Skip to content

⚡ Shapez 2 Optimization Guide

Last updated: July 2026 | Game version: Early Access (Steam)


Shapez 2 is about scaling — building factories that produce hundreds of shapes per minute with minimal space and belt spaghetti. This guide covers belt throughput, balancer designs, compact blueprints, and architecture patterns for endgame automation.


Belt Throughput Basics

Belt Speed Tiers

Belt Tier Shapes/Second Shapes/Minute Unlocked By
Basic Belt 15/s 900/m Start
Fast Belt 30/s 1,800/m Milestone 8
Express Belt 45/s 2,700/m Milestone 16
Max Belt 60/s 3,600/m Milestone 24

Throughput Rules

Rule 1: Never merge beyond belt capacity.
  - Two belts at 30/s each → merging requires Express Belt (60/s max)
  - Merging two 45/s belts is impossible without upgrading belts

Rule 2: Balance before processing arrays.
  - Any array of 4+ machines needs a proper balancer on input
  - Without balancing, the first machine starves while the last idles

Rule 3: Upgrade belts in tiers.
  - Replacing all belts at once is expensive
  - Upgrade the main bus first, then branches, then individual lines
  - Express belts are only needed for the final 60/s push

Belt Balancers

Balancers ensure that every input in a multi-belt system gets equal throughput, regardless of which belts are backed up.

4-to-4 Balancer (Compact)

     Input A    Input B    Input C    Input D
        │           │          │           │
        ▼           ▼          ▼           ▼
    ┌────────┐ ┌────────┐ ┌────────┐ ┌────────┐
    │ Split  │ │ Split  │ │ Split  │ │ Split  │
    └───┬────┘ └───┬────┘ └───┬────┘ └───┬────┘
        │          │          │           │
        ├────┬─────┤          ├─────┬─────┤
        │    │     │          │     │     │
        ▼    ▼     ▼          ▼     ▼     ▼
    ┌────────┐ ┌────────┐ ┌────────┐ ┌────────┐
    │ Merge  │ │ Merge  │ │ Merge  │ │ Merge  │
    └───┬────┘ └───┬────┘ └───┬────┘ └───┬────┘
        │          │          │           │
        ▼          ▼          ▼           ▼
     Output A    Output B    Output C    Output D

Build pattern: Split all 4 inputs → cross-connect to 4 mergers → each output receives 1/4 of each input.

8-to-8 Balancer (Tileable)

Recommended design:
  1. Split all 8 inputs into 16 lines
  2. Cross-connect in pairs to 8 mergers
  3. Each output gets 1/8 of total input

Tiles: 8 wide × 6 long = 48 tiles

N-to-1 Balancing

For feeding a single high-demand line:

Setup: Multiple inputs → load balancer → single output belt

If 3 inputs at 30/s feed 1 belt (max 60/s):
  - You need only 2 balanced inputs (2 × 30 = 60)
  - A 3-to-1 balancer: split one input, merge with the other two, discard excess

Better: balance all 3 inputs into 2 outputs (3:2 balancer) then merge those 2.

When to Use Balancers

Situation Balancer Type Why
Input to 4 identical processors 4-to-4 Each machine gets equal throughput
Loading a 8-belt main bus 8-to-8 Balanced lane use prevents bottlenecks
Unloading production arrays Reverse balancer Ensures all machines output evenly
Multiple mining patches → base N-to-M Handles uneven ore patch output
Any processing array of 3+ Always balance Without balance: first machine starved

Stacker Optimization

Stackers combine shapes. The order and positioning of stackers determines your throughput.

