Light Up · Akari
Light Up, also called Akari, is a bulb-placement logic puzzle. A bulb lights its own white cell and every visible white cell horizontally and vertically until a black wall blocks the beam.
Three rules, one grid
Every white cell must be illuminated. No two bulbs may see each other along an uninterrupted row or column. A numbered black wall must touch exactly the indicated number of bulbs in its orthogonal neighbors. Unnumbered black walls only stop light. These rules create local and long-range deductions at the same time.
Every white cell must be illuminated. No two bulbs may see each other along an uninterrupted row or column. A numbered black wall must touch exactly the indicated number of bulbs in its orthogonal neighbors. Unnumbered black walls only stop light. These rules create local and long-range deductions at the same time. The aim is to make each deduction visible and reproducible rather than depend on guessing. After every bulb or exclusion, scan the affected row, column and nearby wall clues again.
Use X marks to record exclusions
An X is a reasoning mark that says a white cell cannot contain a bulb. A 0-wall rules out every adjacent bulb position, and a placed bulb rules out every other visible white cell in its corridor. Explicit X marks make the remaining illumination candidates easier to read on both desktop and mobile.
An X is a reasoning mark that says a white cell cannot contain a bulb. A 0-wall rules out every adjacent bulb position, and a placed bulb rules out every other visible white cell in its corridor. Explicit X marks make the remaining illumination candidates easier to read on both desktop and mobile. The aim is to make each deduction visible and reproducible rather than depend on guessing. After every bulb or exclusion, scan the affected row, column and nearby wall clues again.
Difficulty changes the information density
The frozen bank uses 5×5, 6×6 and 7×7 boards. Easy sources retain more numbered walls, Medium removes more numbers, and Hard relies more heavily on corridor coverage and interactions between distant constraints. Board size and difficulty are separate controls.
The frozen bank uses 5×5, 6×6 and 7×7 boards. Easy sources retain more numbered walls, Medium removes more numbers, and Hard relies more heavily on corridor coverage and interactions between distant constraints. Board size and difficulty are separate controls. The aim is to make each deduction visible and reproducible rather than depend on guessing. After every bulb or exclusion, scan the affected row, column and nearby wall clues again.
Independent unique-solution verification
The committed bank is regenerated deterministically and an independent solver counts solutions from the published constraints. It checks row and column visibility, illumination coverage and every numbered-wall equality. Each source must have exactly one solution, and all eight rotation/reflection variants are checked again.
The committed bank is regenerated deterministically and an independent solver counts solutions from the published constraints. It checks row and column visibility, illumination coverage and every numbered-wall equality. Each source must have exactly one solution, and all eight rotation/reflection variants are checked again. The aim is to make each deduction visible and reproducible rather than depend on guessing. After every bulb or exclusion, scan the affected row, column and nearby wall clues again.
Deterministic Practice and Daily selection
Practice uses a stable URL seed together with board size and difficulty. Reopening the same seeded URL selects the same source and symmetry. Daily derives the selection from the date. These contracts make puzzles reproducible without a server, account or paid API.
Practice uses a stable URL seed together with board size and difficulty. Reopening the same seeded URL selects the same source and symmetry. Daily derives the selection from the date. These contracts make puzzles reproducible without a server, account or paid API. The aim is to make each deduction visible and reproducible rather than depend on guessing. After every bulb or exclusion, scan the affected row, column and nearby wall clues again.
Clean and Assisted records stay separate
Hint, Check, Undo, Redo and Restart are useful learning tools, but they change the meaning of a personal best. The run becomes Assisted when those helpers are used, while completions without them remain Clean. Local records keep those categories separate.
Hint, Check, Undo, Redo and Restart are useful learning tools, but they change the meaning of a personal best. The run becomes Assisted when those helpers are used, while completions without them remain Clean. Local records keep those categories separate. The aim is to make each deduction visible and reproducible rather than depend on guessing. After every bulb or exclusion, scan the affected row, column and nearby wall clues again.
Autosave restores reasoning, not only completion
The browser stores bulb marks, X marks, move count, assistance state, source, symmetry and seed locally. Reopening the same puzzle can restore an unfinished position, including a temporary contradiction, instead of silently replacing it with a fresh board.
The browser stores bulb marks, X marks, move count, assistance state, source, symmetry and seed locally. Reopening the same puzzle can restore an unfinished position, including a temporary contradiction, instead of silently replacing it with a fresh board. The aim is to make each deduction visible and reproducible rather than depend on guessing. After every bulb or exclusion, scan the affected row, column and nearby wall clues again.
A practical solving sequence
Start with tight numbered walls. A 0 eliminates every adjacent bulb position. If a wall still needs exactly as many bulbs as it has open neighbors, all those neighbors are forced. Then scan unlit cells and ask which remaining candidate positions can illuminate them. Re-check affected corridors after every forced bulb.
Start with tight numbered walls. A 0 eliminates every adjacent bulb position. If a wall still needs exactly as many bulbs as it has open neighbors, all those neighbors are forced. Then scan unlit cells and ask which remaining candidate positions can illuminate them. Re-check affected corridors after every forced bulb. The aim is to make each deduction visible and reproducible rather than depend on guessing. After every bulb or exclusion, scan the affected row, column and nearby wall clues again.
Designed for touch and keyboard
The interface uses explicit Bulb, X and Clear tools rather than a confusing multi-state tap cycle. Right-click can place an X. Keyboard users can focus a white cell and use B for bulb, X for exclusion, Delete for clear and H for hint. Live lit-cell shading shows the effect immediately.
The interface uses explicit Bulb, X and Clear tools rather than a confusing multi-state tap cycle. Right-click can place an X. Keyboard users can focus a white cell and use B for bulb, X for exclusion, Delete for clear and H for hint. Live lit-cell shading shows the effect immediately. The aim is to make each deduction visible and reproducible rather than depend on guessing. After every bulb or exclusion, scan the affected row, column and nearby wall clues again.
Focused search surface, not a doorway farm
This page points to the real Akari game. It does not create separate thin pages for every size, clue, seed, technique or difficulty. The only companion acquisition page is the genuinely different Daily intent, where date stability and the specialist streak are the reason to visit.
This page points to the real Akari game. It does not create separate thin pages for every size, clue, seed, technique or difficulty. The only companion acquisition page is the genuinely different Daily intent, where date stability and the specialist streak are the reason to visit. The aim is to make each deduction visible and reproducible rather than depend on guessing. After every bulb or exclusion, scan the affected row, column and nearby wall clues again.