The Color Memory Game started as one mode — see a colour, recreate it from memory with three sliders — and grew into a family of variants that twist that single idea in different directions. Some shorten the flash. Some hide the live preview. Some give you a deliberately broken tool. Some drop the memory component entirely and test raw discrimination. Some put you in a room with friends who can sabotage your sliders. They all score on the same 0–10 round scale, but each one tests something slightly different about how your eye and short-term memory cooperate.
This guide walks through every variant currently in the game. For each one you’ll find the rules, the scoring, and a paragraph or two on how to actually play well — what trips most people up and what closes the gap between an average run and a great one.
The shared core
Before the variants: most modes express a guess with the same HSV slider system. H is hue (the colour family — red, yellow, green, etc.), S is saturation (how vivid the colour is, from grey at 0 to pure colour at 100), and V is value or brightness (from black at 0 to full brightness at 100). The pick-and-tap modes (Match, Imposter, Stroop and friends) and the typed ones (Hex Guess, the Color Blind Test) skip the sliders but keep the same 0–10 round scale.
Scoring uses CIEDE2000, the perceptual colour difference formula recommended by the International Commission on Illumination. CIEDE2000 measures how different two colours look to a human observer, not how different the underlying numbers are. A guess that’s “close in RGB” can still feel completely different to your eyes; CIEDE2000 reflects that, which is why our scores feel honest. The result of that comparison is mapped to a 0–10 score for the round. Five rounds, 50 points possible — except in the variants that change the maximum.
Solo — Easy
Solo Easy is the unmodified original. A countdown reads Ready, Set, Go, the target colour fills the screen for four seconds, then disappears. Three sliders appear, you dial them to recreate what you saw, and submit. Five rounds, 50 points. There’s a live preview of your current guess so you can compare what you remember against what you’re building in real time.
How to score well in Solo Easy
Almost everyone’s first instinct is to fiddle with all three sliders at once. Don’t. Lock in the hue first— most of your points live in the hue channel, and a wrong hue with perfect saturation and brightness still scores badly. Once the hue feels right, dial saturation, then brightness. The four-second flash is enough time to form one solid impression — try to walk away from each flash with a single sentence in your head: “deep teal, fairly bright”, “washed-out peach”, “pure mid-grey with a hint of warm”. Words survive memory better than pure visuals.
Solo — Hard
Solo Hard takes the four-second flash down to 2.5 and adds two adversarial elements. During the Ready–Set–Go countdown, three decoy colours flash on each word — each generated to be clearly distinct from the target (the game rejects anything within a CIEDE2000 distance of 15), so the countdown floods your eyes with unrelated colours right before the one that matters. Then, while the target itself is on screen, a colour name with no relation to it (Mustard, Turquoise, Crimson) is printed over the swatch — Stroop-style interference at the exact moment you’re trying to memorise the colour.
How to score well in Solo Hard
Treat the decoys as noise — they’re gone before the target arrives, and reacting to them wastes attention. The bigger trap is the name printed over the target. Your memory wants to store a verbal label alongside the visual impression, and the game supplies a wrong one while you’re looking. If Sage Green is sitting on a dusty pink, write your own label instead — “dusty pink, mid-bright” — and repeat it. The 2.5-second flash is short but enough: decide in the first half-second and spend the rest confirming. Once the sliders appear the name is gone, and only whichever label won survives.
Speed
Speed compresses the flash to one second and gives you a ten-second clock to dial your guess. You’re scored out of 12 per round instead of 10: the colour score is still 0–10, plus a time bonus of up to +2 for submitting early. An instant submit nets the full +2; using the entire ten seconds nets zero. Five rounds, max 60.
How to score well in Speed
Don’t rush the dial — rush the decision. Most players who score badly in Speed waste two or three seconds hesitating before they touch the sliders, then panic-tweak in the last second. A confident wrong guess submitted in three seconds usually beats a hesitant accurate one submitted in nine. Form your impression during the half-second the colour is most visible, commit the second the flash ends, and dial confidently.
Memory Stack
Memory Stack shows you a sequence instead of a single colour. Round 1 shows one colour. Round 2 shows two colours in a row. Round 3 shows three. By round 5 you’re watching five colours flash by and have to recreate them in order. Each colour you place is scored against its position in the original sequence.
