CFOP OLL Tutorial
Recap: the two steps of 2-look OLL
When you learned LBL you oriented the last layer in two looks (2-look OLL, Orient the Last Layer): edges first — count the yellow edges facing up (dot / L / line / cross) and use the trigger F R U R' U' F' to push them step by step into a yellow cross; then corners — with the cross done, use the 7 OCLL (Orient Corners of Last Layer) cases (Sune / Anti-Sune / headlights / chameleon / bowtie / Pi / H) to twist the 4 corners yellow. Two separate looks.
CFOP's OLL merges those two looks into one (full 1-look OLL): read edges and corners at a glance and orient the whole top face in a single step. This page teaches how to recognise all 57 cases and what order to learn them — all 57 algorithms sit in the collapsed quick reference at the end of this page for lookup, while drilling them one-by-one happens on the drill page.
Goal: the whole top face yellow
Done = all 9 top stickers (4 edges + 4 corners + centre) yellow-up. Note this is orientation only — the pieces aren't yet permuted to their correct spots (that's the next step, PLL). The greyed top-layer side stickers signal: orientation is finished, side colours are PLL's job.
Recognition method: edges first, then corners
Recognising 57 cases at a glance rests on a two-step decision tree: step 1, count the yellow edges — cross / L / line / dot — which alone rules out most of the field (L-shape is half of all states!); step 2, read the corner pattern — headlights, bars, blocks, fish — to land on a family. The diagram below is the full tree; the 4 folds beneath it list each edge-shape's families (frequency / avg moves / two-axis difficulty), and per-family shapes & distinctions are in group analysis.
cross edge-shape (1 family · 12%)
| Family | cases | freq | avg moves | recog. diff | algo. diff | recognition cue |
|---|---|---|---|---|---|---|
| OCLL | 7 | 12.0% | 8.4 | 2 | 2 | cross done, orient 4 corners only |
L edge-shape (8 families · 50%)
| Family | cases | freq | avg moves | recog. diff | algo. diff | recognition cue |
|---|---|---|---|---|---|---|
| L-shape ★ | 6 | 11.1% | 10.5 | 3 | 4 | two adjacent edges up; corners vary — read side bars/blocks |
| Fish ★ | 4 | 7.4% | 10.0 | 2 | 4 | one corner up forms the fish silhouette (Sune / Anti-Sune) |
| Small Lightning | 4 | 7.4% | 10.3 | 4 | 4 | no headlights; small-lightning corners with a front-slot block |
| Awkward | 4 | 7.4% | 12.0 | 4 | 5 | awkward corners; headlights / bar / neither — three states |
| P-shape | 4 | 7.4% | 7.5 | 3 | 3 | P-shaped corners; side headlights or bar picks the alg |
| Square | 2 | 3.7% | 7.0 | 4 | 2 | square corners; two paired corner-edge blocks |
| W | 2 | 3.7% | 12.0 | 4 | 4 | no headlights; each side shows a different sticker pair |
| H | 1 | 1.9% | 7.0 | 3 | 1 | H-shaped corners with headlights and a bar |
line edge-shape (6 families · 25%)
| Family | cases | freq | avg moves | recog. diff | algo. diff | recognition cue |
|---|---|---|---|---|---|---|
| Knight-Move ★ | 4 | 7.4% | 10.5 | 4 | 4 | no headlights; chess-knight corners with a block |
| I-shape | 4 | 5.6% | 10.8 | 3 | 4 | two edges in a line; i-shaped corners |
| T-shape | 2 | 3.7% | 7.0 | 3 | 2 | T-shaped corners; line + block or headlights |
| C-shape | 2 | 3.7% | 9.5 | 3 | 3 | C-shaped corners; headlights vs bar |
| Big Lightning | 2 | 3.7% | 9.0 | 4 | 3 | big-lightning corners; the edge line stands out |
| Arrow | 1 | 0.9% | 9.0 | 3 | 3 | arrow-shaped corners with headlights and a bar |
dot edge-shape (1 family · 12.5%)
| Family | cases | freq | avg moves | recog. diff | algo. diff | recognition cue |
|---|---|---|---|---|---|---|
| Dot ★ | 8 | 12.5% | 12.0 | 1 | 5 | zero edges up; build the cross from scratch (usually two triggers) |
How frequency & difficulty are derived
Frequency (how often a case occurs in real solves): the premise is a WCA-standard scramble (random-state — the full cube is sampled uniformly from all legal states); old-style random-move scrambles and hand scrambles do not satisfy this, but every timer app and competition uses random-state, so the premise holds. Under it, there are 216 legal top-layer orientation states (4 corner orientations constrained by total twist=0 → 3³=27 corner states; 4 edge orientations constrained by total flip=0 → 2³=8 edge states; 27×8=216); fully enumerating them yields exact frequencies. Why enumeration = real-solve frequency (uniform after solving to OLL): ① random-state is uniform overall → the top-layer orientation is marginally uniform (each top state is realised by the same number of legal whole-cube states); ② F2L move choice depends only on the first-two-layer pieces, not on the top orientation; ③ F2L moves act on the top orientation as a permutation (moves are reversible, i.e. a bijection on the state space); ④ a uniform distribution is invariant under any permutation → after F2L the top orientation stays uniform. E.g. L-shape 50% (108/216 states), Dot 12.5% (27/216 states).
Recognition difficulty (1-5, lower = easier to spot): blends the family's landmark richness (how findable its headlights / blocks / bars are) with edge-shape crowding. e.g. Dot = 1 (no yellow edges at all — easiest), Awkward = 4 (no fixed landmark — hardest).
Algorithm difficulty (1-5, lower = easier to execute): weighted move count — base move-count tier, +1 for wide-layer turns (M/E/S), +1 for ≥2 F/B face turns. e.g. H = 1 (~7 moves, no wide), Dot = 5 (~12 moves + wide).
Why 16 families + 57 cases (extension)
The 216 orientation states partition by edge-shape into four groups (8 edge states → cross 1 / L 4 / line 2 / dot 1); within each group the corner states cluster by landmark into families — cross → 1 family (OCLL), L → 8 families, line → 6 (including Knight-Move / I-shape), dot → 1, total 16. Edge-shape is the first cut (count edges to pick the group), corner landmark the second — together spanning all 57 cases.
Among the 216 orientation states, two states that differ only by a U move (AUF) count as the same OLL case — you AUF the top layer and execute the same algorithm. Burnside's lemma counts equivalence classes (orbits) under a group action: the 4 rotations (identity / 90° / 180° / 270°) fix 216, 2, 12, and 2 states, giving (216+2+12+2)/4 = 58 orbits. Subtract the fully-oriented state (the OLL-skip) → 57 OLL cases. The 2 states fixed by 90° have all corners oriented and all 4 edges the same (both up or both down); the 12 states fixed by 180° = 3 corner-twist classes × 4 edge-flip classes.
Learning path: four stages by coverage
You don't have to grind all 57 at once. Learn in occurrence order: high-frequency families first (more real-game coverage per minute of practice), low-frequency tail last. OCLL — these 7 cases overlap with 2-look's corner step, giving you a 12% coverage head start. The roadmap below shows the four stages, each labelled with cumulative coverage + difficulty.
OCLL — 7 cases (Sune / Anti-Sune / headlights / chameleon / bowtie / Pi / H), cross done, corners only. They overlap with 2-look's corner step — 12% coverage head start.
Mastery — Just review; no new algorithms.
The four most frequent families; nail these and you're past half. Note frequency ≠ ease: Dot is the heaviest on the whole page (8 cases + build-the-cross-from-scratch + 12-move algs), but it's the most frequent so it must come early; L-shape (6 cases) is the edge-shape workhorse (L is half of all states); Fish is the most iconic; Small-Lightning confuses with L — check the front-slot block.
Mastery — Drill each family until no alg lookup, recognition < 2s before moving on.
Four mid-frequency families; together you reach 78%. Knight and Awkward both lack headlights — read how the corner blocks are offset; Awkward's 12-move alg is on the heavy side; P-shape's alg is short (~7.5 moves) and easy; I-shape has edges in a line with i-shaped corners.
Mastery — Drill each family until no alg lookup, recognition < 2s before moving on.
Seven long-tail families, each rare (1–2 cases, ≤ 3.7%) with short algs — just recognise them. The shape name is the cue: a block, letter shapes, bolts, an arrow. Rare enough you might go a week without seeing one — just match the shape and look up the alg.
Mastery — Recognise the shape and be able to look up the alg; no need to memorise.
Family by family: shapes & distinctions across the 16 families
Below we walk the 16 shape families one by one — each family's full set of case top-views side by side, so you can see how same-family cases relate (orientation / mirror / AUF variants) and how families differ. This is structure & relationships only — not per-case recognition features or algorithms; the algorithms are collected in the quick reference at the end of the page. Grouped by edge-shape: cross → L → line → dot.
Cross edges (1 family · 12.0%)
Cross done (all 4 edges up); the 7 cases orient the 4 corners only — the Sune / Anti-Sune / headlights / chameleon / bowtie / Pi / H. They differ by how many corners are up (1/2/4) and whether headlights are present. The 7 cases split into three groups by corners-up count: 1 up (Sune/Anti-Sune), 2 up (headlights/chameleon/bowtie), 4 up (Pi/H) — count corners-up first to pick the group, then read headlights to land the case. That's OCLL's built-in taxonomy.
