.

Editing is spreadsheet-shaped:

• Tab / Shift+Tab move between cells (Tab past the last cell adds a row); Enter adds a row below — press it on an empty last row to leave the table.
• The pipes align themselves as you type — no manual space-pushing.
• Ctrl+Alt+R adds a row, Ctrl+Alt+C a column (+Shift deletes one); Ctrl+Alt+A cycles the column’s alignment, as does Alt+click on a rendered header cell.
• Hover the table for + Row · + Col · Format · Copy (Copy puts the table on the clipboard tab-separated, ready for Excel or Sheets) — these work whether the source is showing or folded away.
• Paste from Excel/Sheets and the copied cells become a table — pasted outside one they insert a fresh table; pasted inside one they fill cells from the caret. CSV converts via Ctrl+K → Paste as table.

A table is numbered once it has a caption line — : Your caption {#tbl-name} below it — which /table writes for you. A table without one is perfectly fine, it just has no number to carry, so its pill reads ▦ Table. Wide tables scroll in the editor and shrink to fit on export so nothing gets cut off.

Equations#

Display math does not fold away like a table: its source line stays visible, with the rendered equation below it. A labeled equation ({#eq-name} after the closing $$) is numbered, and @eq-name references it with a live chip, as described under Figures.