LucaArmstrong
Member
Here's a new method I made the other day which is a cross between the Roux and Mehta methods. There are example solves near the bottom and the steps are shown below (white on bottom)
Step 1 - Left block:
Make a 2x3x1 block on the left like in Roux
Step 2 - Belt:
Solve the remaining two edges which don't have yellow or white on them (complete the belt). Although the edges can be difficult to find for this step, planning left-block in inspection means the belt edges can be tracked reducing pauses. You can also plan the belt during inspection if the left-block solution is easy enough.
Step 3 - 6CO:
Orient the 6 remaining corners. You can use the 6CO algorithms from Mehta to do this or any other way
Step 4 - Orient edges:
Place any oriented edge in the DR position and orient the other 6 edges with M and U moves LSE style. (Since there are 7 edges left for this step, there will be at least one oriented edge because of parity which can be placed in DR)
Step 5 - Right block:
Solve the right 2x3x1 block without disturbing orientation, in other words, using just U, R2, M2 and Rw2 moves.
Step 6 - CP-bar:
The goal of this step is to permute the 4 yellow corners while forming the white edges into a bar on top of the cube. The first part of this step is to move both white edges to the top of the cube. Then you can do an algorithm which solves the corners while moving 2 white edges into a bar. I've just been using PLL algorithms for this, but are probably shorter ones which do the same job since there are another 4 edges whose permutations don't need to be preserved. For the sake of space I'm not showing the cases here.
Step 7 - Solve the cube intuitively:
In this step, you keep the white bar on top with the white edges in the UF and UB positions, and solve the cube fairly intuitively. The way you do this is by recognising one of a few cases depending on which edges are on the left and right relative to their corners, and which centre is on top. I made a list of these cases which are in fact move optimal, and can always be solved in 9 moves or less. Of course, you could solve the left and right edges like in Roux and continue, but this won't always give the lowest number of moves.
Using this method I've been averaging 25-30 seconds per solve compared to 16 seconds with CFOP. I don't think this method could be used for speedsolving but I did come up with two other variants. The first is where you solve the right block before orienting the edges which in practice would save a few moves but might make it harder to find the DR edge for making the right block. For this reason, I actually think the original method is slightly faster. The other which I think could be used for speedsolving is where you solve the right block first, then do EODFDB, orienting the edges while solving the white edges into the D layer. Then doing PLL would solve the cube. I did some working out and this would save roughly 6 moves per solve with the added advantage of PLLs being easier to recognise and being able to be executed fully at high TPS.
Here are a couple example solves using the original method. (I'm not a Roux user so my first block solutions probably aren't great)
Scramble 1 - F2 R' B' L B2 R' D L2 F2 L' B2 U2 F2 U2 R F2 L U2 B
left block - z2 F2 B' L2 D R' U' R' U' L' U' L
belt - Rw U' Rw' R' U R'
6CO - R U R U2 R' U R'
orient edges - R2 M U' M' U2 M' U' M U' M' U' M
right block - M2 U' Rw2 U' R2 U2 R2
CP-Bar - U + Rb perm (R2 F R U R U' R' F' R U2 R' U2 R) + U
solve - U' M U2 M' + M2 U M2 U2
Scramble 2 - R' D2 L' B2 R U2 F2 D2 F2 R F2 R2 U R2 B' U F' R D' L U
left block - z2 y U' B F D F' D L D' M2 L U' L'
belt - R
6CO - U2 R2 U R U2 R' U' R U' R'
Orient edges - R2 M U' M' U2 M' U' M U' M' U' M
right block - R2 U' M2 U2 M2 U R2
CP-Bar - U' M2 U' + Jb perm (R U R' F' R U R' U' R' F R2 U' R')
solve - U' M U2 M' + U2 M2 U M2 U
I'd be interested to hear anyone's thoughts on this method and the variants
Step 1 - Left block:
Make a 2x3x1 block on the left like in Roux
Step 2 - Belt:
Solve the remaining two edges which don't have yellow or white on them (complete the belt). Although the edges can be difficult to find for this step, planning left-block in inspection means the belt edges can be tracked reducing pauses. You can also plan the belt during inspection if the left-block solution is easy enough.
Step 3 - 6CO:
Orient the 6 remaining corners. You can use the 6CO algorithms from Mehta to do this or any other way
Step 4 - Orient edges:
Place any oriented edge in the DR position and orient the other 6 edges with M and U moves LSE style. (Since there are 7 edges left for this step, there will be at least one oriented edge because of parity which can be placed in DR)
Step 5 - Right block:
Solve the right 2x3x1 block without disturbing orientation, in other words, using just U, R2, M2 and Rw2 moves.
Step 6 - CP-bar:
The goal of this step is to permute the 4 yellow corners while forming the white edges into a bar on top of the cube. The first part of this step is to move both white edges to the top of the cube. Then you can do an algorithm which solves the corners while moving 2 white edges into a bar. I've just been using PLL algorithms for this, but are probably shorter ones which do the same job since there are another 4 edges whose permutations don't need to be preserved. For the sake of space I'm not showing the cases here.
Step 7 - Solve the cube intuitively:
In this step, you keep the white bar on top with the white edges in the UF and UB positions, and solve the cube fairly intuitively. The way you do this is by recognising one of a few cases depending on which edges are on the left and right relative to their corners, and which centre is on top. I made a list of these cases which are in fact move optimal, and can always be solved in 9 moves or less. Of course, you could solve the left and right edges like in Roux and continue, but this won't always give the lowest number of moves.
Using this method I've been averaging 25-30 seconds per solve compared to 16 seconds with CFOP. I don't think this method could be used for speedsolving but I did come up with two other variants. The first is where you solve the right block before orienting the edges which in practice would save a few moves but might make it harder to find the DR edge for making the right block. For this reason, I actually think the original method is slightly faster. The other which I think could be used for speedsolving is where you solve the right block first, then do EODFDB, orienting the edges while solving the white edges into the D layer. Then doing PLL would solve the cube. I did some working out and this would save roughly 6 moves per solve with the added advantage of PLLs being easier to recognise and being able to be executed fully at high TPS.
Here are a couple example solves using the original method. (I'm not a Roux user so my first block solutions probably aren't great)
Scramble 1 - F2 R' B' L B2 R' D L2 F2 L' B2 U2 F2 U2 R F2 L U2 B
left block - z2 F2 B' L2 D R' U' R' U' L' U' L
belt - Rw U' Rw' R' U R'
6CO - R U R U2 R' U R'
orient edges - R2 M U' M' U2 M' U' M U' M' U' M
right block - M2 U' Rw2 U' R2 U2 R2
CP-Bar - U + Rb perm (R2 F R U R U' R' F' R U2 R' U2 R) + U
solve - U' M U2 M' + M2 U M2 U2
Scramble 2 - R' D2 L' B2 R U2 F2 D2 F2 R F2 R2 U R2 B' U F' R D' L U
left block - z2 y U' B F D F' D L D' M2 L U' L'
belt - R
6CO - U2 R2 U R U2 R' U' R U' R'
Orient edges - R2 M U' M' U2 M' U' M U' M' U' M
right block - R2 U' M2 U2 M2 U R2
CP-Bar - U' M2 U' + Jb perm (R U R' F' R U R' U' R' F R2 U' R')
solve - U' M U2 M' + U2 M2 U M2 U
I'd be interested to hear anyone's thoughts on this method and the variants
Last edited:
