Last updated: July 18, 2026
shapez 2 Wires & Signals Guide
Wires are the difference between a factory that produces one shape and a factory that produces any shape. In shapez 2, the wire layer is a full signal system: shapes, booleans, colors, and error states all travel on the same copper, and buildings like Belt Filters change their behavior based on exactly which signal they receive. Most players hit the same wall — a filter that blocks everything, a gate that never fires, a MAM that stalls for no visible reason — and the fix is almost always understanding what signal is actually on the wire. This guide covers every signal type, the complete Belt Filter truth table, the logic gates that matter, and the debugging habits that make wiring problems trivial.
Signal Types — What Can Travel on a Wire
Every wire in shapez 2 carries exactly one of six signal states. Two of them — null and conflict — are error states that halt whatever they touch. Learning to read these states is 80% of wire debugging.
| Signal | Source | Carries | Common Use |
|---|---|---|---|
| Shape signal | Shape Analyzer, Global Receiver, Signal Producer | A full shape code (layers, quadrants, colors) | Tell Belt Filters which shape to pass; feed simulated buildings in a MAM |
| 1 (On) | Constant Signal building, logic output | Boolean true | Force a filter or gate open; enable a production line |
| 0 (Off) | Constant Signal building, logic output | Boolean false | Force a filter closed; divert all shapes to the side output |
| null | Disconnected or unpowered wire | Nothing — absence of signal | Not intentional. A null at a Belt Filter blocks the belt entirely, which is the #1 wiring bug |
| Conflict | Two different signals wired into the same line | Unresolvable state | Never intentional. Conflicts block filters and gates — isolate sources with diodes or separate wires |
| Color signal | Fluid Analyzer, Signal Producer | A color code (red, green, blue, mixed) | Route paint to the correct Painter; verify mixer output in a MAM |
Belt Filter Truth Table
The Belt Filter is the most important wire-controlled building in shapez 2. It reads the signal on its control wire and routes physical shapes accordingly — matching shapes continue forward, everything else exits the side output. But the filter responds to all five signal states, not just shape signals, and the two error states block the belt completely. This is the table to memorize.
The practical takeaway: a Belt Filter is not just a filter, it is a wire-controlled diverter. Feed it a shape signal and it filters; feed it a boolean and it becomes a remotely controlled gate; feed it nothing and it becomes a wall. Every stalled-belt bug in a wired factory is one of the bottom two rows of this table.
| Control Signal | Main Output | Side Output | Notes |
|---|---|---|---|
| Matching shape signal | Matching shapes pass | Non-matching shapes divert | The standard filter behavior. One filter per shape type on a shared belt. |
| 1 (On) | Everything passes | Nothing | Filter fully open. Use a Constant Signal to bypass filtering during testing. |
| 0 (Off) | Nothing | Everything diverts | Filter fully closed. Useful as a wire-controlled diverter or overflow switch. |
| null | Blocked — belt stops | Blocked | The filter has no signal and halts the belt. Check the wire connection and the signal source power. |
| Conflict | Blocked — belt stops | Blocked | Two signals are fighting on one wire. Separate them or add a diode. Place a Display upstream to see it. |
Logic Gates — The Ones That Actually Matter
shapez 2 gives you a full logic toolkit, but a working MAM only needs a handful of gates. The critical rule: logic gates operate on booleans, not shape signals. You cannot AND two shape signals — convert them to booleans with Analyzers or Comparators first, then combine.
| Gate | Behavior | MAM Use | Common Pitfall |
|---|---|---|---|
| AND | Outputs 1 only when all inputs are 1 | Enable a production line only when shape demand AND material buffer are both ready | A shape signal is not a 1 — compare shapes with an Analyzer first, then feed the boolean into AND |
| OR | Outputs 1 when any input is 1 | Merge multiple demand sources so any request starts production | Two different shape signals into OR create a conflict, not an OR — convert to booleans first |
| NOT | Inverts a boolean input | Stop a line when a storage buffer is full (NOT full = run) | NOT on a null input stays null — it does not become 1 |
| XOR | Outputs 1 when exactly one input is 1 | Toggle between two exclusive modes, e.g. paint quadrant A or B | Rarely needed — most toggle logic is cleaner with a NOT + AND pair |
| Comparator | Outputs 1 when two shape signals are identical | Verify a produced shape matches the request before delivering to the Hub | Rotation matters — normalize quadrants with a Simulated Rotator before comparing |
| Diode | Passes signal one direction only | Prevent backfeed between signal sources sharing a trunk wire | Forgetting diodes is the main cause of conflict states in compact MAM wiring |
Debugging Wires — The Display Method
Never debug wires by staring at the layout. Place a Display on the wire segment in question and read what it shows. The Display renders the live signal — a shape icon, a 1, a 0, or the distinct null/conflict indicators — so you are always working from ground truth. Then apply a simple rule: start at the consumer (the filter or gate that misbehaves) and walk Displays upstream toward the source, one segment at a time. The first segment where the signal changes from expected to wrong is the break.
