The Visual editor

Visual mode turns supported LaTeX into an editable reading surface. It retains source ranges and writes changes back to the same .tex file used by the LaTeX editor. It is useful for drafting prose while keeping equations, figures and headings in context.

1. Open a .tex source and select Visual.

2. Click a supported paragraph, heading, list item or caption.

3. Edit the text. Inline emphasis is retained where supported.

4. Press Enter in ordinary prose to begin a new paragraph.

5. Use source undo/redo when needed, and compile to check the actual output.

The unchanged parts of the source are preserved. Visual editing is not an HTML-to-LaTeX conversion of the entire document. It shares autosave, conflict checks and undo history with LaTeX mode.

Click an equation to open its small LaTeX editor. The local math preview provides immediate feedback while you type, using the macros your preamble defines (\newcommand, \DeclareMathOperator, and files such as math_commands.tex). Apply writes the equation back with its original delimiters. If a math command is unsupported by the preview, inspect the source and compiled PDF; preview support and TeX support are different.

Numbers read as the PDF prints them. Sections (lettered after \appendix), figures, tables, algorithms, theorems, footnotes and equation rows are counted in order, and wherever the last build resolved a \label, Visual uses the number it printed. \ref, \eqref, \cref and \autoref show those numbers ("Fig. 2", "Section 3"), and ?? before a first build.

Click a citation or reference to edit its command. Citations show the build's own labels: [3] in numeric styles, "Smith et al. (2020)" for natbib's \citet and "(Smith et al., 2020)" for \citep. Before a first build they fall back to your bibliography's authors and years.

Use Split → LaTeX + Visual while learning. Editing a short passage and comparing its source is a practical way to understand commands without losing the readable view.

After a build, Visual measures the PDF itself: the page size, margins and columns, the body font, size and leading, paragraph indent and spacing, how each heading level is set (size, weight, small capitals, spacing, run-in), the title block, the abstract, captions, the running header, page numbers and review line numbers. The toolbar then reads Layout measured from the last PDF. Before a first build, Visual reads the same things from the document class options, geometry, titlesec and the template's style files.

Pages are set the way TeX sets them:

• Paragraphs continue across pages and columns, keeping at least two lines on each side of a break. Click the continued part to go on typing there.

• Figures and tables go where LaTeX places them: here for [h] when they fit, otherwise at the top of this or the next page. figure* spans the columns, wrapfigure floats beside the text, and subfigures share a line.

• Space merges and drops as TeX's \addvspace does, and footnotes collect at the foot of their page or column.

• For paragraphs unchanged since the last build, Visual adjusts word spacing slightly so each takes as many lines as it does in the PDF. Page breaks then follow the PDF closely.

Documents set in Computer Modern are drawn in Latin Modern from your TeX installation, with real small capitals. Times documents use the system Times.

Tables render from tabular (booktabs rules, \multicolumn, \multirow, \resizebox), with their cells editable. Pseudocode from algorithmic shows with its keywords, nesting and line numbers, verbatim blocks as typed, and theorem-like environments with their names and numbers. The reference list is taken from the build's .bbl.

Included files (\input, \include) appear where they are, as the PDF shows them, and their text can be edited in place: the edit goes into that file. Commands the document defines for itself with \newcommand or \def appear as they expand (\method shows the method's name); click one to edit its LaTeX.

Fit page changes magnification instead of narrowing and reflowing the document into a different reading width.

Comments and layout commands are hidden from the reading surface by default. Enable LaTeX controls to reveal them. Unsupported environments, and custom commands that do more than substitute their arguments (conditionals, catcode changes), retain source controls or placeholders. Use LaTeX for preamble changes, environment restructuring or template-specific formatting.

For example, writing a paragraph is a good Visual edit; introducing a custom theorem environment or changing document-class options belongs in source. A placeholder does not mean the content has been deleted.

Refresh PDF after typing compiles after a short idle period, approximately 1.5 seconds. If an edit arrives while a build is running, another refresh can be queued. Disable the toggle when frequent builds interrupt your workflow, and use Recompile explicitly.

Exact layout: Visual follows the PDF closely but does not run TeX. Line breaks inside a paragraph can still differ, and pages drift after long edits until the next build. Use the PDF for submission limits and final proofreading.

Selections can become comments, agent context, source checks or linked results through the available selection actions. The source anchor matters: if another edit changes or duplicates the passage before an action runs, VersoFold may require a fresh selection.

Dragging a snippet onto a Visual gap maps the insertion point back to source. Figure templates then open the image picker. Review inserted captions and labels before compilation.

Related: source editor, snippets, check the sources, linked results.