The difference between a level that works and a level that is remembered is almost always one moving part. This page is about those moving parts.

Joints: the core advanced tool

A joint connects two objects so that they move in relation to each other rather than independently. The pin joint is the one you will use constantly — it holds two objects together at a single point, which is what turns a flat object into a hinge.

Once you have a hinge, the catalogue of things you can build opens up:

  • Doors and gates that swing instead of sliding.
  • Traps that drop when a support is removed.
  • Swings, pendulums and rope-like motion.
  • Machines with visible moving parts, which read as intentional rather than as scenery.

The reason this matters is comprehension. A level whose mechanics are visible is a fair level. A level where the player cannot tell why they died is a level people quit.

Contraptions: what the community builds

The wiki documents a category of popular user-made creations: hazards and mechanisms assembled from ordinary objects and a joint, because none of them exist as a single editor item.

The chomper. Also called a snapper, scissors or clapper. It is two interactive items — usually blade weapons — connected at their back ends with a pin joint. When it clamps onto the character, dismemberment follows quickly. It looks like a designed object and is entirely improvised.

That is worth stating plainly for anyone starting out: the impressive things in featured levels are usually two objects and a joint. There is no secret tool you have not unlocked.

The wiki is explicit that only physical objects which change the playing experience count as creations. Artwork does not count, and vehicles do not count, because those are editor features rather than contraptions.

Triggers: making the level react

The trigger tool fires events in response to the player. This is what separates a level made of static geometry from a level that behaves like a designed encounter.

Use triggers for:

  • Set pieces that begin when the player crosses a line, such as a vehicle crashing into scenery.
  • Call-and-response hazards where a safe route opens only after an earlier action.
  • Scripted reveals, including text and sound. The documentation includes an in-level text object used to label sections, and a documented set of sounds that can be fired from the trigger tool.

The design principle worth carrying over: telegraph, then fire. A trigger the player cannot anticipate reads as a cheat. A trigger the player notices a moment early reads as a great moment.

Rails and vehicle-specific sections

The rails object only makes sense with Explorer Guy, whose mine cart can attach to them. Rails replace the normal physics movement with a guided path, which is the single biggest behavioural change available in the editor.

If you build a rail section, you are committing the player to a specific character. Decide whether that is the intent, or whether the level should be winnable another way.

Working inside the shape and art budgets

Two documented limits shape every advanced build:

  • 900 shapes.
  • 10,000 art units.

Both can be raised in-game, but designing around them produces better levels anyway. Some practical habits:

  1. Use polygons for irregular and extreme shapes. The polygon tool spans the full range, from a 2 x 2 triangle to the entire 20,000 x 10,000 stage.
  2. Reuse geometry. One wall section, copied, costs the same as one wall section built by hand and takes a fraction of the time.
  3. Keep decoration off the shape budget where possible by using the art tool for anything purely visual.
  4. Give yourself headroom. A level that uses all 900 shapes cannot be fixed when testing reveals a problem.

Performance is a design constraint

Heavy levels run worse, and the game has had performance reports since launch. A level near the object limit on a machine near the minimum spec will stutter, and stutter breaks timing-based sections.

This is the same advice the official FAQ gave about detail: complexity that does not improve play is not worth its cost. If a decorative structure is competing with a timing window for frame budget, the timing window wins.

The final test

The old official guidance is the best summary of what “advanced” should mean. A level should be fun, understandable, and between roughly 30 seconds and 2 minutes long — and if dying makes the player want to try again rather than quit, you have built something worth publishing.

Length and difficulty are not substitutes for that. A two-minute level that reads clearly will outlast a ten-minute level that cannot be understood.