Stacker Sandwich Technique

Standard approach (1 stacker):
  Shape A ──► Stacker ──► Shape A+B
  Shape B ──►          (1x throughput)

Sandwich approach (2 stackers, parallel):
  Shape A ──► Split ─┬─► Stacker1 ─┬─► Shape A+B
  Shape B ──► Split ─┤             │
                     ├─► Stacker2 ─┤
                     └─►          Merge ──► 2x throughput

Quad-Stacker (4x Throughput)

For high-volume shape requirements (e.g., crystals for milestones):

Input A (60/s) ──► 4-way Split ─┬─► Stacker 1 ─┬─► Merge 1
Input B (60/s) ──► 4-way Split ─┤              │
                                ├─► Stacker 2 ─┤
                                │              ├─► Merge 2 → Output (60/s)
                                ├─► Stacker 3 ─┤
                                │              │
                                └─► Stacker 4 ─┘
Throughput: 4 stackers × 15/s each = 60/s (capped by belt)

Stacker Priority

Stacker Setup Space (tiles) Throughput Best For
Single stacker 2×3 15/s Early game, low volume
Stacker sandwich (2) 4×4 30/s Mid-game
Quad-stacker (4) 8×6 60/s Endgame high demand
Octo-stacker (8) 16×8 120/s (dual belts) Mega-factory only

Bus Architecture

A main bus carries all your shapes through a central highway. Production modules tap off what they need.

8-Belt Main Bus Layout

          Production Modules
    ┌────────────┼────────────┐
    │  Belt 1:  │ Shape A (raw)  │
    │  Belt 2:  │ Shape B (raw)  │
    │  Belt 3:  │ Shape C (raw)  │
    │  Belt 4:  │ Shape AB (stacked) │
    │  Belt 5:  │ Shape AC (stacked) │
    │  Belt 6:  │ Shape BC (stacked) │
    │  Belt 7:  │ Shape ABC (complex)│
    │  Belt 8:  │ Painted shapes     │
    └────────────┴────────────────────┘
          Bus lanes
     (all running same direction)

Tapping the Bus

For each production module:

  1. Split one belt from the bus (using Splitter)
  2. Process the shapes (stack, paint, cut)
  3. Merge processed shapes back onto the bus

  Never tap a bus belt with 100% of its throughput.
  Always leave 20% headroom on main bus lanes.
  Tap from the RIGHT side of the bus (keep left clear).

Bus Lane Allocation

Game Stage Bus Width Content
Early (M1-8) 2-4 belts Raw shapes only
Mid (M9-16) 4-6 belts Raw + simple stacks
Late (M17-24) 6-8 belts Raw + stacks + painted
Endgame (M25+) 8-12 belts Full complexity range
💡 Bus tip: Use the "lane balancer" module every 20 tiles on the bus
to rebalance lanes. Uneven consumption will slowly drain one belt
while another overflows. A 4×2 balancer every 20 tiles fixes this.

Blueprint Archetypes

The Universal Processor (10×10 tiles)

┌──────────────────────────┐
│  INPUT ──► Splitter      │
│              │           │
│        ┌─────┴─────┐    │
│    Belt A        Belt B  │
│        │           │     │
│  ┌─────────┐ ┌─────────┐│
│  │Stacker  │ │Painter  ││
│  └────┬────┘ └────┬────┘│
│       │           │      │
│        └────┬─────┘      │
│             │            │
│         ┌───▼───┐       │
│         │ Merger│       │
│         └───┬───┘       │
│             │           │
│  OUTPUT ────┘           │
└──────────────────────────┘

Throughput: 15-30/s depending on belt upgrade
Best for: Any shape that needs stacking + painting

Compact Stacker Block (6×6 tiles)

┌────────────────┐
│  A ──► SPLIT ──┤
│           │    │
│  B ──► ──┤    │
│           ▼    │
│       STACKER  │
│           │    │
│        ┌──▼──┐ │
│        │MERGE│ │
│        └──┬──┘ │
│     ┌───┐ │    │
│     │OUT │◄────┘
│     └───┘      │
└────────────────┘

Input: 2 belts (30/s each) → output: 1 belt (60/s after stacking)

Painter Block (8×4 tiles)

┌──────────────────────────┐
│  Shape In ──► Splitter   │
│                  │        │
│           ┌──────┴──┐    │
│      Paint A     Paint B  │
│           │          │    │
│           └────┬─────┘    │
│                │          │
│           ┌────▼────┐    │
│           │  Merger  │    │
│           └────┬─────┘    │
│                │          │
│       Shape Out (colored) │
└──────────────────────────┘

Paint A and Paint B alternate colors on the output belt.
For 4 colors: chain two of these blocks.

