Engine capabilities

This page is the public maturity snapshot for the engine revision dated 2026-08-09. A method existing in source is not, by itself, proof that it is a production authoring contract.

Production authoring

These are the preferred tools for real projects:

Automatic root-tape layout and lazy reveal
Plain and rich-run text
LaTeX math
CSS-like width, height, alignment, wrapping, and margins
Flex rows and columns with independently addressable items
DataPath for sampled paths, trajectories, contours, routes, and vectors
DataPlot for named sampled series and markers
Diagram for explicit semantic node-edge structures
StateTransition for synchronized allowlisted property changes
ElementMorph for recompiled content or geometry
Isolate and overlay camera focus
Additional camera-facing tapes with automatic curtain switching
Explicit validated tape selection and local movement with scroll_tape()
Portrait 9:16 and landscape 16:9 settings
Strict project/DSL checking before render

These capabilities still require an actual preview. Structural validation does not prove legibility, pacing, camera quality, or subject accuracy.

Specialized capabilities

These are implemented but need closer visual testing:

Equation-backed mathematical surfaces
Tape-local cube/sphere and multipart solids
Solid lift, rotation paths, labels, and camera inspection
Full-tape PNG/PDF export
Complete desktop project archive import/export
Reel cutting
Legacy grid-board and quadratic helpers

The legacy grid/quadratic methods remain for compatibility. New projects should prefer generic visuals and project-local helpers.

Production spatial composition

These are exercised by the orbital flagship and engine tests:

Additional camera-facing tapes
Automatic world → tape, tape → tape, and tape → world curtain switching
Explicit scroll_tape() / serialized TapeScroll
Stable-ID registered world objects through add_object()
Stable-ID world morphs
add_camera_inspect() paths

Only one tape—or the free world—is visible at a time. Simultaneously visible tapes and arbitrary physical tape transforms are not current contracts.

Experimental capabilities

Use these only when a project genuinely needs them and can provide render evidence:

Low-level free world objects through add_world_object()
General camera keyframes and observe_object()
Relative world placement

The structures and some tests exist, but the world-camera model and authoring ergonomics are not yet a flagship-safe default.

Not currently available

Arbitrary physical transforms for tapes
Simultaneously visible tape contexts
Generic timed traversal or cursor animation over DataPath / DataPlot
Shared reactive clocks and declarative data bindings
Timed audio/media tracks in the canvas timeline
Automatic subject-specific simulation or graph layout
Production builders for every nominal Manim primitive accepted by historical DSL loading

Keep these as explicit limitations. Do not imitate them with lesson-specific branches in the engine and then advertise the workaround as a general feature.

Open spatial-authoring question

Orbital Mechanics, Feedback Control, SN2 Reaction, and DNA to Protein expose repeated manual 3D positioning and movement work. They are evidence used to test possible cross-subject abstractions, not templates that Matemium’s engine will be adapted to.

Persistent relative movement, sampled pose traversal, live anchor bindings, surface-relative movement, and geodesics are under discussion. No listed approach or illustrative API has been accepted or rejected, and none is a launch capability. See the open discussion and evaluation criteria.

Desktop delivery capabilities

The current desktop can:

Inspect saved source and list every populated tape in a scene
Export one complete tape as natural-aspect PNG or PDF at native or fixed detail
Archive an entire editable project workspace as .matemium.zip
Validate an imported archive, assign a new project identity, and open it locally
Browse imported images, video, audio, and generated output history

Evidence levels

Use precise language:

Level	What it establishes
Declared	A method, class, or document exists
Importable	The scene source imports and instantiates
Structurally valid	The compiled DSL passes checks
Renderable	A real render completes
Visually accepted	The output was watched and inspected
Domain reviewed	Facts, equations, data, and causal claims were checked
Production ready	Relevant tests, packaging, documentation, and final media are accepted

Never promote a capability based only on a lower evidence level.