The three failure modes, in order of frequency: a null from a broken or unpowered source wire, a conflict from two sources sharing a line, and a correct signal feeding a filter pointing the wrong direction. The first two are visible on a Display instantly. The third is visible by clicking the filter and checking its output arrows — a surprisingly common bug after copy-pasting filter rows.
For large MAMs, leave permanent Displays installed at the Global Receiver output, every filter feed, and the final delivery comparator. When throughput stalls, you diagnose in seconds instead of tracing a hundred tiles of wire. Treat Displays like fuses in an electrical panel: cheap, permanent, and exactly where the failures happen.
Wiring a MAM: The Signal Pipeline
A Make Anything Machine is a signal pipeline with physical modules hanging off it. The pipeline has five stages, and every stage is a wire problem before it is a belt problem:
- Acquire. A Global Receiver reads the current objective and emits the target shape signal. Verify it with a Display before building anything downstream.
- Decompose. Simulated Unstackers split the signal into layers; each layer feeds Simulated Cutters to isolate quadrants. A four-layer shape becomes up to sixteen quadrant signals.
- Normalize. Every quadrant signal passes a Simulated Rotator set to the NE reference. Skipping this stage is the most common cause of MAMs rejecting shapes they can actually build.
- Route. Comparators match each normalized quadrant against your operation modules, and the resulting booleans drive Belt Filters that admit the right physical pieces into the right stackers, painters, and rotators.
- Verify. A final Comparator checks the assembled output against the original request. Only a match opens the delivery filter to the Hub; anything else routes to trash. This one gate is the difference between a MAM and a machine that jams the objective with wrong shapes.
Build and test the pipeline one stage at a time with Displays at every junction. A MAM wired incrementally works on the first objective switch; a MAM wired all at once fails in one of five stages and gives you no hint which.
Frequently Asked Questions
Q: Why is my Belt Filter blocking everything even though the wire is connected?
Place a Display on the wire segment feeding the filter. Nine times out of ten you will see either null (the signal source is unpowered, destroyed, or the wire is broken further upstream) or conflict (two sources on one wire). A connected wire does not guarantee a valid signal — the Display shows what the filter actually receives, not what you think you built. Fix nulls by tracing the wire back to its source; fix conflicts by adding a diode or splitting the sources onto separate wires.
Q: How do Global Receivers work in a MAM?
A Global Receiver reads the current objective shape from the Hub network and outputs it as a shape signal on your wire layer — no physical belt connection needed. Wire that signal into a chain of Simulated Unstackers to split it into layers, then Simulated Rotators to normalize each quadrant, then Comparators to route each quadrant to the correct operation module. This is what lets a MAM switch targets automatically when the objective changes, instead of you rebuilding Signal Producers by hand.
Q: What are simulated buildings and why do I need them?
Simulated buildings process shape signals on the wire layer without touching physical shapes. A Simulated Unstacker takes a shape signal and outputs one layer; a Simulated Rotator outputs the signal rotated to a reference orientation; a Simulated Cutter halves it. You need them because a MAM must reason about a requested shape — decompose it into quadrants and layers, compare each piece against what your modules can produce — and all of that reasoning happens as signals, not as physical items.
Q: Should I normalize to NE before comparing shapes?
Yes. Community convention is to normalize every shape signal to the northeast (NE) reference orientation with a Simulated Rotator before any comparison. A shape and its 90-degree rotation are the same item for delivery purposes but different signals — comparing without normalizing rejects valid shapes. One Simulated Rotator at the start of each comparison chain eliminates an entire class of false mismatches.
Q: How many Displays is too many?
There is no such thing during debugging. Displays cost almost nothing and show you the live signal on any wire segment. Experienced MAM builders leave permanent Displays on the four or five most failure-prone junctions — the Global Receiver output, each filter feed, and the final comparator — because when production stalls at 2,000 shapes per minute, you want to see which junction went null in five seconds, not after ten minutes of wire tracing.