How to score well in Memory Stack
This is the variant where verbal labels really shine. By round 3 or 4, you cannot hold every colour in pure visual memory; you have to encode them as words. “Mustard, dusty pink, navy, seafoam” survives a five-second gap; four pure visual impressions don’t. The flash on each colour is short, so practice naming colours quickly. A good colour vocabulary is the single biggest skill upgrade for Memory Stack.
Blind Sliders
Blind Sliders removes the live preview. You see the target for four seconds. After that, the sliders appear — but the swatch showing what you’ve currently dialled stays hidden until you submit. You have to dial entirely from your memory of the colour and your understanding of how the sliders move.
How to score well in Blind Sliders
Blind Sliders rewards calibration. Spend a few rounds in Solo Easy first to learn what hue 200 looks like vs hue 240, what saturation 70 looks like vs saturation 90 — then carry that mental map into Blind. Build the guess from memory of the slider positions, not from a guess at the colour. The score is unforgiving but the skill transfers: people who play Blind regularly become noticeably better in every other variant.
Gradient
Gradient shows a two-colour gradient as the target. You match both endpoints with two sets of HSV sliders. The score is the average of the two colour scores, so a brilliant guess at one end and a poor one at the other averages out.
How to score well in Gradient
Treat the gradient as two independent colour memory tasks, but encode them together as a relationship: “goes from warm pink to cool blue, with brightness increasing”. Remembering the trajectory makes both endpoints stickier than remembering them in isolation. Don’t spend all your time on one end — split your dialling effort evenly.
Flags
Flags swaps the abstract swatch for a real national flag. It waves on screen for five seconds, vanishes, and you dial back all three of its band colours — the round score is the average of the three. Five flags per game, drawn from a 40-flag pool and ordered easiest to hardest: flags with a white or black band come early (those bands are near-free points), and the all-colour flags like Colombia’s land last. If the game can place your country and its flag is in the pool, you finish on your own flag. Every reveal comes with a short fact about the design.
How to score well in Flags
Bank the free bands fast — white is zero saturation at full brightness, black is zero brightness — and spend the saved time on the chromatic bands, which are more vivid than most Solo targets but rarely the pure primary you assume. Knowing the flag beats remembering it: if you already know which of the three bands is the odd one, five seconds is plenty for the other two.
Chameleon
Chameleon turns the recall task into painting. A procedurally generated painting fills the screen for four seconds. Then a pure-white chameleon climbs onto it, covering a patch, and you get a brush, the painting’s own palette, and sixty seconds to repaint the body so the picture looks whole again. The covered patch can span multiple colour regions — if the boundary between sky and hills ran through the chameleon, you have to remember where. Scoring compares what you painted over each hidden region against the true colour underneath, on the usual CIEDE2000 curve; average 9.3 or better and the chameleon visibly vanishes into the scene. Five paintings per game, drawn from a ten-scene gallery and ordered easy to hard.
How to score well in Chameleon
Spend the four seconds on geography, not colour — the palette is handed to you on the brush bar, so the real questions are which colour goes where and where the boundaries cross the body. Trace the edges that will pass through the chameleon and note their angles. While painting, extend the visible lines of the surrounding painting through the body instead of guessing; the canvas around the outline is a ruler. Sloppy boundaries blend two regions’ paint and drag both scores down.
Broken Slider
Broken Slider locks one of the three sliders to a deliberately wrong value at the start of each round. You can’t move it; you have to compensate with the other two. If hue is locked but saturation and brightness still move, you’re forced to pick the closest possible match given an incorrect hue. The locked slider is shown clearly — you know which one is broken before you start dialling.
How to score well in Broken Slider
Accept the loss the broken slider forces on you and play the compensation game. If the broken slider is hue, you can’t match the colour family but you can match the apparent vividness and brightness, which still claws back a respectable score. If it’s saturation, push the hue and value into the closest feasible space and accept that you’ll be greyer or more saturated than the target. Most players underestimate how much can be recovered with the two working sliders.
Imposter
Imposter drops the memory component entirely. You’re shown a grid of swatches — all but one are identical. Spot the imposter and tap it. The grid size grows each round, and the difference between the imposter and the rest gets subtler as you progress. Scored on speed and accuracy.
How to score well in Imposter
Don’t scan linearly — let your eye pick out the swatch that feels different. Human peripheral vision is excellent at identifying odd-one-out colour differences when you let it work without conscious focus. Defocus your gaze across the whole grid for a beat; the imposter will usually pop. If it doesn’t, the difference is in saturation rather than hue; tilt your head or look slightly off-centre to break the symmetry of your foveal view.