L edges (8 families · 50.0%)
Two adjacent edges up in an L — the largest family. All 6 cases share the identical top yellow pattern — two adjacent yellow edges and zero yellow corners — so the top face gives no distinguishing clue; you must read the sides. They are not AUF variants of each other either: AUF only rotates the top layer and cannot change corner twists, so the 6 are six distinct corner-twist states (if any two were AUF-equivalent they would collapse into one case — the count of 57 is already the post-AUF figure). The real differences lie in the side-landmark configuration. Grouped by headlight-set count: (1) one headlight set — oll-47/48 have just one block and no bar (left/right F-wrap, mirrors), oll-49/50 carry a bar plus multiple blocks; (2) two headlight sets — oll-53/54 (one set carries the bar; left/right wide-layer wrap, mirrors). Recognition order: count headlight sets (1 vs 2), then check bar/block to land the case.
L edges + the iconic fish silhouette. None of the 4 are AUF variants (AUF can't change corner twists; the count of 57 is already post-AUF). They split by up-corner count: (1) one corner up (single-head fish) — oll-09/10, left/right mirrors (classic single-head / Sune-style mirror); (2) two corners up — oll-35 (no blocks) / oll-37 (blocks on all four sides), same up-corner count but entirely different side landmarks. Recognition: spot the fish, count up-corners (1 vs 2), then read side blocks to land the case.
The single up-corner (fish head) is Fish's unique signature — not confused with other L-edge families.
L edges + small-lightning corners, no headlights (this is the dividing line from L-shape, which always has ≥1 headlight set). All 4 cases have exactly one corner up and carry side blocks, but they are not AUF variants (AUF only repositions; it can't change twists, else they'd collapse to one). They differ by which corner is up and the block layout. Recognition: confirm no headlights to land in this family, then read the up-corner + block positions.
Most confused with L-shape — check for headlights (see the L-shape comparison above).
L edges + the "awkward" corner shape with no fixed landmark — headlights, bar, or neither across three states — the hardest family to read. The 4 cases need each corner-offset pattern checked. Its 4 cases are three-state varied (headlights / bar / neither) with no single landmark handle — the hardest family in all of OLL. They aren't simple orientation variants; each needs its corner-offset pattern memorised.
No-headlights is shared with W; distinguish by the specific corner shape.
L edges + P-shaped corners. The 4 cases are not AUF variants; they split by headlight-set count: (1) no headlights — oll-31/32, both two corners up with three blocks, differing by which side the blocks sit on (front F-gate / back B-gate); (2) one headlight set with a bar — oll-43/44, left/right mirrors (left/right wide-layer). Recognition: check headlights first to pick the group, then block side / mirror direction.
L edges + square corners (two paired corner-edge blocks), no headlights. The 2 cases are mirrors. The 2 cases are a left/right mirror pair; the square is itself 4-fold symmetric with no orientation variety, hence just this pair.
L edges + W-shaped corners, no headlights, a different sticker pair on each side. Both cases have two corners up (on a diagonal), but they are not rotations / AUF-related (after AUF collapse they remain two distinct cases) — they differ by which diagonal holds the up-corners (URF+ULB vs URB+ULF) and the side blocks.
L edges + H-shaped corners with headlights and a bar. A single case (oll-57). The H shape is bilaterally symmetric with no orientation variants, hence a single case — the rarest structure in the L-edge group.
Line edges (6 families · 25.0%)
Two opposite edges in a line + chess-knight corners, no headlights, with a block. All 4 cases have exactly one corner up, but they are not AUF rotations (AUF only repositions; the 4 are four distinct twist states). They split by block count: (1) three blocks — oll-13/14, left/right mirrors (front-left / front-right corner); (2) one block — oll-15/16, left/right mirrors (back-left / back-right corner). Recognition: spot the knight + block, then read block count and direction.
Two edges in a line + i-shaped corners. All 4 cases have zero corners up, but they are not orientation variants — they're distinguished by headlight-set count (increasing symmetry): oll-51 (1 set), oll-52 (2 sets + bar), oll-55 (3 sets + 2 bars), oll-56 (3 sets, no bar/blocks). oll-55/56 both have 3 headlight sets and differ by bar/block presence.
Most confused with Knight-Move — check the up-corner count (I-shape 0 vs Knight-Move 1; see the Knight-Move comparison above).
Line edges + T-shaped corners; the line pairs with a block or headlights. Both cases have two corners up, but they are not mirror/rotation related — they split by headlights: oll-33 (no headlights + blocks on all four sides) / oll-45 (one headlight set + one block).
Line edges + C-shaped corners. Both cases have two corners up; they split by whether a headlight+bar is present: oll-34 (no headlights, no blocks) / oll-46 (one headlight set with a bar + multiple blocks).
Line edges + big-lightning corners; the edge line stands out. Both cases have two corners up + three blocks and no headlights; they form a left/right mirror pair (left/right F-wrap) — not orientation variants.
Line edges + arrow-shaped corners with headlights and a bar. A single case. The arrow is highly symmetric with no orientation variants — a single case, the rarest in the line-edge group.
Dot edges (1 family · 12.5%)
Zero edges up; build the cross from scratch (usually two triggers: flip to L/line first, then to cross). The 8 cases are the largest family, differing by how many corners are already up (0/2/4) + landmarks. Easiest to spot (no yellow edges at all) but the longest algorithms (avg 12 moves). The 8 cases split into three groups by corners-already-up count: 0 (flip all), 2, 4 — then subdivide by landmark config. Cases differ in starting corner state: the cross-building trigger is the same, but the starting corner orientation sends you to different landing cases.
Common recognition traps
The usual OLL recognition pitfalls:
- Looking only at corners and forgetting to count edges — L-shape is half of all states, so fixing the edge-shape first cuts the field down sharply instead of hunting across 16 families.
- Holding a mirror family backwards — many families have left/right mirrors (e.g. Sune / Anti-Sune); holding it wrong runs the wrong alg. Fix the headlights or fish-head direction first.
- Mistaking a partial cross inside a dot for an L or line — dot means "no yellow edges up at all"; don't let yellow-up corners distract the edge count.
- Missing a symmetric case — a few cases (e.g. oll-20, four-fold symmetric) turn up even less than 1/57 (< 1% after AUF collapse); failing to recognise them is normal — just look them up.
Go drill the algorithms
Algorithm quick reference · all 57
R U2 R' U' R U R' U' R U' R'R U2 R2 U' R2 U' R2 U2 RR2 D R' U2 R D' R' U2 R'r U R' U' r' F R F'R' F R B' R' F' R BR U2 R' U' R U' R'R U R' U R U2 R'F' L' U' L U L' U' L U FF R U R' U' R U R' U' F'R' F R' F' R2 U2 y R' F R F'R B' R B R2 U2 F R' F' Rl' U' L U' L' U L U' L' U2 lr U R' U R U' R' U R U2 r'R U R' U' R' F R2 U R' U' F'R U R' U R' F R F' R U2 R'R U2 R2 F R F' R U2 R'F R U' R' U' R U R' F'r U R' U R U2 r'R U2 R' U2 R' F R F'F' L' U' L U F y F R U R' U' F'F R U R' U' F' U F R U R' U' F'M U R U R' U' R' F R F' M'R2 U R' B' R U' R2 U R B R'R U' R' U2 R U y R U' R' U' F'R' U2 R U R' U R y F R U R' U' F'R' U' F U R U' R' F' RR U B' U' R' U R B R'f' L' U' L U ff R U R' U' f'l' U2 L U L' U lr U2 R' U' R U' r'L' U' L U' L' U L U L F' L' FR U R' U R U' R' U' R' F R F'M' U M U2 M' U Mr U' r' U' r U r' y' R' U RR' F R U R' F' R y' R U' R'l' U' l L' U' L U l' U lr U r' R U R' U' r U' r'f R U R' U' R U R' U' f'R U R' U R d' R U' R' F'R U2 R2 U' R U' R' U2 F R F'F R U R' U' R F' r U R' U' r'R U R' U' R' F R F'F R U R' U' F'R U R2 U' R' F R U R U' F'R' U' R' F R F' U RL F' L' U' L U F U' L'R' F R U R' U' F' U RR U R' U' M' U R U' r'R U2 R2 F R F' U2 R' F R F'F R U R' U' F' f R U R' U' f'f R U R' U' f' U' F R U R' U' F'f R U R' U' f' U F R U R' U' F'R U R' U R' F R F' U2 R' F R F'F R U R' U y' R' U2 R' F R F'M U R U R' U' M' R' F R F'M U R U R' U' M2 U R U' r'This page gives the recognition framework + learning path. All 57 algorithms are in the folded quick reference above (grouped by edge-shape, look them up any time); to drill them to fluency and speed, practise by family on the drill page until you no longer look up the alg. The next step is PLL: orientation is done in OLL; PLL permutes the top-layer pieces (corners and edges) into their correct spots.
Drill the 57 OLL cases on /learn