A light reflection puzzle asks you to place mirrors on a grid so that a single ray of light reaches a specific exit edge. In Mirror Reflection Puzzle, that grid is 5×5, the boards are fixed, and the ray is traced cell by cell so you can see exactly where your prediction diverges. There are five levels, each one builds toward a longer route, and a perfect run finishes on exactly 1,000 points. Slash and backslash tiles turn the ray by ninety degrees according to the equal-angle law of reflection, as described by OpenStax University Physics Volume 3, while empty cells let it continue straight. The five fixed boards gradually move from a single turn to a four-mirror route, so every new signal asks you to hold a slightly longer path in mind. No random layouts, no timers, and no network calls are involved; the entire ray model runs locally in your browser and the same five layouts are replayable indefinitely.
What a Light Reflection Puzzle Asks You to Do
Every board in this genre has three visible ingredients: a fixed entry point, a marked exit edge, and an otherwise empty grid where you place the mirrors. The light enters the grid from the entry, travels one cell at a time in one of the four cardinal directions, and either leaves through the requested exit or fails. Empty cells do not change the ray's direction; only a slash or backslash tile changes it. Because the entry and exit positions are fixed, the puzzle is not about choosing where the ray starts or stops. It is about deciding which cells deserve a mirror and which orientation each mirror should take.
Mirror Reflection Puzzle packages that idea into five fixed levels. Level one asks for a single turn, and the boards gradually move toward longer routes, until the final board asks you to keep a four-mirror route in your head at once. The boards never shuffle and the ray model is narrow on purpose: only four travel directions exist, mirrors are ideal forty-five-degree reflectors, and there are no thickness, wavelength, or intensity rules. That narrowness is what lets the tracer give a precise cell-by-cell answer every time you press Trace ray, which is far more useful for learning than a flat right-or-wrong verdict would be.
How Slash and Backslash Mirrors Turn the Ray
The two mirror types behave as complementary reflectors. A forward slash turns a right-moving ray into an upward-moving ray, an upward ray into a rightward ray, a leftward ray into a downward ray, and a downward ray into a leftward ray. A backslash does the opposite pairing: right becomes down, down becomes right, left becomes up, and up becomes left. Both tiles always rotate the ray by exactly ninety degrees, never by forty-five or one-eighty. The direction rule is the same one used in standard optics (the angle of reflection equals the angle of incidence), as also explained by The Physics Classroom, but it is expressed on a discrete grid where only the four cardinal travel directions are allowed.
The exact mappings you need to memorize are summarized below.
| Incoming direction | Slash (/) turns to | Backslash (\) turns to |
|---|---|---|
| Right | Up | Down |
| Up | Right | Left |
| Left | Down | Up |
| Down | Left | Right |
If you walk through any cell mentally, you can verify the table by remembering that a slash "/" looks like a forward slope and a backslash "\" looks like the reverse. The two mirror types together cover all eight possible (incoming direction × mirror type) combinations, and the source code verifies each one with a vector oracle before the puzzle is ever shipped.
How to Play Mirror Reflection Puzzle in Five Levels
The whole interaction reduces to three repeatable moves: pick a cell, choose a mirror orientation, and trace the ray. You can do each step with a mouse, a touchscreen, or a keyboard, so the same workflow works on a desktop, a laptop, or a phone.
- Open Mirror Reflection Puzzle in your browser. The first level loads with a 5×5 grid, one marked entry cell, and a highlighted OUT edge that the ray must leave through.
- Select any grid cell by clicking or tapping it. A selected cell cycles between empty, slash, backslash, and empty each time you press it. With a keyboard, use the arrow keys to move the selected cell and type `/`, `\`, or `0` to place a slash, place a backslash, or clear the cell back to empty.
- Place only the mirrors you can justify at this point. There is no penalty for empty cells, and mirrors that fall outside the active path do not affect the ray. Keeping the board sparse makes it easier to audit the intended route later.
- Press Trace ray or hit Enter to run the tracer. The visited cells highlight one by one in the order the ray actually travelled, and each visited cell shows the outgoing direction.
- Read the result in the status panel. A trace that reaches the OUT edge clears the level and awards exactly 200 points. A trace that leaves through any other edge is labelled a wrong edge, and a trace that revisits the same cell with the same incoming direction is labelled a loop.
- Adjust the mirrors and trace again. If you reach the OUT edge, the next board loads automatically. If your layout was already wrong once before and is now wrong in a different way, the run ends.
- Press Restart from the interface at any time to return to level one with an empty placement map, zero mistakes, and zero points. The same five layouts will reload because nothing in the game is randomised.
A Backward-from-OUT Solving Method
Working forward works fine on the early levels, but the longer boards reward a backward pass. Begin at the OUT edge and ask which direction the ray must be travelling in the last cell before it exits. Decide which mirror orientation in that cell would produce that outgoing direction from the cell before it. Repeat one cell at a time until your reconstructed path meets the natural entry path. The result is a chain of slash or backslash choices that you can place and trace as a single hypothesis.
When you are mid-solve, trace early and often. The highlighted path shows the first cell where your prediction diverged from reality, which is far more informative than a flat "wrong" message would be. Restart is also a learning tool: because the layouts never change, you can study a route on a second pass without worrying that random seeds will move the cells around. Mirrors outside the active path are harmless, but a sparse board is much easier to audit, so remove any tile you cannot justify.
Scoring, Mistakes, and the Frozen Final State
Scoring is intentionally transparent. Every solved level awards exactly 200 points, and all five levels combined produce a perfect score of exactly 200 × 5 = 1,000. Mistakes are counted on the layout, not on the press: a wrong trace only counts if the current mirror placement is distinct from every earlier wrong layout you tried. Checking the same failed layout again does not consume another attempt, which removes the punishment for simply retrying your previous mistake. A second distinct failed layout ends the run. Completed and failed states are frozen, so extra clicks or key presses cannot alter the recorded result.
| Event | Outcome | Score change |
|---|---|---|
| Ray reaches the exact OUT edge | Level cleared, next board loads | +200 |
| Ray exits through any other edge | Wrong-edge failure, retry allowed | No change |
| Ray repeats a cell with the same incoming direction | Loop failure, retry allowed | No change |
| Second distinct wrong layout | Run ends, board frozen | Score held at last cleared level |
| All five levels cleared | Run complete | Exactly 1,000 |
Best scores are stored in your browser's local storage as an optional convenience. The live run stays usable even if that storage is unavailable, and no account, download, or upload is required. There is no timer, no hidden penalty, and no third-party image or script that could change the rules.
What the Ray Model Does Not Cover
It is worth being explicit about the limits of the model so you do not expect features that are not part of the puzzle. This is not a precision optics simulator. It does not model mirror thickness, wavelength, refraction, scattering, brightness, polarization, imperfect surfaces, or any real measurement. The ray is a discrete entity that travels in one of four cardinal directions and reflects at perfect ninety-degree angles. OpenStax and The Physics Classroom are cited only for the equal-angle reflection rule that anchors the direction mapping, and the five board layouts themselves are original game fixtures rather than copies of textbook diagrams.
The implementation is equally deterministic. A pure reducer stores the mirror map and traces from the authored entry one cell at a time, the eight direction mappings are tested with a vector oracle, and literal winning paths are replayed for every level. Separate probes confirm both wrong-edge termination and loop detection. In other words, every visible behaviour of the tracer, including when and why a trace stops, is checked by tests that you can rerun in the browser, which is why the puzzle feels less like a guess and more like an audit you can complete with confidence.
If you're weighing options, Creature Fusion Generator: 5 Briefs, 1,000 Points covers this in detail.