RUBIK'S CUBE
OLL and PLL trainer
Thirty-one last-layer cases — 21 PLL and the 10 two-look OLL — to recognise at a glance. You see the case drawn from above, recall the algorithm in your head, then reveal it and judge yourself. The ones you miss go back to the start, the ones you know move away: this is box repetition, not a list to re-read. Free, no sign-up.
Space reveals the algorithm, then 1 = «Again», 2 = «Got it». The menu lets you drill PLL only, OLL only, or everything.
How it works
Each case is drawn the way you see it on the table: the last layer from above, with the three adjacency rows along the edges. For OLL what matters is which pieces are already yellow; for PLL what matters is the arrows, that is where each piece has to go. You look, you recall the algorithm, and only then do you reveal it.
The judging is yours and has exactly two answers. «Got it» moves the case up one box; «Again» sends it back to the first. The three boxes are to learn, in progress, mastered, and the draw weights them six to three to one: the cases you do not know come round six times more often than the ones you already have. That is the Leitner method, the one with the boxes of cards.
What is in it
- 21 PLL — all of them, from the two- and three-piece cases (Ua, Ub, Aa, Ab) up to full permutations like E and V.
- 10 two-look OLL — first the three edge cases (dot, line, L shape), then the seven corner ones: Sune, Antisune, H, Pi, T, headlights and bowtie.
- Filter — drill everything together, or just one of the two families.
- Counters — how many cases are still to master and how many have reached the third box.
Why two-look
Full OLL is 57 algorithms. The two-look version uses ten of them and solves the same layer in two passes instead of one: edges first, then corners. It costs a few extra moves per solve, but it is learned in a week instead of three months — and it is the right step for someone who has just finished the layer method. Once the ten are automatic, the 57 become an addition rather than a wall.
Frequently asked questions
Are the algorithms correct?
They are verified by the page itself, not transcribed and hoped for: at start-up each of the 31 is run on the cube engine and checked to make sure it touches the last layer only, leaving the two below untouched. The result of that self-check is written in the menu, so you can see it too.
What notation are they in?
Standard notation: U R F D L B for the faces, an apostrophe for anticlockwise, a 2 for a half turn, lower case for the wide two-layer moves. The same as any algorithm list you will find.
Is progress saved?
Yes. Each case's box stays in the browser's local storage: close it and carry on where you were. There is no account and the data never leaves your device. The menu lets you wipe it all and start again.
Do I need to be able to solve the cube already?
You need the layer method. OLL and PLL are the two final stages of CFOP, and they make sense once the first two layers happen without thinking. If you are not there yet, start with the 3D cube: the solver in beginner mode shows the layer method stage by stage.
Does it work on a phone?
Yes, with the buttons under the diagram. On a keyboard it is quicker: space to reveal, 1 and 2 to judge yourself.
How it is built
The real problem with an algorithm trainer is drawing the case, and it is not a graphics problem. Every figure has to be exactly what you see after the algorithm has acted, otherwise you learn to recognise a wrong drawing — which is worse than not practising at all.
So the drawings here are not drawn: they are computed. Each case starts from its own algorithm, which is run backwards on a solved cube; the result is the state you are in when that case turns up in front of you. The figure is therefore a consequence of the algorithm, not an illustration of it — and if an algorithm were mistranscribed, the drawing would change and the self-check would notice. It is the same principle as the 3D cube, where the move permutations are also born from the geometry instead of hand-copied tables.
It comes from the same way of working I use on real products: Miraviso is a concrete case, and if you need a hand with something similar this is where to start.