Match
Match is pure discrimination: the target swatch stays on screen the whole time, above a grid of candidates. Exactly one candidate is the target; the rest are decoys shifted in one or more HSV channels. The grid grows from four swatches to nine across the five rounds while the shifts shrink, until round five sits near the just-noticeable-difference threshold. Ten seconds per round: a correct pick scores 10 plus a time bonus of up to +2 that drains as the clock runs, so an instant pick banks the full bonus. Sixty points possible.
How to score well in Match
Compare channel by channel, not holistically. Most decoys are single-channel shifts — a touch warmer, a touch greyer, a touch darker (the biggest grids add two combined shifts) — so hold the target and ask one question at a time: same hue? same vividness? same brightness? Every decoy fails one of the three; the match fails none. Move quickly in the early rounds, where the gaps are wide, and bank the bonus for the hard finish.
Spot the Difference
Spot the Difference puts two copies of the same mosaic side by side — identical except for one tile, whose colour is shifted slightly in one grid. Tap that tile. The mosaics grow from 6 to 20 tiles across five rounds while the shift shrinks toward the just-noticeable threshold, and every tile has its own colour, so the difference hides inside genuine variety. Ten seconds per round; a correct tap scores 10 plus a time bonus of up to +2 that shrinks as the clock runs. Sixty points possible.
How to score well in Spot the Difference
Scan in matched pairs with a rhythm — left tile, right tile, next — rather than staring at one grid and trusting memory to check the other. On the big grids sweep row by row; random darting revisits some tiles and skips others. If nothing surfaces by the halfway mark, the shift is probably saturation or brightness rather than hue: look for the tile that reads flatter or duller than its twin.
Color Blind Test
The Color Blind Test is five Ishihara-style plates: a disc of coloured dots in which some dots form a digit. Type the digit you see. Each correct read is worth 10 — there’s no time bonus, because reading the plate at all is the whole challenge — and the hue gap between figure and background shrinks every plate, so the fifth is a genuine squint. Fifty points possible. It’s a screening-style game, not a medical diagnosis; if the early plates give you trouble, an optician runs the real thing.
How to score well in the Color Blind Test
Lean back. The digit is a cluster of dots sharing a hue family, and it pops at distance or with a soft focus, where the dot-by-dot noise blurs away and the shape survives. Tracing individual dots is the trap. And if a plate stays stubbornly blank, guess the most plausible digit rather than passing — a wrong read scores the same zero as no read, so a guess is free.
Hue Sort
Hue Sort is the Farnsworth-Munsell idea in game form. A row of chips forms a smooth gradient between two locked end tiles; the middle chips arrive shuffled, and you swap them until the row reads as one clean sweep. Rounds grow from four chips to eight. There’s no clock — the test is purely of your eye — and scoring compares the colour at each position against the colour that belongs there, on the usual CIEDE2000 curve, so a chip one slot out of place still scores well. Fifty points possible.
How to score well in Hue Sort
Work outward from the locked ends: find the chip closest to the left anchor, then the next, rather than ordering the whole middle at once. Pairwise questions — which of these two sits nearer the blue end? — are more reliable than holistic ones. Before submitting, defocus and read the row as a single band; a kink in the gradient jumps out the moment you stop reading chips individually.
Stroop Test
The Stroop Test is the classic interference task. A colour word — RED, BLUE, PURPLE — appears printed in a different ink, above a row of swatches. Tap the swatch matching the ink and ignore what the word says; your reading reflex will fight you the whole way, which is the point. The choice row grows from three swatches to six across five rounds, each on a six-second clock. A correct tap scores 10 plus a speed bonus of up to +2, for a 60-point maximum.
How to score well in the Stroop Test
You can’t stop yourself reading the word, but you can start from the ink. Fixate on one letter’s edge rather than the word’s centre — reading needs the whole shape, colour doesn’t — then go straight to the swatches without re-checking the text. Most errors come from a second glance, which gives the word a fresh chance to vote.
Name That Color
Name That Color is an endless run. A shade fills the screen with four names under it; tap the name that actually belongs to the shade. Every third round flips direction — you get a name and pick its swatch from four. The wrong options aren’t random: they’re names whose colours sit perceptually close to the target, and both the names and the gaps tighten as you climb, from everyday vocabulary toward the deep cuts — early rounds offer red versus blue, deep runs offer chartreuse versus citrine. One wrong pick ends the run. Ten points per cleared round with no cap, ten seconds per decision.
How to score well in Name That Color
This is a vocabulary game wearing a perception costume. The game follows each name’s commonly documented value (CSS named colours where they exist), so it rewards convention, not opinion — learning the reference shades pays off directly. When two names both feel right, decide by undertone: is the swatch warmer or cooler than the canonical version you’re picturing? On flipped rounds, pre-visualise the named colour before looking, so the four swatches can’t anchor you.
Hex Guess
Hex Guess shows a swatch and asks you to type the six-digit hex code that produces it. That’s the whole game — five rounds, no multiple choice, no clock. Scoring uses the same CIEDE2000 closeness curve as the slider modes rather than exact-match, so a near-miss code still earns most of the round; three-digit shorthand like f80 is accepted too. Fifty points possible, and probably the most transferable skill in the whole collection if you write CSS for a living.
How to score well in Hex Guess
Read the colour as three numbers. The hue family sets the channel order — orange means red high, green mid, blue low — then estimate each channel coarsely (c0-ish, 80-ish, 40-ish) instead of agonising digit by digit. A handful of anchors (808080 mid-grey, ffa500 orange, 4169e1 royal blue) plus interpolation beats trying to memorise a swatch book.
Color Mixer
Color Mixer removes memory entirely: the target stays on screen while you work. Instead of HSV sliders you get the three channels of light — red, green and blue, 0 to 255, starting from mid-grey — and you mix until your swatch matches the target. No clock. Scoring is the usual CIEDE2000 curve, five rounds, fifty points. It’s the variant that teaches how screens actually make colour, and it’s harder than it sounds: additive mixing runs against every paint-mixing instinct you own.
How to score well in Color Mixer
Set the dominant channel first — it carries the hue family — then shape the exact hue with the second channel and use the third to control how washed-out the result is: closing the gap between channels pulls toward grey, widening it adds vividness. To lighten without shifting hue, raise all three roughly in proportion. Small steps, one slider at a time, checking against the target after each.
Color Sequence
Color Sequence is the Simon memory game played on four fixed pads — green, red, yellow and blue. A sequence lights up and you repeat it by tapping; each cleared round deals a fresh sequence one pad longer, starting at two with no cap, and the first wrong tap ends the run. Every cleared round is worth ten points, so the leaderboard total is simply rounds × 10 — clear five and you’ve matched the 50-point games, clear more and you beat them. Solo only, with its own leaderboard.
How to score well in Color Sequence
Chunk it. A nine-pad sequence is impossible as nine items and manageable as three groups of three — rehearse it as rhythm while it plays. Saying the colours under your breath as they flash builds a verbal track alongside the visual one, and the verbal track is the one that survives. There’s no speed bonus, so never tap ahead of your own certainty.
Daily challenge
The Daily challenge gives every player in the world the same five colours, seeded from the calendar date. The run is the standard Solo Easy format — four-second flash, three sliders, no decoys. You can post your score once per nickname per day to the daily leaderboard, where you’ll see what percentile you landed in. Rounds reset at midnight UTC.
How to score well in Daily
Don’t play Daily as your warm-up. The leaderboard rewards your best single run today, and you only get one entry per nickname. Play a few rounds of Solo Easy first to wake your eye up, then play Daily as a deliberate session. Pay extra attention to the first colour — most players score lowest on round one because they haven’t calibrated yet.
Challenge — vs the character roster
Challenge is a 1v1 against a roster of fifteen original characters. Each one plays a specific variant at their own skill level, with their own commentary between rounds. Smudge and Doodle play the easy variants at low accuracy. Speed bots like Blip, Static, and Frost play increasingly tighter Speed rounds. Oracle plays Blind Sliders at near-perfect accuracy. The bots taunt, gloat, and complain in character; the scoring is identical to the underlying variant.
How to score well in Challenge
Pick your fight. Bots are tier-labelled (Trivial, Easy, Medium, Hard, Genius) — that’s the bot’s skill, not the difficulty of the variant. A “Trivial” bot in Solo Easy is a lay-up; a “Genius” bot in Blind Sliders is genuinely hard to beat. Once you’ve found one you can consistently beat, work up the tiers. Wins and losses post to the per-character vs-bots leaderboard.
Arena — head-to-head
Arena is automated 1v1 matchmaking. You queue, get paired with the next player searching, and the mode is drawn at random from Easy, Hard, Speed, Blind Sliders, and Gradient — you can’t warm up for the exact test you’ll face. If no other player is searching within about 18 seconds, a bot opponent steps in so the match still happens. Both players face identical targets, and whoever scores higher across the five rounds wins. There’s no chat — you see only your own sliders and the round outcomes. It’s the closest thing to ranked play in the game.
How to score well in Arena
Play your own game. The temptation in head-to-head is to rush because your opponent might be rushing — don’t. Outside the Speed draw, Arena doesn’t reward pace; it rewards the higher absolute score across five rounds. Use every moment of viewing time the drawn mode gives you, dial deliberately, submit when you’re happy. Most Arena losses come from one wildly off round, not a slightly slower set.
Color Whispers (multiplayer)
Color Whispers is “telephone” played with colours. Three or more players sit in a chain. The first player sees the target, dials a guess, and submits. The second player sees only the first player’s guess, dials their own version of it from memory, and passes it on. By the end of the chain, the colour has drifted — sometimes a lot. Everyone in the chain sees the original target and the full chain at the end, and scores are based on how close each player’s guess is to the person before them, not to the original.
How to score well in Color Whispers
Match the input you were given, not the colour you think the original was. Trying to “correct” what you assume is already a drifted version usually drifts it further. Treat your round as a normal Solo round with whatever swatch you were shown as the target.
Sabotage (multiplayer)
Sabotage is a free-for-all for two or more players: five rounds, each opened by a twelve-second pick phase where everyone secretly spends tokens on debuffs aimed at opponents. Lock in, and the round plays as a normal memorise-and-dial round with whatever landed on you applied. There are six debuffs. Squint halves your viewing time, Shuffle rearranges the slider order, Blackout hides the live preview while you dial, Decoy flashes fake colours before the target and overlays a wrong colour name, Lock freezes the hue slider for the first five seconds, and Rush halves the guess timer. Tokens regenerate as you play: one per round, plus one for the round’s best score and one for scoring 8+ out of 10. The final scoreboard is straight cumulative accuracy — debuffs only ever affect the round they’re played in, and tokens never convert to points.
How to score well in Sabotage
Read the room. If you’re leading, expect the pile-on and play for consistency — with your view time squinted and your timer rushed, a safe six beats a gambled two. If you’re trailing, spend on the leader: Rush is the most reliable two-token play (it deletes ten seconds of dialing time), and Squint is the best value at one. Duplicate debuffs on the same target don’t stack, so spread them around instead of doubling up. And dump your whole balance by the final round — unspent tokens are worth nothing when the game ends.
Hide & Seek (multiplayer)
Hide & Seek is multiplayer only — two to eight players, and the only mode where everyone shares one canvas. Each round, every hider places a gingerbread man somewhere on the same busy painting and gets two minutes to paint him into the background, stroke by stroke, while the seeker waits blind. Then the hunt: the seeker sweeps the canvas through a rifle scope with a limited stock of paint bolts — two per hider, plus two — against the clock. Every thirty seconds a whistle blows and every surviving gingerbread man opens his glowing eyes for a heartbeat. Hiders can whistle voluntarily too, exposing only themselves for longer. Each voluntary whistle can add 0.25 points up to the normal round cap if they survive, but all of those whistle points are forfeited if they are found. Hiders also score for survival time and close-call bait; the seeker scores for finds and accuracy. The seeker role rotates until everyone has hunted once.
How to score well in Hide & Seek
As a hider, pick ground with detail — in a flat field, your eye-flash is the only event in the neighbourhood, while clutter gives the seeker ten things to check on every whistle. Paint edge to edge; a halo of mismatched background is what scopes catch. Voluntary whistles are safest when the scope is sweeping the opposite side of the painting, but repeated whistles can still lead it back to you. As the seeker, budget your bolts: every near miss hands the hider a bait bonus, so hold fire until a whistle, watch one region at a time, and shoot seams, not hunches.
Tips that apply everywhere
Three habits raise your scores across every variant. Calibrate your monitor — at the very least, turn off any night-shift / blue-light filter while you play, and make sure the room isn’t lit by anything heavily tinted (warm bulbs and direct sun both shift your perception). Encode colours as words — even a sloppy verbal label survives memory better than a pure visual impression. Always tune hue first — it’s where most of the points live, and dialling saturation and brightness before you’ve found the right hue family wastes time.
If you’ve been playing one variant exclusively, try a couple of rounds of Blind Sliders — it forces you to internalise what each slider position looks like, and the calibration carries over to every other mode.