Endgame Throughput Strategy

Milestone 24+ Optimization

At endgame, you need 60/s throughput on complex shapes. Here's how:

1. Parallel processing arrays
   - 8 processing lines in parallel for each complex shape
   - Each line: 4 stackers + 2 painters = 6 machines
   - Total: 48 machines per complex shape

2. Dedicated resource lines
   - Each raw shape gets its own 60/s belt
   - 4 mining outposts per raw shape type
   - Direct feeds (no bus) for highest-demand shapes

3. Output balancing
   - Every block of 4 machines gets a 4-to-4 balancer
   - Output merges to 2 Express belts (120/s total)
   - Feed directly into Hub input

Throughput Math for Endgame

Component Max Throughput Limiting Factor
Single belt 60/s Belt speed cap
Single stacker 15/s Machine speed cap
Stacker array (4) 60/s Belt input limit
Painter 30/s Faster than stackers
Crystal mixer 15/s Machine speed cap
Hub input 120/s (2 belts) Hub belt slot limit

Unlocking Full Potential

To reach 60/s on complex shapes:
  1. Upgrade all belts to Express (60/s)
  2. Build quad-stackers for each stacking step
  3. Use parallel painting lanes (2 painters per color)
  4. Balance EVERYTHING with 4-to-4 balancers
  5. Feed the Hub with 2 belts (left + right input)

Total footprint: ~100×150 tiles for one full-complexity shape line

Color Mixing Optimization

Compact Color Wheel (12×12 tiles)

For efficient crystal/color production:

          Crystal In (60/s)
         4-way Splitter
       ┌────┬──┴──┬────┐
       │    │     │    │
       ▼    ▼     ▼    ▼
     ┌──┐ ┌──┐ ┌──┐ ┌──┐
     │R │ │G │ │B │ │W │
     └┬─┘ └┬─┘ └┬─┘ └┬─┘
      │    │     │    │
      └────┼─────┼────┘
           ▼     ▼
        ┌────┐ ┌────┐
        │Mix1│ │Mix2│  ← Create secondary colors
        └──┬─┘ └──┬─┘
           │      │
           └──┬───┘
           ┌───────┐
           │ OUTPUT│ 60/s max
           └───────┘

Paint Efficiency

Paint Type Shapes Painted Crystal Cost Per Shape
Single color 1 shape 1 paint unit
Striped (2-color) 1 shape 2 paint units
Checkered (4-color) 1 shape 4 paint units
Gradient (8-color) 1 shape 8 paint units

Pro tip: Always produce the minimum color variety needed. If a shape only needs red and blue, don't produce the full rainbow — it wastes crystal throughput.


Blueprint Library Essentials

Save these 5 blueprints for rapid expansion:

Blueprint Size Purpose
4-to-4 Balancer 4×6 Universal balancing
Quad-Stacker 8×6 60/s stacking
Compact Painter 8×4 2-color painting
Universal Processor 10×10 Stack + Paint module
Dual Hub Feeder 4×8 120/s hub input

Common Bottlenecks & Fixes

Symptom Cause Fix
Machines idle randomly Unbalanced input Add a 4-to-4 balancer
Belt backed up but machines starved Wrong merge order Stack before merging, not after
Stackers at 15/s while belts are 60/s Single stacker bottleneck Build quad-stacker array
Hub not accepting full throughput Only 1 input belt used Split to both Hub inputs
Painters skipping shapes Paint supply empty Add paint buffer chest
Complex shape output below 30/s Crystal bottleneck Upgrade crystal mixer to quad

All throughput numbers based on Shapez 2 Early Access (Steam, v0.x). Belt speeds and machine rates may change with updates. Always verify in sandbox mode before committing to a build.