Step 3
Now, you have to solve the last four corners. Mastering cases recognition takes time, don't expect to be good at this step before several months. You should learn cases progressively, using two easy sequences (orientation and permutation for example) for the ones you don't know.
Many other web sites teach you how to solve four corners. Depending on the algorithm followed, other pieces may or may not be moved (last layer edges for example, and middle slice edges in Waterman's method). In this method, U-edges and M-slice are free.
The corners generate eight configurations of orientations (if you look at what should be U color) and six possible permutation cases. No need to learn 6x8=48 cases. For many of them, it's only a matter of symmetry, or applying a reverse sequence. You have to remember "only" 24 unique sequences if you want to master this step. Progressively, you'll find out how to detect patterns for fast case recognition, from different point of views (and with a lot of experience, you'll even begin to analyze the configuration and ready your fingers before the end of Step 2).
For many of the sequences given below, you need to adjust U before execution. Do the reverse sequence, you'll see where to start from.
| Analyze
the orientation pattern and find the right column |
 |
 |
 |
 |
 |
 |
 |
I (A1)' Sym(A1) |
R'UL'U2RU'R'U2R2 (A2)' (A3)' (A4)' (A5)' Sym(A2, A3, A4, A5) |
(A2)' (A3)' (A4)' (A5)' Sym(A2, A3, A4, A5) |
(A2)' (A3)' (A4)' (A5)' Sym(A2, A3, A4, A5) |
(A2)' (A3)' (A4)' (A5)' Sym(A2, A3, A4, A5) |
R2U'RF'R'Ur2UFUF' (A6)' Sym(A6) |
 |
R'U'RU'R'U2R (C1)' Sym(C1) |
R'FU2F'RFR'U2RF' (C4)' Sym(C4) |
R'FRF'U2F'U2F (C2)' Sym(C3) |
(C3)' Sym(C2) |
R'ULU'RUR' (C5)' Sym(C5) |
RU2R'U'RU'BUB'U'R' (C6)' Sym(C6) |
 |
(B1)' Sym(B1) |
(B3)' Sym(B4) |
(B4)' Sym(B3) |
(B2)' Sym(B2) |
(B5)' Sym(B5) |
(B6)' Sym(B6) |
 |
(R'DR'DR')U(R'DR'DR')' (D1)' Sym(D1) |
R'F2R'U'RF2R'UR2 (E5)' Sym(D3) |
(E3)' Sym(D2) |
RU'r'U'F'UF (F5)' Sym(D5) |
(F3)' Sym(D4) |
RUR2F2rFR'F2R (D6)' Sym(D6) |
 |
R2F'UFU'F'U'F2RF'R (F1)' Sym(E1) |
RU'L'UR'U2B'UBL (E2)' Sym(E2) |
(D3)' Sym(E5) |
RB'UR'B'RU'R'B (F4)' Sym(E4) |
(D2)' Sym(E3) |
FRUR'U'F' (E6)' Sym(E6) |
 |
(E1)' Sym(F1) |
R'UrU2R2FRF'R (F2)' Sym(F2) |
(D5)' Sym(F5) |
(E4)' Sym(F4) |
(D4)' Sym(F3) |
R'URUBU2B'RB'R'B (F6)' Sym(F6) |
 |
(E6)2 (G1)' Sym(G1) |
R'F2RF'U2RU'L'B'U (G4)' Sym(G4) |
R'FRUFU'RUR'U'F' (H2)' (H4)' Sym(G3) |
(G2)' Sym(G2) |
RU'L'UR'ULUL'UL (H3) ' (H5)' Sym(G5) |
R'U'RU'R'UF'UFR (G6)' Sym(G6) |
 |
(R'FRUR')F(R'FRUR')' (H1)' Sym(H1) |
(G3)' Sym(H2, H4) |
(G5)' Sym(H3, H5) |
(G3)' Sym(H2, H4) |
(G5)' Sym(H3, H5) |
(E6)3 R2B'D'R2E'F2R2URF2 (H6)' Sym(H6) |
Notes:
- In this step, a quarter-turn metric is the best speed metric.
- Because M-slice is free, instead of an R move at the beginning or at the end of a sequence, you can do r, l or L.
- Many, many more possibilities, ask me if you've got a problem with a case.
- I did not propose a fast recognition technique based on easy color patterns, because you'll find them out automatically, and because you can achieve better recognition skills with a deeper analysis.
An
old video
showing sequences for all the casesSome other fast sequences I use:
| 1 |
2 |
3 |
4 |
5 |
6 |
|
| A |
(Don't
stop now, stupid!) |
(U2)R'UL'U2RU'R'U2R2 | (U)R'UL'U2RU'R'U2R2 | R'UL'U2RU'R'U2R2 | (U')R'UL'U2RU'R'U2R2 | R'UL'U2RU'BL'B2RB'R |
| B |
R'U'RU'R'U2R | L'URU'BR'F'LU'L'FL |
B'RBR'U2R'U2R |
LU2L'U2L'BLB' |
R'ULU'RUR' | RU2R'FR'F'RU'RU'R' |
| C |
RUR'URU2R' |
L'U2LU2LF'L'F |
FR'F'RU2RU2R' |
LU'R'UF'RBL'ULB'L' |
RU'L'UR'U'R |
R'U2RB'RBR'UR'UR |
| D |
R'DR'DR'URD'RD'R | RBL'B2RB'LBR'BR' |
(U')L'B'RB2L'BR'B'LB'L |
(U')L'B'R'BL'B'RB |
RBLB'R'BL'B' |
RUR2F2rFR'F2L |
| E |
x'(RU'R')z'(RU'R)z(R'UR)z'(R'UR') |
(U')FR2DR'URD'R2U'F' |
R'UL'URU'LU2R'UR |
RB'UR'B'RU'R'B |
R'FRUR'FRUFU2F' |
(U')FRUR'U'F' |
| F |
x'z'(RU'R)z(R'U'R)z'(R'UR')z(RUR') |
R'UrU2R2FRF'R | (U')R'U'RURB'R'B |
FR'U'RF'R'UF'R |
(U')RUR'U'R'FRF' |
R'FRF'RU2R'U'F'U'F |
| G |
FRUR'U'RUR'U'F' |
R'F2RF'U2RU'L'B'U |
(U')R'FRUFU'RUR'U'F' |
RB2R'BU2R'ULFU' |
R'FR'F'R2U2B'RBR' |
(U2)R'U'RU'R'UF'UFR |
| H |
(U)R'FRUR'FRU'R'F'R |
FURU'R'UF'U'R'F'R |
RUR'URUL'UR'U'L |
(U2)FURU'R'UF'U'R'F'R | (U2)RUR'URUL'UR'U'L | (U)FRUR'U'RUR'U'RUR'U'F' |
Possible improvement:
Solving the last corners with non matching 1x2x3 blocks is easy while solving slowly (useful for fewest-moves) and allows you to optimize the first two steps much more. But fast recognition in such circumstances is very difficult. However, it should be possible thanks to a recognition system based on locating L/R colors first (instead of the orientation pattern above).
James Straughan wrote a table that directly gives you a solution to all cases when side blocks don't match (and of course, it works when they do).
Updated 2010/05: James proposed another table that some may prefer, because case recognition is done by always looking at the same positions for a given pattern of L/R stickers.